What is hate speech and how it can be controlled
Hate speech often comes up in public debate, but what criteria distinguish it from legal freedom of opinion? The 5 point test of the Ethical Journalism Network approaches the hate speech phenomenon through the status of the speaker, the reach of the speech, the goals of the speech, the content and form of speech and the economic, social and political climate. The status of the speaker includes attempts to reach a new audience using sensational claims because the talker has some public position. The reach of the speech includes both the private or public aspect as well as the repetition of the message. A hostile opinion, which has been repeatedly stated indicates a bigger extent of the purpose than once expressed. The content and form of the speech include that the humans may have offensive opinions and this is not wrong as long as it doesn’t incite hate towards some specific group of people. The consequences under the law must also be considered when selecting words. Especially, people tend to use more harsh language when they are speaking anonymously. Economical, social and political climate means the circumstances, where it is especially dangerous to express opinionated claims. The consequences may be more serious during troubled times than during steady ones. To avoid hate speed during critical times, the impact of the speech should be carefully analyzed beforehand. To sum up, the recognition of hate speech depends on several factors concerning expression, context, and circumstances. [1]
Globally, the definition of hate speech depends on the country concerned. Laws restricting public speech vary country by country. Privately, individuals may express even their offensive opinions. They may do so to deliberately provoke feedback from others. This way a person may learn the existing rules. Freedom to express one’s thoughts is important. If the control is too critical, this right may be lost. On the other side, it is very important to be aware of rules and legislation concerning public speech. The problem is, a private conversation may be copied and without permission presented in a wrong context publicly.
The amount of hate speed can be be reduced through moderators who review and approve new posts on social media. In addition to human moderators, some tools have been used to recognize and decrease the prevalence of hate speech. There have been trials to use bots as controllers of unsuitable speech. These haven’t been accepted very well by the users. Some discussion forums have included special meters on web sites indicating the proportion of unsuitable expressions. The speaker can check the meter display to see if the expressions are valid.
Machine learning is increasingly used to recognize the characteristics of hate speech. However, it is difficult to analyze natural speech, because there are so many ways to express the same thing. Occasionally, the speech may accidentally be interpreted as hate speech because of the context. For example, jokes may be misinterpreted as hate speech. Most of these tools recognize and comment on hate speed after the text has been published. It could be better to have tools that could be used to revise a text before publishing it. This kind of tool could inform the legal consequences if the text is published in its original form.
One challenge in preventing hate speech using tools is that the channels of distribution are changing. For example, hate speech sent by email or instant messengers is especially insulting and stressful, although it often is private. One reason for the failure of proper communication may be that electronic communication is a very young phenomenon when compared to face-to-face meetings. It takes time to adopt new circumstances. The availability of hate speech recognition tools may facilitate this development, especially if they can be used voluntarily to revise the message before it is sent.
[1] Five-point test of speech. Ethical Journalism Network. Available: https://ethicaljournalismnetwork.org/resources/publications/hate-speech [Accessed: 30.9.2019].