The present volume has explored a number of themes that are inextricably linked to discriminatory discourse. As made clear from the beginning, its main objective has not been to provide an exhaustive account of hate speech, but rather to show that research from the perspective of discourse analysis can shed further light on this social phenomenon that has, unfortunately, been increasingly gaining momentum lately. What we hope to have made evident through all the preceding argumentation is that there is much more to hate speech than meets the eye. This is especially true in the online setting, which is typically characterised by intense emotional content and expression, especially when it comes to posts made by the general public in reaction to current affairs (cf. Yus 2011; Musolff 2017; Santana 2014).

The central objective of this volume has been to show that legislation against hate speech in the EU may be an effective first step towards combatting the phenomenon, but it might not be adequate on its own to contain the present situation. This is because hate speech has multiple ways of being expressed. In this volume, we have identified several strategies of Othering that can be used to express such an unfavourable position towards members of a minority: categorisation and stereotyping, hate concealed as patriotism, metaphorical language, sarcasm, allusions and constructed dialogue can all be ‘subtle’ ways in which discrimination emerges in public discourse. And while we are not in a position, as linguists, to suggest that such strategies belong to the category of prosecutable hate speech, we think that it is safe to assume that they do form part of what we have dubbed soft hate speech in Chap. 1. The reason for this is that all these strategies create a fertile ground for hard hate speech to emerge since they slowly but steadily legitimise discrimination and potentially even violence against vulnerable groups. As Waldron (2012: 4) puts it,

[the] sense of security in the space we all inhabit is a public good, and in a good society it is something that we all contribute to and help sustain in an instinctive and almost unnoticeable way. Hate speech undermines this public good, or it makes the task of sustaining it much more difficult than it would otherwise be. It does this not only by intimating discrimination and violence, but by reawakening living nightmares of what this society was like – or what other societies have been like – in the past. In doing so, it creates something like an environmental threat to social peace, a sort of slow-acting poison, accumulating here and there, word by word, so that eventually it becomes harder and less natural for even the good-hearted members of the society to play their part in maintaining this public good.

Indeed, in all the national corpora that have been collected for the C.O.N.T.A.C.T. project, one can see that such indirect means of calling for discrimination against a minority group are commonly encountered across the board. And even though we identify them as ‘soft’ hate speech, they may have the same perlocutionary effect as hard hate speech. To mention an example, well-known conceptual metaphors such as parasites were often used in Nazi speeches with a view to outcasting and demonising the Jewish community, and have been found to be conducive to the abhorrent behaviour to which this particular minority group was subjected during the Second World War (cf. Billig 1977; Wodak and Richardson 2013). Such metaphors seem to be reworked into apparently ‘mild’ negative qualifications in our data too; and although they may seem mild at first, they have the potential to lead to some destructive behaviour too. For instance, in both the Lithuanian and Cypriot datasets, homosexuals are described as being ‘selfish’ or ‘useless’ in statements that may seem puzzling at first. However, when looking more closely at the relevant comment threads, these characterisations seem to be based on the assumption that same sex couples cannot straightforwardly reproduce and therefore do not contribute to society at large. Such statements can be taken to communicate many inferences, with a simple one being that since same sex couples benefit from the community in which they live without contributing new members to it, they behave like leeches, which are after all a type of parasite. Therefore even though the terms ‘leeches’ or ‘parasites’ are not used, similar actions and reactions could emerge on the basis of such inferences (cf. Baider forthcoming). Working on creating counter narratives based on these inferences may then succeed to debunk the implied consequences as well as the fallacies conveyed therein. Clearly, further research should focus on the inferences that can be drawn from other comments in the relevant threads and their possible contribution to a potential escalation of violence in order to test the above hypothesis (cf. KhosraviNik 2017).

Linguistic creativity in instances of both hard and soft hate speech is also a common characteristic across the board. For instance, we observed in the Italian data the linguistic and ironic calque of ‘different sexually-oriented’ (‘diversamente orientati sessualmente’), which echoes the Italian phrase for ‘disabled people’ (‘diversamente abile’), implying therefore that homosexuals are incapacitated in some respect. Moreover, researchers working on online exchanges among members of extreme-right groups have noticed that they are often prone to using codified language, which will enable them to avoid being tracked by automatic hate speech detection software: e.g. using ‘juices’ instead of ‘Jews’, ‘jewrope’ instead of ‘Europe’, etc. (cf. Baider and Constantinou 2017). Irony and humour are also sure ways to get the attention of further readers and build connivance outside the already convinced circle of followers. In any case, this ‘playful’ dimension of hate speech would also be well worth exploring; if nothing else, it would enable us to tweak already existing software for the automatic detection of hate speech so that they also take such comments into account.

Apart from the identification of this repertoire of strategies, which is arguably not exhaustive, we have also attempted in this volume to show that the general public’s perception of what actually constitutes hate speech and how it should be regulated is far from uniform. So, even though the young people we interviewed are, quite expectedly, ardent supporters of freedom of expression, they still generally feel that hate speech is an issue that needs to be combatted. Yet, many do not seem to be sensitised towards what exactly hate speech is and also justify at times the negative sentiment of the general public towards a particular minority. To our mind, this carries two implications. On the one hand, even though the EU is pushing for the regulation of hate speech at a transnational level, it seems sensible for it to leave space for each member state to target hate speech within its national context, with its own particularities and needs. On the other hand, it seems that while legislation does help, it is not enough on its own to contain the situation, since most participants in our interviews showed ignorance of the relevant laws and repercussions for the expression of hate online. What emerged from the interviews, instead, was that the most effective weapon in the fight against hate speech is education, broadly construed.

Against this background, when it comes to policy-making, the C.O.N.T.A.C.T. consortium can therefore make two recommendations. For one, it is necessary to conduct extensive research on the different forms that hate speech can take, both online and offline, as well as the underlying reasons for the emergence of such speech. It may sound banal to point this out, but it is only through the profound understanding of these reasons in the first place, at both the national state and transnational levels, that effective policies of inclusion can be developed; and this is something that seems to be often disregarded by those in charge. Then, it is equally, if not even more important for the general public to develop an awareness on matters of discrimination. This is something that can only be accomplished through wide-reaching awareness-raising events, a responsible approach to the relevant issues by the media, and, of course, the establishment of an agenda that promotes inclusion and tolerance at all levels of education. The latter has also been pointed out in a very recent European Agency for Fundamental Rights press release, according to which, “promoting inclusion and mutual respect through education and strong positive narratives are essential to prevent incitement to hatred and counter hate speech in the digital age” (FRA 2013: 1).

In closing, we hope to have shown that linguists have an important role to play in this picture (cf. Olsson and Luchjenbroers 2013). Since it is intention that lies at the very core of most legal definitions of hate speech, contextualising and qualitatively analysing such speech seems central to not only tackling this complex phenomenon but also to safeguarding freedom of expression on the many platforms that the internet offers. We therefore believe that this is an endeavour that can only be accomplished by encouraging collaboration and constructive dialogue between policy makers, legal practitioners, linguists and computer scientists specialising in the automatic detection of hate speech, as well as involving higher education institutions more directly in the implementation of the relevant EU agency directives.