The Geometry of Culture: Analyzing the Meanings of Class through Word Embeddings

Abstract

We demonstrate the utility of a new methodological tool, neural-network word embedding models, for large-scale text analysis, revealing how these models produce richer insights into cultural associations and categories than possible with prior methods. Word embeddings represent semantic relations between words as geometric relationships between vectors in a high-dimensional space, operationalizing a relational model of meaning consistent with contemporary theories of identity and culture. We show that dimensions induced by word differences (e.g., man -woman , rich -poor , black -white , liberal -conservative ) in these vector spaces closely correspond to dimensions of cultural meaning, and the projection of words onto these dimensions reflects widely shared cultural connotations when compared to surveyed responses and labeled historical data. We pilot a method for testing the stability of these associations, then demonstrate applications of word embeddings for macro-cultural investigation with a longitudinal analysis of the coevolution of gender and class associations in the United States over the 20th century and comparative analysis of historic distinctions between markers of gender and class in the U.S. and Britain. We argue that the success of these high-dimensional models motivates a move towards "high-dimensional theorizing" of meanings, identities and cultural processes.

The Cinderella Complex: Word embeddings reveal gender stereotypes in movies and books

“…After three waves of feminism [38], words like brave and independent are more likely to associate with female roles [39]. Females' increasing entry into professional occupations enhances their perceived competence, and the improvement of their education level also helps 16 break the gender stereotypes [25]. In a recent study, Gard et al analyzed gender stereotypes in the past century using word embeddings and found that gender bias was decreasing, especially after the second-wave feminism in the 1960s [15].…”

“…We 3 suggest that the emerging word embedding techniques [12,13] provide new tools to automate sentiment labeling and scale up the analysis of stories. Although word embeddings have been used to explore social and cultural dimensions in large-scale corpora [14][15][16], to our limited knowledge, we firstly apply them to analyze the shape of stories and quantify gender stereotypes.…”

Machine learning as a model for cultural learning: Teaching an algorithm what it means to be fat

“…Our paper also makes several contributions to computational sociology. We use embedding methods, which are increasingly being used in sociology and the social sciences for empirical analyses of culture (e.g., Kozlowski et al (2019); Jones et al (2019); Stoltz and Taylor (2019)). Our paper theoretically motivates these methods by drawing parallels to culture and cognition, using Marr's levels of analysis (Foster, 2018;Marr, 1982), and proposing that embeddings model schemata (specifically, those having to do with semantic memory and word meaning).…”

“…An abundance of previous work on scale development suggests that a dimension is one popular representation of gender (Kachel et al, 2016). Additionally, dimensions also operationalize binary opposition -a core structure of meaning in structuralism, as noted by Kozlowski et al (2019). In a binary opposition, meaning emerges from the opposition of two poles, such as the concept of gender being defined by the opposition between masculinity and femininity (Lévi-Strauss, 1963).…”

What are the Biases in My Word Embedding?

“…There are multiple reasons why such an algorithm is useful. First, social scientists can use it as a tool to study human bias, as data analysis is increasingly common in social studies of human biases (Garg et al, 2018;Kozlowski, Taddy, and Evans, 2018). Second, finding bias is a natural step in "debiasing" representations (Bolukbasi et al, 2016).…”