Helping journalists, producers and conference planners find the female guests, speakers and expert sources they need.

The future of women’s voices in media: an interview with Gender Gap Tracker’s lead SFU researcher

SFU News with Maita Taboada 28 March 2019

As a woman, do you feel underrepresented in the media? Does it seem like the majority of people being quoted are male? While 21st century society moves toward gender equity, is this reflected in the news?

Those questions were answered earlier this year with the launch of Informed Opinions’ Gender Gap Tracker powered by Simon Fraser University. In real time, this online tool tracks the number of women and men quoted in major news platforms in Canada — and the disparity is significant.

On March 26, 2019, SFU hosted “Tracking the Gender Gap in Canadian Media — Taking a Big Data Approach,” a free public event held at SFU’s Big Data Hub. The Gender Gap Tracker’s lead researcher, SFU linguistics professor Maite Taboada presented the history, functionality and impact of the tool. A team of SFU undergrads, graduates and postdoctoral fellows from SFU’s Discourse Processing Lab (Fatemeh Torabi Asr, Vasundhara Gautam, Junette Gonzales, Mohammad Mazraeh and Alexandre Lopes) helped build the Gender Gap Tracker. Most of the research team joined Taboada to answer audience questions.

SFU News caught up with Taboada to learn more about her research, the collaborative effort behind the Gender Gap Tracker, how this tool will benefit society and what it means for the future of women’s voices in media.

Q: As the Gender Gap Tracker’s lead researcher, tell us about your work in computational linguistics, the Discourse Processing Lab and the expertise you needed to deliver on this project.

For a long time now, I have worked on research spanning computer science and linguistics. I am interested in understanding how language works—in particular, how we put sentences together to make coherent discourse. I am also very interested in the language of evaluation: how we use language to convey opinion. I have applied the results of those theoretical investigations in applications such as sentiment analysis, moderation of online comments and detection of fake news. I find this work—going back and forth between theory and applications—to be really satisfying, because I want answers to the big questions, but I also like to be able to make an impact with my research.

Q: Tell us about how the collaboration between SFU and Informed Opinions? Why was it a good fit?

Informed Opinions approached SFU about building this tool. Initially it was just an informal set of conversations, but the goal of Informed Opinions and the passion of its founder and catalyst, Shari Graydon, to bring gender equality to the news is contagious, and soon everyone in my lab was behind this project.

Q: The phrase that is being used to describe the Gender Gap Tracker is “closing the gap in female media representation.” Can you provide some examples of how you see this tool benefitting society?

We can bring meaningful change to newsrooms by making them aware of the lack of equality in their sources. We want to start a conversation about truly diverse sources, not just with respect to gender, but all forms of diversity. Personally, I would like to have news sources be fully representative of Canadian society.

At the same time, we are making the data available to all researchers. There are lots of interesting questions that can be asked, such as what areas of the news women are quoted more often, how much space they are given and how we portray women’s opinions.

Q: The Gender Gap Tracker uses big data to analyze women’s voices in media. How exactly does this work?

The Gender Gap Tracker works in three stages. The first stage collects articles from online news sources and stores them in a database. This is done by scraping the listed news sources through either their RSS feed or via a media-oriented scraping tool in Python.  

The second stage takes the collected articles, identifies the people in the article and pulls out whether they are speaking or not, and what their gender is. Various natural language processing tools are used for this stage. This involves doing Named Entity Recognition (finding the people mentioned) and finding quotes by analyzing the structure of sentences.

The third stage draws on the information produced from the first two and produces the website and visualizations.

Q: What’s next for the Gender Gap Tracker? We hear a Francophone version is being developed.

Yes, we are working on a French-language version. This is a bit more challenging, because we do not have a lot of French expertise within my team, so Alain Désilets, a researcher at the National Research Council, is helping us. It is also challenging because there are fewer French language resources freely available.

That brings up an important point: we couldn’t have done this without open-source software. Most of the tools that we use—from scraping to language analysis—are open source, and we are making our own code available to the community.

Q: What advice do you have for other social scientists embarking on their own research projects?

This project was so unique, that there were so many ways in which it may not have happened. I would say that it is important to be willing to take risks. This is one of the thrills of research.