It is nearly impossible to find a single female journalist who would repudiate facing incivility and hate speech on social media. In fact, according to a study by Committee to Protect Journalists (CPJ), it is the largest safety concern for female journalists, because as much as 85% of respondents said they feel less safe than they did five years ago. Such harassment isn’t limited to one country or culture, slowly it has become pervasive and sadly, Pakistan is no exception. In reality, it is worse in many ways because other countries at least acknowledge it a problem, but here few are interested to even talk about it.
But on 12th August 2020, the abuse must have reached its pinnacle, which is why some 51 Twitter using prominent female journalists and writers jointly released a statement asking the government to restrain its officials and affiliates from this online attack. After reading the statement, I was prompted to verify the veracity of their claims. Hence, the only two questions I was interested to investigate were (1) if the government or its sympathizers are involved in this campaign, and (2) do male journalists are subjected to a different kind of conversation compared to female journalists?
Analysis on 68,352 different replies received by male and female journalists
To find the answers to my questions, I decided to base my work on empirical evidence and facts so that no bias clout my assessment. I began by randomly choosing 16 prominent male journalists on Twitter and the same number of female journalists that were among the signatories of the statement. Next, I used Twitter’s REST API through R programming language’s package “rtweet” and collected all the replies received by both male and female journalists in my sample.
In total, I managed to collect 68,352 replies both in English and Urdu between 4th to 13th August 2020. Out of all these replies, I filtered only those that were in English (mainly because we still have not developed tools that can computationally process Urdu), and to have equal representation from both groups, I randomly selected 4,000 replies sent to both male and female journalists.
The analysis indicates that usually male journalists receive a lot more responses compared to females but on 12th August, the day they published their statement, the number of replies significantly increased (see Figure1). But I was interested in what was said in those responses. So, to do that, I took advantage of Natural Language Processing (NLP) and calculated the co-relation of words used in the replies for each group. Though most words in Figure 2 reflect nothing vile, especially for male journalists, however, the words such as “lifafa”, “lies”, and “oratcard” were more likely to be used in replies sent out to female journalists.
Figure 1: Frequency of replies to both female and male journalists
Figure 2: Correlation of words used in the replies
When I plotted the co-occurrence of words in a network, the picture became clearer because then, I could see how certain words were always associated with women journalists, and the strength of co-occurrence signified by the color makes apparent the context of words.
Figure 3: Network of co-occurring words
After getting the answer of my first question that yes, women are more likely to face hateful speech on social media I started to find if government and its sympathizers are behind such online incivility. To do that, I began to look in to the profiles of those users who have sent replies to journalists in my sample. The analysis revealed that 5,690 users sent in total of 8,000 randomly selected replies indicating some users may have either responded to both male and female journalists or replied more than once. Out of these unique users only 4,180 users have written something in their profile description. My plan was to identify those users through their Twitter profile description.
Figure 4: Words used in the profile descriptions of responders to both male and female journalists
People tend to reveal their qualifications levels through social media profiles
Looking at words used in the profiles of people who sent replies to either male or female journalists gives some hints about their identity but does not directly attribute them to be a part of the government or their sympathizers (provided what they have written in their description is true). If anything, Figure 4 shows often people through their profile description reveal their level of education “MPhil”, interests’ “journalism”, “cricket” or career “editor”, “bank” and “assistant”.
However, the network analysis of co-occurring words in Figure 5 offers some more insights and lend support to the claims made by female journalists in their statement. The most common words that often appear together in the profiles of people who responded to male and female journalists are “Imran”, “khan”, “cricket”, “Pakistani”, “PTI”, and “Insafian”. Thus, I found the answer to my second question, which also provided support to the claim of female journalists.
Figure 5: Networks of co-occurring words in responder’s profile descriptions
As a citizen of an independent country who is constitutionally granted the freedom of expression is free to express his or her opinion. However, no privilege allows anyone to be uncivil or hateful merely because other person holds difference of opinion on any subject matter. Most importantly, not the journalists and definitely not the female ones, because neither do we want them to be fearful of reporting on issues that matters nor it is part of our culture and values.
And as for the officials and sympathizers, the incumbent government is already battling at numerous fronts and by attacking media you are not doing it any favors. Besides, being insulted and threatened online should not be a part of anyone’s job, and we can certainly do better.
Waqas Ejaz is an Assistant Professor at the Department of Mass Communication in the National University of Science and Technology (NUST). He can be reached at: firstname.lastname@example.org