What We Teach About Race and Gender: Representation in Images and Text of Children’s Books

Abstract

Books shape how children learn about society and norms, in part through representation of different characters. We introduce new artificial intelligence methods for systematically converting images into data and apply them, along with text analysis methods, to measure the representation of skin color, race, gender, and age in award winning children's books widely read in homes, classrooms, and libraries over the last century. We find that more characters with darker skin color appear over time, but the most influential books persistently depict characters with lighter skin color, on average, than other books, even after conditioning on race; we also find that children are depicted with lighter skin than adults on average. Relative to their growing share of the U.S. population, Black and Latinx people are underrepresented in these same books, while White males are overrepresented. Over time, females are increasingly present but appear less often in text than in images, suggesting greater symbolic inclusion in pictures than substantive inclusion in stories. We then present analysis of the supply of, and demand for, books with different levels of representation to better understand the economic behavior that may contribute to these patterns. On the demand side, we show that people consume books that center their own identities. On the supply side, we document higher prices for books that center nondominant social identities and fewer copies of these books in libraries that serve predominantly White communities. Lastly, we show that the types of children's books purchased in a neighborhood are related to local political beliefs.

Ceilings Made of Glass and Leaving En Masse? Examining Superintendent Gender Gaps and Turnover Over Time Across the United States

“…Diverse leaders provide "mirrors" for diverse students to see themselves reflected and "windows" of opportunities for students to see the world through other perspectives-for example, young Black girls not internalizing social norms in which leadership is seen as a "white man's world" (Bos et al, 2022). Diverse leaders also have the potential to invalidate the discriminatory yet common-either intentionally or unintentionally-practice of instilling the belief in individuals and communities that members of underrepresented groups are inherently deficient (Adukia et al, 2022). Moreover, research shows that students of color are more likely to be exposed to teachers of color when their school has a leader of color (Goff et al, 2018), which has a significant positive impact on academic outcomes for all students but particularly for students of color (Bartanen & Grissom, 2021;Egalite et al, 2015;Gershenson et al, 2017;Grissom et al, 2017;Joshi et al, 2018;Lindsay et al, 2021).…”

Computational Narrative Understanding: A Big Picture Analysis

“…Grounded in NLP, computational narrative understanding has largely prioritized written narratives for understandable reasons. However, such textdriven approaches leave out large portions of storytelling behavior, including movies and television (Arnold et al, 2019;Papalampidi et al, 2019), usergenerated streaming content, illustrated content in comic strips (Edlin and Reiss, 2023), graphic novels, or children's books (Adukia et al, 2021), and finally video games, which might have stronger or weaker narrative structures. While textual narratives are largely unimodal in nature (though the physical and visual dimensions of books has been a vibrant area of study for a long time (Collective, 2019)), these other narrative forms are all crucially multi-modal in nature.…”

“…2023)). Nevertheless, investigations into multimodal narrative understanding, such as the relationship between text and illustrations in children's books or graphic novels is in need of more attention (see Adukia et al (2021) for an example exploring the visual qualities of children's book illustrations with respect to race). Understanding the kinds of gestural or pictorial preferences that are foregrounded given certain textual cues could give us insights into the way humans translate language into image (and vice versa) across different cultural domains.…”

Proceedings of the Big Picture Workshop

“…Grounded in NLP, computational narrative understanding has largely prioritized written narratives for understandable reasons. However, such textdriven approaches leave out large portions of storytelling behavior, including movies and television (Arnold et al, 2019;Papalampidi et al, 2019), usergenerated streaming content, illustrated content in comic strips (Edlin and Reiss, 2023), graphic novels, or children's books (Adukia et al, 2021), and finally video games, which might have stronger or weaker narrative structures. While textual narratives are largely unimodal in nature (though the physical and visual dimensions of books has been a vibrant area of study for a long time (Collective, 2019)), these other narrative forms are all crucially multi-modal in nature.…”

“…2023)). Nevertheless, investigations into multimodal narrative understanding, such as the relationship between text and illustrations in children's books or graphic novels is in need of more attention (see Adukia et al (2021) for an example exploring the visual qualities of children's book illustrations with respect to race). Understanding the kinds of gestural or pictorial preferences that are foregrounded given certain textual cues could give us insights into the way humans translate language into image (and vice versa) across different cultural domains.…”