Imagined social structures: Mirrors or alternatives? A comparison between networks of characters in contemporary Dutch literature and networks of the population in the Netherlands

Abstract

We study whether patterns of segregation and homogeneity in personal networks in the Netherlands are comparable to the network structures of characters in modern Dutch literature and examine the degree to which social divides in terms of sex, education, ethnicity, and age are reflected in literary novels. A representative sample of people living in the Netherlands (the Survey of the Social Networks of the Dutch, SSND 2014, n = 967 respondents and 3424 network members) is employed as well as a representative sample from Dutch literary fiction published in 2012 and submitted to the Libris Prize (n = 170 books and 1292 characters). While controlling for author characteristics, our multilevel regression models reveal that networks in books are more diverse in terms of ethnicity, education and age. However, higher levels of education show more closure in books than in the actual world, and the same holds for characters with a higher age. We conclude with a discussion of the relationship between literature and society. IntroductionIn Dutch society, like in other modern western societies, severe social divides constitute challenges for the future. These are the divides according to differences in educational levels, ethnic background, age, and sex. People with different backgrounds in these regards are out of each other's sight and they have only little contact with each other. As a result, their networks do not overlap, and the social worlds of different groups are largely separated. In social science, divides are studied as patterns of social segregation and homogeneity in networks and it is argued that strong patterns of segregation hamper intergroup solidarity and diminish social cohesion (e.g. Briggs, 2001;Markovsky & Lawler, 1994;Pettigrew, 1998). Importantly, if social divides recur in cultural domains, such as in fiction writing, social cohesion is probably even more threatened than actually thought, because the actually existing divides are not even bridged in the imagined social structures. Hence, the degree of network segregation in fiction writing can be perceived as an additional indicator of the strength of segregation and divides in the actual world.This paper examines this issue and studies the question whether network patterns with regard to homogeneity and segregation of characters in books reflect network patterns of people in the Netherlands. Much research shows that people's networks are considerable homogenously composed in terms of important social background characteristics such as age, race, education, and sex. How is this in contemporary fiction writing? Is it likely that two befriended characters in a novel differ in ethnic background and/or education? And how likely is it that friends in a novel diverge in age? We employ two sets of data, a national representative dataset on

Character Constellations

“…Currently, the Libris dataset has been augmented to a total of 2,137 characters and thus covers a larger portion of character interactions in the books, taking into account a fair number of less visible side characters as well. Moreover, the methodical setup of the present chapter differs from the research in Volker and Smeets (2019). The results reported in this chapter broadly confirm the findings reported in this previous study.…”

“…The positive mean assortativity coefficient for descent suggests that segregation on this demographic factor is also apparent in present-day Dutch literary fiction. However, this segregation pattern is probably less dominant than it is in present-day Dutch society, which is demonstrated by a previous comparison of a part of the present dataset with a dataset on real-world networks of Dutch people showing that the fictional networks are significantly less segregated by means of descent than real ones (Volker & Smeets, 2019). Divides in society are also caused by homophilous associations with regards to education (Marsden, 1987;Verbrugge, 1977).…”

“…There is one study that explores the idea of gender homophily in networks of characters in English-language fiction in depth (Kraicer & Piper, 2019), which is closest in nature to the analyses carried out in this chapter. Preparatory work for the present research was done in Volker and Smeets (2019), 2 in which the Libris dataset was compared with actual personal networks of Dutch people through the framework of homophily. In this comparative analysis, we found that the character networks in the Libris corpus are less homophilous in terms of descent, education, and age than actual networks in Dutch society.…”

Relationship Change, Network Change, and the Use of Single Name Generators in Longitudinal Research on Social Support

“…The six-name generators used here were chosen to represent significant dimensions of social support, but just as single name generators can exclude active network members who did not provide one support, these six name generators could exclude alters who are active in the network or important in respondents’ lives but did not provide these specific social supports. Data collected with more than six name generators (e.g., Fischer 1982b, 2018; Völker and Flap 2002) could plausibly use similar methods to show that using six name generators could result in some ties being mistakenly classified as dormant. There is no unambiguously and objectively correct way of delimiting which kinds of support should be included in a person’s support network.…”

Character Constellations

“…The analysis exclusively focused on the categories of friend and enemy, and ignored the other categories in case of double, triple, or more relational annotations. 9 Volker and Smeets (2019) compare the networks of people living in the Netherlands with the networks of characters in the Libris corpus. The research does not focus on social balance theory, but the dataset of actual personal networks does seem to conform to an overrepresentation of social balance as opposed to social imbalance.…”