Topography of Character's Body: a Case of Russian Children's Literature

This poster presents quantitative data on the representation of characters’ bodies in the corpus of Russian children’s literature, visualized as a series of body heatmaps. This visualization is inspired by the research on the topography of social touching in different cultures
(Suvilehto et al., 2015). But unlike physical touch data in that study, our literary topography represents what parts of a character’s body the author’s pen is allowed to touch. The central question of the research is how the selectiveness of authors in describing their characters’ bodies is related to the demographic features of characters, such as gender and age. Our data reveal gender differences in the representation of female and male characters, point out differences in the representation of adult and child characters, and provide comparative material for the study of character embodiment in literary fiction.

Corporeality is a fundamental attribute of fictional characters. In literary criticism, references to character’s face and body features as a means of characterization have a long tradition of study under the rubric of
literary portrait (
Eder et al., 2010). Yet only recently, with the advances in computational methods and literary corpora appeared quantitative data that suggest large-scale regularities and historical trends in the representation of characters’ bodies. These trends can be related to transformations both in society and in literary taste. For instance, in the corpus of 19th century English novels the frequency of body parts mentions was observed to grow over time, as the urbanization progressed and the attention of the authors shifted from direct and evaluative characterization to physical details of characters (Heuser and Le-Khac, 2012). Presumably, representation of characters’ bodies depends on social stereotypes.
Plenty of research confirmed that depiction of male and female characters in children’s books is strongly affected by the normative view of gender roles in a society, even in contemporary literature that is supposed to be progressive in terms of gender
(Anderson and Hamilton, 2005). The usage of words denoting body parts was also shown to be associated with character’s gender and to change with time on a large scale in the corpus of English-language fiction (Underwood et al., 2018). We are not aware of any similar quantitative work on Russian children’s literature. The prominence of children as characters in children’s literature allows us to address the question of differences in the representation of bodies of children and adult characters.

We base our research on the
corpus of about 500 works that includes prosaic fiction for children and youth written by Russian authors during 1930—2000s in broadly realistic genre (

http://detcorpus.ru

), which is a sample of available digitalized texts. To obtain frequency data on the references to character’s bodies in the corpus, we have to compile a list of words denoting body parts and to locate the occurrences of these words that refer to certain characters. First, we automatically extract the list of characters using procedure similar to that of BookNLP (Vala et al., 2015), adapted to Russian and with some modifications. We then build a series of classifiers to select only human characters, and detect their gender and age category. Finally, references to body parts are attributed to characters with the use of syntactic parsing and coreference resolution. The resulting data allow to build heatmaps for various groups of characters, and to test statistical hypotheses about their differences. Our research is still work in progress, but preliminary results show that the representation of characters’ bodies varies by gender. However, gender differences were not the same for children and adult characters. Comparative analysis of contexts in which body parts of different categories of characters are mentioned, showed that some frequency peaks correspond to the formulaic features in portrait descriptions. For example, boy characters very often scratch their napes.

The primary contribution of our work is that we present empirical data on the representation of characters’ bodies in the Russian children's literature, that allows to quantitatively explore the association of normative images of gender and childhood with its literary representation. These data may serve as a comparative material to the existing quantitative research in this field in English and American literature. Furthermore, the analysis of contexts of body parts mentions suggests the existence of formulaic features of character portraits that may be seen as a part of literary tradition.
Our heatmap visualization is meant to draw attention of researchers in digital humanities to the prospects of further comparative and historical research of body as a specific aspect of literary character.

Bibliography
Anderson, D. A. and Hamilton, M. (2005). Gender Role Stereotyping of Parents in Children’s Picture Books: The Invisible Father.
Sex Roles. 52 (3), 145–151.

Eder J., Jannidis F. and Schneider R. (2010).
Characters in fictional worlds: Understanding imaginary beings in literature, film, and other media. Walter de Gruyter.

Heuser R. and Le-Khac L.A. (2012). Quantitative Literary History of 2,958 Nineteenth Century
British Novels: The Semantic Cohort Method
. Stanford Literary Lab Pamphlet
Series
. URL: https://litlab.stanford.edu/LiteraryLabPamphlet4.pdf.

Suvilehto, J. T., Glerean, E., Dunbar RIM, Hari, R. and Nummenmaa, L. (2015). Topography of social touching depends on emotional bonds between humans.
Proceedings of the National Academy of Sciences. 112 (45): 13811–13816.

Underwood, T., Bamman D. and Lee S. (2018). The Transformation of Gender in English-Language Fiction.
Journal of Cultural Analytics. Feb. 13, 2018.

Vala, H., Jurgens, D., Piper, A. and Ruths, D. (2015). Mr. bennet, his coachman, and the archbishop walk into a bar but only one of them gets recognized: On the difficulty of detecting characters in literary texts
. In Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, pp. 769-774.