Research and advocacy of progressive and pragmatic policy ideas.
A study of responses to Malaysian hate speech samples, towards a framework for hate speech categorisation and response.
By Jia Vern Tham & Nelleita Omar25 November 2020
Hate speech has become a major concern in multi-ethnic societies around the world and Malaysia is not exempt. Slurs and hateful speech targeting ethnic and religious groups, from comments by ordinary individuals to statements by prominent figures, are prevalent.
The long-term consequences of hate speech could be extremely serious. Studies in other countries have shown that what appears to be merely ‘uncivil discourse’ could erode inter-ethnic trust over the long term to the point of instability.
Given the risks from hate speech in Malaysia, we argue that the current institutional and policy response in Malaysia is insufficient. Firstly, the understanding of what constitutes hate speech needs to be localised, or it runs the risk of being divorced from local values and historical context. Secondly, improvements are necessary to deal with the whole range of hate speech expressed today. Existing laws that deal with deterrent and punitive action against incitement need to be complemented with more pre-emptive and rehabilitative measures against less serious levels of hate speech.
To see if we could categorise hate speech systematically in Malaysia (and beyond), we embarked on research that sought to understand how ordinary Malaysians assess the seriousness of a range of offensive and hateful speech samples. Using the findings of the research, we then proposed a holistic framework by which to categorise and manage hate speech in Malaysia.
Our three core research questions are:
What is hate speech to Malaysians? In particular, how does the public discern between ‘less serious’ vs. ‘more serious’ speech, and whether there are differences across the major ethnic groups of Peninsular Malaysia*
Following from the above, what might a people-informed hate speech intensity categorisation look like for Malaysia?
What could be appropriate societal responses to different intensities of hate speech?
*Due to time, budget and conceptual constraints, this study focused on Peninsular Malaysia only. A separate study on Sabah & Sarawak’s assessment of hate speech would be a valuable extension of research in this area.
For the purposes of our study, we define hate speech as speech that has public impact (i.e. not private exchanges) and speech deemed offensive or hateful targeting a group trait such as race or religion.
Both quantitative and qualitative methods were employed. The study required respondents to indicate how they reacted to 15 samples of offensive / hateful speech.
The speech samples were chosen to correspond to Peninsular Malaysia’s most significant cultural-political fault lines. 9 speech samples were selected as targeting race (Malay, Chinese and Indians) and six speech samples were selected as targeting religion & culture (Muslims vs. non-Muslims and non-Malays).
For the quantitative phase of the study, a face-to-face survey of 200 Malays, 200 Chinese and 200 Indians was conducted in major urban areas throughout Peninsular Malaysia. Respondents could choose to answer the survey in English, BM or Mandarin. Respondents were asked to score the 15 speech samples based on how offensive (O), fearful (F) and threatening to society (T) they found the sample to be.
For the qualitative phase of the study, three focus groups, each consisting of 7 Malay, Chinese and Indian participants respectively were convened over two days in Klang Valley. Focus groups were conducted in the participants’ language of choice. Participants were asked to rate the same speech samples against a simpler 3-level scale: L1 for Not Very Serious, L2 for Moderately Serious and L3 for Very Serious. The reasons behind their ratings were probed by a skilled moderator, to aid in the analysis for common themes.
Based on the responses of both the survey and the focus group, we found that speech samples can be classified into three categories, with each category exhibiting consistent general characteristics.
The first category consisted of speech samples that obtained relatively high scores from survey respondents overall, as well as speech samples that obtained high scores from the targeted group. Based on the qualitative responses of focus group participants to samples considered ‘very serious’, it was clear that content is a key factor, in two ways. First, hate speech that disrespected religion, threatened violence or encouraged hatred against a group were considered as very serious across all ethnicities. Second, hate speech that demeaned the humanity or status of a particular ethnicity were found to be very serious by the targeted group. The theme of ‘context’ was also stronger for this category of speech compared to other categories – many of the speech samples were seen to have the potential to trigger general unrest in society.
The second category consisted of speech samples that obtained middling scores in the survey. Based on the qualitative responses of focus group participants to samples that were considered middling or ‘moderately serious’, we surmise that content is again a driving factor. Here, the content or substance of the message is seen as offensive but not intensely so due to various factors, such as whether the content has become ‘normalised’ to the respondent. Respondents also felt that speech samples can be ‘promoted’ or ‘demoted’ to this category of speech by the presence of mitigating or aggravating factors such as the status of the speaker, the perceived intent behind the speech, the reach of the message, amongst others.
The third category consisted of speech samples that were rated relatively low in intensity by the majority of survey respondents. Based on the qualitative responses of focus group participants to speech samples rated as ‘not very serious’, we surmise that the nature of these speech samples’ content is again key. The content or substance of the message is seen as negative in general but not to the extent of causing offence or fear – for example, the message is perceived as nonsensical rather than insulting. The impact of the content is also lessened by the presence of mitigating factors such as the ‘ordinary’ status of the speaker and the limited reach of the message.
Based on the above, we put forward that the capacity of a speech’s (or act’s) content to cause feelings of offence, fear and threat to societal stability is the first key factor in determining the speech’s seriousness in terms of hatefulness. The second key factor is the presence of mitigating or aggravating factors which includes the status of the speaker, the perceived intent behind the speech, the reach of the message and the unique impact of the speech given Malaysia’s historical, cultural and political context (see Figure 2).
To address the full range of hate speech in Malaysia requires us to have a more comprehensive, cohesive and proportionate approach.
Existing legislative responses and police enforcement towards hate speech in Malaysia should be more clearly focused on criminal levels of hate speech i.e. ‘very serious’ speech that by its content and other factors is an incitement to violence or mass violence. Less serious forms of hate speech, which range from mildly provocative speech to incitement to marginalise, should be addressed by more pre-emptive and rehabilitative measures such as public education and the articulation of codes of conduct (see Figure 1).
Based on the findings of our research, we propose a hate speech management framework that consists of (i) transparent categorisation of hate speech according to levels of intensity, and (ii) a whole-of-society response.
The experience of the research process shows that it is possible to develop a categorisation framework outlining the general characteristics of different levels of hate speech seriousness. As a first principle for any categorisation framework to be developed, be it for an organisation or for national policy, we propose that it be kept simple but sufficient. Starting with a 3-point categorisation scale, from ‘not very serious’ to ‘moderately serious’ to ‘very serious’ enables sufficient differentiation and avoids becoming too complex or unwieldy* (see Figure 2).
*Whether ‘very serious’ hate speech should be further defined as ‘incitement’ could be a future extension of research work in this dynamic subject.
The second principle we propose for hate speech categorisation is to prioritise the targeted group’s perception or rating of the hate speech’s seriousness. Unsurprisingly, our research shows differences in the level of offence and fear experienced by targeted groups vs. non-targeted groups. To avoid underestimating a speech or an act’s potential to cause societal harm, the categorisation of a statement’s seriousness should be weighted more heavily on the targeted group’s perception.
Building from the above, we propose a response framework that engages the whole of society (see Figure 3). The response framework also needs to take into account the scale and the scope of implementation; for example, the resources required to manage hate speech for a school or company would be very different from a media platform. Lastly, the response has to be both proportionate (i.e. commensurate with the offence) and constructive (i.e. include rehabilitative measures).
‘Not very serious’ hate speech could be mainly addressed in pre-emptive and rehabilitative ways by civil society-led actions focused on cultural awareness and building bridges between communities. Indeed, fledgling civil society efforts are already in place in Malaysia, which include diverse programs and advocacy by Architects of Diversity, Article 19, Tribeless, Pusat KOMAS, CIJ and many others, though there are still few specific references to hate speech. Apart from such efforts, we also advocate for organisations to start articulating and integrating codes of conduct on hate speech and civil discourse within relevant policies (for our other recommendations, see Figure 4).
‘Moderately serious’ hate speech could be addressed by educational or bridge-building programs but for cases that require mediation or investigation, we propose a national arbiter function, such as in a commission or a tribunal, that provides an avenue for resolving civil complaints related to hate speech. Apart from case resolution, this function could also set out national hate speech guidelines that could help inform the codes of conduct of various organisations, from companies to media to political parties. Examples of such a function in other countries include the Australian Human Rights Comission, the New Zealand Human Rights Commission and the City Committees under the Anti-Hate Speech Ordinance in Japan.
Certain cases of ‘very serious’ hate speech could be dealt with by the aforementioned national commission or tribunal but cases that pass a clearly defined criminal threshold (e.g. incitement to violence) should be dealt with via the criminal justice system. The Sedition Act, the Penal Code Sections 298A and 503-505, the Communications and Multimedia Act and the Printing Presses and Publications Act all have provisions that broadly address hateful speech. However, the existing definitions and underlying assessments used are still very broad and therefore lends itself to the risk of misuse or being disproportionately punitive. We therefore advocate the establishment of a national framework for hate speech categorisation and response which, at the minimum, clarifies the difference between criminal and non-criminal acts of hate speech (for our other recommendations, see Figure 4).
The full range of hate speech should be addressed if we are to balance freedom of speech with the obligations of living in a diverse multi-ethnic society. We hope this study contributes towards the body of work that informs guidelines and frameworks for managing hate speech in Malaysia, be it for a department, an organisation, a group of concerned citizens or for the country.
The Centre extends heartfelt thanks to research consultant Dr. Murni Md. Nor and project sponsors Westports Berhad, MARI and USAID for their invaluable support throughout the study.