Analyzing gender bias in children’s television shows

Abstract

In this research gender bias was analyzed in two of the most watched children’s television shows in the Netherlands, Sesamstraat and Het Klokhuis, due to the impressionability of the target audience of these shows. Automated speech recognition has been used to generate a textual corpus of 20 episodes per year per show in the time period from 2011 until 2020. In this research, bias is defined as the existence of prejudice towards certain (groups) of people. The (groups of) people this paper focuses on, are distinct in gender, male and female. A model calculating word distances (Word2Vec) was used to calculate the distance between words from six LIWC categories and two lists containing male or female related words. Using linear regression, the change over time in gender bias in each LIWC category was calculated per year to show in what aspects gender bias has changed in these two television shows over the last decade. Aside from the LIWC categories ‘Groom’ and ‘School’ in which a slight decrease in the slight bias towards women was found, no trend was found in the other categories. This could mean several things, for example 1) there is no trend in gender bias in these shows over the last decade, or 2) the corpus was of insufficient size to get proper results. The limitations of this research make the research valuable as a steppingstone towards further research into the field of gender bias related to children’s television shows.