Studying Muslim Stereotyping through Microportrait Extraction

Studying Muslim Stereotyping through Microportrait Extraction Antske Fokkens♣, Nel Ruigrok♥, Camiel Beukeboom♦, Sarah Gagestein♠ and Wouter van Atteveldt♦ ♣ Computational Lexicology and Terminology Lab. Vrije Universiteit Amsterdam, the Netherlands. ♠ LJS Nieuwsmonitor. Amsterdam, the Netherlands ♦ Department of Communication Science. Vrije Universiteit Amsterdam, the Netherlands ♥ Taalstrategie. Amsterdam, the Netherlands. {antske.fokkens,c.j.beukeboom,w.van.atteveldt}@vu.nl,nelruigrok@nieuwsmonitor.org,sarah@taalstrategie.nl Abstract Research from communication science has shown that stereotypical ideas are often reflected in language use. Media coverage of different groups in society influences the perception people have about these groups and even increases distrust and polarization among different groups. Investigating the forms of (especially subtle) stereotyping can raise awareness to journalists and help prevent reinforcing oppositions between groups in society. Conducting large-scale, deep investigations to determine whether we are faced with stereotyping is time-consuming and costly. We propose to tackle this challenges through the means of microportraits: an impression of a target group or individual conveyed in a single text. We introduce the first system implementation for Dutch and show that microportraits allow social scientists to explore various dimensions of stereotyping. We explore the possibilities provided by microportraits by investigating stereotyping of Muslims in the Dutch media. Our (preliminary) results show that microportraits provide more detailed insights into stereotyping compared to more basic models such as word clouds. Keywords: stereotyping, digital social science, text analysis 1. Introduction Research in social sciences and computational linguistics shows that language use plays a crucial role in creating stereotypes about social categories (like Muslims). Sev- eral fields within Social Sciences have studied how stereo- typic beliefs are reflected in language and consequently be- come shared knowledge (Ruscher et al., 2005; Klein et al., 2008; Beukeboom, 2014, among others). These studies show that people (unconsciously) express stereotypic be- liefs they hold about described individuals or groups in both linguistic content (what information is provided) and form. Computational linguistic studies show that language mod- els learned from corpora reflect such human bias, including stereotypes (Caliskan et al., 2017, e.g.). Insights from computational linguistics and communication science are complementary. Distributional semantic mod- els applying purely mathematical models to large corpora of text reveal that biases are present in the texts. Yet, these abstract language models do not provide means to reveal how bias is expressed in language exactly. Social science studies fill this gap, for example by studying the frames that media use in their coverage (Entman, 1993). These studies, however, only cover limited sets of data as they often rely on experimentally manipulated sentences or manual anno- tation to study biased language use. Moreover, the need for double-blind annotations on each new set under investiga- tion makes it challenging to study the expression of stereo- types over time and across a high variety of sources. Au- tomatic approaches that can help tackle this problem can therefore greatly enhance the possibility of investigating stereotyping in natural language. Previous work has shown that NLP can be used to identify various forms of fram- ing (Card et al., 2016). These general classifiers perform deeper insight than pure …

Content truncated to 500 words