Our ever-present culture clash over how and when children learn about race becomes especially fraught between the pages of children’s books. Publishers have sought to diversify children’s literature since the 2020 rise of the #BlackLivesMatter movement, but gains remain small.
Books written by historically underrepresented authors increased by 3% in 2020, to 26.8%, while books written about racially diverse characters increased by only 1% to 30%, according to Kathleen Horning, director of the University of Wisconsin-Madison’s Cooperative Children’s Book Center. This progress has been further threatened by right-wing efforts to banish the discussion of race from classrooms altogether.
What accounts for the stubbornness of racial bias in children’s books?
To answer this question, the National Bureau of Economic Research (NBER) used artificial intelligence tools to mine a century’s worth of award-winning children’s books for data about race and gender representation. The study examines winners of the American Library Association’s (ALA) children’s literature awards and is divided into two samples: “mainstream” books, or winners of the Newbery or Caldecott awards, and “diversity” books, or winners of prizes like the Coretta Scott King or Rise Feminist awards. Together, these books make up what the authors call the “canon” of children’s literature.
Despite the recent increase in characters with darker skin, the study found, children are generally depicted in illustrations with lighter skin than adults, and the “mainstream” books show lighter skin tones overall than their “diversity” counterparts. Dark-skinned children rarely see themselves represented in the pages of their books, even when publishers are being more intentional about the nuances of representation.
But as important as the study’s conclusions are, they also come with their own biases, many of which are encoded in the very AI tools that produced these results.
The study was conducted using an adaptation of Google’s Vision Cloud technology, which came under fire in early 2021 for producing racist results when presented with photographs of Black subjects. Computer scientists have gotten better at solving for these problems in their code, but what happens when biased AI is turned loose on a set of illustrations? What does it mean to analyze not just the visual markers of race but how those visual markers are interpreted and transformed by artists?
Racial reckoning between the pages
In the United States, the history of children’s book illustration and racial representation are intimately linked. The expansion of cheap print technology in the first half of the 19th century made illustration more economically viable during a period when racial reckoning was at the forefront of the national conversation.
Harriet Beecher Stowe’s “Uncle Tom’s Cabin” — a racist work that has by turns been credited with drumming up abolitionist feeling and condemned for reinforcing racial stereotypes — was released in book form in 1852 with seven illustrations by Hammatt Billings. In these images, Little Eva was as White as the printed page, while Tom shared the tones and values of the shadows.
As “Uncle Tom’s Cabin” persisted in print, it was abridged and adapted for children. After Stowe’s copyright lapsed, many of these abridgments took liberties with the already-contested message of the novel, revising the ending to endorse the logic of Jim Crow segregation. Illustrations often exaggerated Billings’ tendency to represent Black and White characters as visual opposites. In 1900, the Philadelphia publisher Henry Altemus Co. released a children’s version of “Uncle Tom’s Cabin” showing Eva and Topsy on its cover. Eva, a White child, is rendered as a line drawing; her hair, skin, and clothes all share the tone of the White cloth cover. Topsy, a Black child, meanwhile, is rendered in solid Black, with a bright red dress and just enough facial detail to differentiate her features.
Seeking to push back on the trend of reducing Black characters to exaggerated silhouettes, W.E.B. Du Bois’ “The Brownie’s Book,” the first magazine published for Black children, contained some illustrations representing Black figures in the complete absence of tone. In the first issue, a sketch entitled “Once Upon a Time in Uganda” is illustrated with an airy line drawing of a mother telling a story to her three children, inaugurating what would become one of the magazine’s signature styles.
Hilda Wilkinson Brown and Laura Wheeler Waring are credited with illustrating this image. Both were groundbreaking Black women artists and educators. Their larger bodies of work display an incredible sensitivity to skin tone, but when faced with the challenge of illustrating for children, both artists often completely eliminated tone, a choice that suggests an interest in challenging the sorts of children’s illustrations that collapsed the range and dimensionality of Black subjectivity into Whiteness’ monolithic opposite. If the Whiteness of the page had become the default tone of protagonists, then these artists were claiming for their subjects the same status.
If their illustrations were presented to the economic research bureau’s algorithm, however, they would be offered up as casualties of pervasive racism and colorism in children’s illustration, while Billings’ illustrations in “Uncle Tom’s Cabin” would be celebrated. Though the NBER sampled over a century of children’s books, their algorithm does not account for the changing historical contexts against which these illustrations were produced and the nuances of their confrontations with the racial ideologies of their respective periods.
Next-gen artists bring the range, diversity kids deserve
These contexts have informed not only these illustrators’ representative strategies but also the composition of the children’s literary canon. The Newbery medal was first awarded in 1922 to Hendrik Van Loon’s “The Story of Mankind,” an illustrated history of civilization that enshrined a racist, Eurocentric point of view into American children’s education. The first Black recipient of the Newbery medal was Arna Bontemps, who received the award for his illustrated children’s history book “Story of the Negro” in 1949.
The Caldecott Medal, similarly, was not awarded to a single Black artist for the first 30 years of its existence, a fact that prompted a group of Black librarians to confer the Coretta Scott King award in 1970. This award was not officially recognized by the ALA until 1982 and not presented annually until 1986.
Reliance on these awards as a significant basis of analysis determined the NBER’s conclusions before the first algorithm was run. We do not need artificial intelligence to know that the ALA was among the most important cultural gatekeepers in 20th-century literary history.
The winners of these awards, however, have never represented the full range and diversity of children’s literature. The representative possibilities introduced by the newest generation of children’s book authors and illustrators give us reason to hope Black and Brown children won’t be rendered as reductive caricatures or mimeographed reproductions, but as richly imagined subjects who can inhabit the full range of fictional worlds.
Emily Gowen is a dissertation fellow at the Boston University Center for the Humanities, the 2022–23 John B. Hench Post-Dissertation Fellow at the American Antiquarian Society, and an affiliate at Boston University’s Center for Antiracist Research.