Search

Concerns over online hatespeech have prompted governments to strengthen social media governance. However, claims by policy-makers and political activists regarding the effectiveness and likely consequences of legal regulations remain largely untested. We rely on qualitative interviews and two expert surveys to examine the behavior of public relations professionals in response to online hatespeech when having the option of using the new user-complaint mechanism under the German Network Enforcement Act (NetzDG). Our findings reveal that strategies depend on whether professionals work at public sector institutions, business, or civil society organizations and political parties. Public sector institutions are likely to report to the platform, but not under NetzDG. Civil society organizations are likely to choose content moderation, counterspeech, and other forms of intervention. Businesses deploy a wide range of strategies. In practice, Germany's procedural approach relying on user-complaint mechanisms to deal with online hatespeech is not used by experts as a means to combat online harassment., Policy & Internet, ISSN:1944-2866

Zusammenfassung Das Thema Hatespeech ruckt immer mehr in den Fokus der Offentlichkeit und der Forschung. Im Gegensatz zu Hatespeech im Internet wird jedoch Hatespeech unter Jugendlichen, die von Angesicht zu Angesicht im Schulkontext ausgeubt wird, kaum beachtet. Hier setzt die vorliegende Studie an, in der Schuler*innen (n = 55), Lehrkrafte (n = 18) und Sozialpadagog*innen (n =16) auf der Basis leitfadengestutzter Interviews dazu befragt wurden, was mogliche Beweggrunde und Motive dafur sind, dass Schuler*innen Hatespeech in der Lebenswelt Schule und online ausuben. Die Ergebnisse zeigen, dass mogliche Beweggrunde fur Hatespeech Angst vor Statusverlust, Gruppendruck, Provokation, Spas, politisch-ideologische Uberzeugung und Kompensation von Frust- und Minderwertigkeitsgefuhlen sind. Daruber hinaus wird verdeutlicht, dass sich hinter diesen Grunden fur Hatespeech oftmals Grundmotive nach Macht und Zugehorigkeit abzeichnen. Die Ergebnisse werden in Bezug auf anschliesende Forschung und praktische Implikationen diskutiert. Schlagworter: Hatespeech, Hassrede, Grunde, Motive, Schule ----- “Then everyone just goes along with it.” A qualitative study on reasons and motives of hate speech among students Abstract Interest in the topic of hate speech has increased steadily in both the public realm and that of scientific research. Seldom addressed, however, is the proliferation of hate speech amongst adolescents, experienced face-to-face in the school context. To this end, the present study interviewed students (n = 55), teachers (n = 18) and social workers (n = 16), using guideline-based interviews to discuss reasons and motives for students practicing hate speech both online and in the school environment. Results showed that reported reasons for hate speech include fear of diminishing status, peer group pressure, provocation, fun, political-ideological convictions, and compensation for feelings of frustration and inferiority. Additionally, reasons for hate speech are often associated with the basic motives need for power and affiliation. The findings are discussed in relation to future research and practical implications. Keywords: hate speech, reasons, motives, school ----- Bibliographie: Ballaschk, Cindy/Wachs, Sebastian/Krause, Norman/Schulze-Reichelt, Friederike/Kansok-Dusche, Julia/Bilz, Ludwig/Schubarth, Wilfried: „Dann machen halt alle mit.“ Eine qualitative Studie zu Beweggrunden und Motiven fur Hatespeech unter Schuler*innen, Diskurs Kindheits- und Jugendforschung / Discourse. Journal of Childhood and Adolescence Research, 4-2021, Online-First, S. 1-18. https://doi.org/10.3224/diskurs.v16i4.01

Ahmad Dhani telibat dalam kasus ujaran kebencian dan akhirnya divonis bersalah atas cuitannya di Twitter. Penelitian ini bertujuan untuk mengetahui bagaimana TvOne membingkai berita kasus ujaran kebencian yang dilakukan oleh Ahmad Dhani. Model framing yang digunakan dalam penelitian ini adalah model framing milik Robert N. Entman yang berfokus pada pemilihan isu dan penonjolan aspek dari suatu berita. Penelitian ini menggunakan metode kualitatif dengan menggunakan tiga berita TvOne pada periode November 2017 sampai Februari 2019 sebagai unit analisisnya, yang dianalisa dengan melihat teks, durasi dan scene . Berdasarkan element framing Entman yaitu Define Problem, Diagnose Cause, Moral Judgement, dan Treatment Recommendation , hasil penelitian menunjukkan tiga hal dalam pemberitaan tersebut, yaitu ketidakberimbangan narasumber, pengulangan narasi Ahmad Dhani tidak bersalah, dan kontroversi UU ITE di Indonesia. Ahmad Dhani was involved in the case of hate speech and was ultimately convicted of his post on Twitter. This study aims to find out how TvOne framed the case of hate speech committed by Ahmad Dhani. The framing model used in this study is Robert N. Entman's framing model, focusing on the selection of issues and highlighting aspects of a story. This study used a qualitative method using three TvOne clips from November 2017 until February 2019 as the unit of analysis, analyzed by looking at the text, duration and scene. Based on Entman's framing elements such as Define Problem, Diagnosis Cause, Moral Judgment, and Treatment Recommendation, the results of the study identified three frames in the news, including the imbalance of the sources, the narrative of Ahmad Dhani's innocence, and the controversy over the ITE Law in Indonesia.

This article proposes the study and analysis of the state-of-the-art in video game panorama, focusing on the tendency to use hate speech among young players. The immersion of the player in the symbolic arena of the game, where everything becomes possible, raises questions about the relationship between video games in the virtual world, and the player’s behavior in the physical world. It is shown that anonymity and the creation of communities and game groups can lead to the exclusion of and attacks to minorities; chat communications can facilitate sharing interests and game techniques as well as insults in times of tension between players, leading to imperative reflection on the role of gaming platforms in the control of shared content. Considering the boundless possibilities of video games, this article also reflects on game literacy and on how games have the potential to become powerful learning tools., Journal of Digital Media & Interaction, vol. 3 n.º 6 (2020): Journal of Digital Media & Interaction, Vol.3, No.6

At first, social media came as a solution for the community to facilitate communication and interaction virtually. However, due to the absence of strict regulation, it later changed into a means to spread the utterances of hate in the form of certain terms. The term cannot be understood by everyone. Therefore, researchers tried to dismantle the terms used, namely in terms of meaning and also the ideology of the use of the term. As for the term utterance of hatred which is used as material for analysis, among them are the short axis people, the people of the flat earth, the people of napkins, the children of the tablecloths, the children of the lewd, the children of camels / camel urine, the people of the micin, the children of cebong, the children of the camps, and the children of the negligee. The analysis was carried out using Fairclough's theory namely the dimensions of the text, discourse practice and social practice. Meanwhile, to dismantle its ideology is by using the theory of Jagger and F. Maier which consists of context, output of text, means of rhetoric, content and ideological statements, peculiarities and positions of discourse.

One of the consequences of the popularization of Internet access is the spread of insults and discriminatory messages, the so-called hatespeeches. They are comments that aim to discriminate against someone or a group of people because they belong to a certain group, usually minority, or have some characteristic common to other people. Fighting hates peech is a growing demand in real and virtual life as it profoundly affects the dignity of its victims. Detection of hatespeech is a difficult task because, in addition to natural language being inherently ambiguous, it requires a certain level of understanding of its linguistic structure. In many discourses, discrimination does not happen explicitly or with typical expressions: it is necessary world knowledge to recognize them. In addition, sometimes it is necessary to understand the context of the sentence to perceive its hateful content. Sarcasm is another huge challenge (even for humans) since its presence requires knowledge of the community and potentially of the user responsible for the comment for understanding their intent. Several approaches have been proposed for the hatespeech recognition task . Many authors consider the use of N-grams, of which those based on characters are more effective than those based on words. Combined or not with N-grams, lexical features were also evaluated, such as the presence or absence of negative words, classes or expressions indicative of insult, punctuation marks, letter repetitions, the presence of emoji, etc. Linguistic features were inefficient when used alone, such as POS tag, and the relationship between the terms of the dependency tree resulting from the syntax analysis. Recently, the most successful approach has used a neural network to create a distributed representation of the sentences present in a corpus of hatespeech, indicating that word embeddings training is a promising path in the area of hatespeech. Language drastically affects the tasks of Natural Language Processing (NLP), since most, if not all, words differ from one language to another, as well as their syntax, morphology, and linguistic construction. Thanks to this, works in English are not directly applicable in corpora of Portuguese language. In addition, corpora in Portuguese for hatespeech are rare, making researchers in the area to do all the construction work. In this dissertation we studied the use of deep cross-lingual Long Short-Term Memory (LSTM) model, trained with a hatespeech dataset translated from English in two different ways, preprocessed and vectorized with several strategies that were represented in 24 scenarios. The main approaches adopted included the training of embeddings through word index vectors (State of the Art technique), TFIDF vectors, N-grams vectors, with or without GloVe vocabulary, tested with the dataset constructed and labeled in this work and with another available in Portuguese. The inverted process was also tried out: we translated our corpus into English and compared the performance with its original version. With the embeddings resulting from the training process in each scenario, we used a Gradient Boosting Decision Tree (GBDT) as a means of improving classification. In fact, the results obtained with LSTM were improved in many scenarios. We achieved accuracy of up to 70 % in the experiments using the model written with the corpus in English and our dataset translated into this language. In others, traditional and successful techniques such as TFIDF vectors associated with an LSTM have not proved sufficient. Two important contributions of this work are: (i) proposal of an alternative research approach to attack the problem based on the translation of corpora and (ii) provision of a dataset of hatespeech in Portuguese to the community. Uma das consequências da popularização do acesso à Internet é a disseminação de insultos e mensagens discriminatórias, os chamados discursos de ódio (do inglês, hatespeech). São comentários que visam discriminar alguém ou um conjunto de pessoas por pertencerem a um certo grupo, normalmente minoritário, ou por possuírem alguma característica também comum a outras pessoas. O combate aos discursos de ódio é uma demanda crescente na vida real e virtual pois eles afetam profundamente a dignidade de suas vítimas. Detecção de discursos de ódio é uma tarefa difícil porque, além da linguagem natural ser inerentemente ambígua, ela exige certo nível de compreensão de sua estrutura linguística. Em muitos discursos, a discriminação não acontece de forma explícita ou com expressões típicas: é preciso ter conhecimento de mundo para reconhecê-las. Além disso, algumas vezes é necessário entender o contexto da frase para perceber seu teor odioso. O sarcasmo é outro desafio enorme (até para humanos) uma vez que sua presença exige conhecimento da comunidade e potencialmente do usuário responsável pelo comentário para o entendimento de sua intenção. Diversas abordagens foram propostas para reconhecimento do hatespeech. Muitos autores consideram N-Grams, dentre os quais aqueles baseados em caracteres mostram-se mais efetivos que aqueles baseados em palavras. Combinadas ou não aos N-Grams, features léxicas também foram estudadas, como a presença ou não de palavras negativas, classes ou expressões indicativas de insulto, sinais de pontuação, repetições de letras, presença de emojis etc. Features linguísticas mostraram-se ineficientes quando utilizadas isoladamente, como as POS tag, e a relação entre os termos da árvore de dependência resultante da análise sintática. Recentemente, a abordagem mais bem sucedida usou uma rede neural para criar uma representação distribuída das sentenças presentes em um corpus de discursos de ódio, indicando que o treinamento de word embeddings é um caminho promissor para a área. A língua afeta drasticamente as tarefas de Processamento de Linguagem Natural (PLN), uma vez que a maioria das palavras, se não todas, são diferentes de uma língua para outra, além de sua sintaxe, morfologia e construções linguísticas. Por esta razão, os trabalhos em língua inglesa não são diretamente aplicáveis em corpora de língua portuguesa, por exemplo. Além disso, corpora em português para discursos de ódio são raros, fazendo com que pesquisadores da área precisem realizar todo o trabalho de construção. Nessa dissertação, foi estudado o uso de um modelo deep cross-lingual Long Short-Term Memory (LSTM), treinado com um dataset de discursos de ódio traduzido do Inglês de duas diferentes maneiras, pré-processado e vetorizado com variadas estratégias que foram representadas em 24 cenários. As principais abordagens adotadas consideraram: o treinamento de embeddings através de vetores de índices de palavras (técnica Estado da Arte), vetores TFIDF, vetores N-Grams, com ou sem vocabulário GloVe, testados com o dataset construído e rotulado neste trabalho e com outro disponível em português. O processo invertido também foi experimentado: traduzimos o nosso corpus para o inglês e comparamos o desempenho com sua versão original. Com os embeddings resultantes do processo de treinamento em cada cenário, usamos uma Gradient Boosting Decision Tree (GBDT) como forma de melhorar a classificação e, de fato, os resultados obtidos com a LSTM foram melhorados em muitos cenários. Alcançamos precisão de até 70% nos experimentos usando o modelo treinado com o corpus em Inglês e nosso dataset traduzido para esta língua. Em outros, técnicas tradicionais e bem sucedidas como vetores TFIDF associados à uma LSTM não se mostraram suficientes. Duas importantes contribuições deste trabalho são: (i) proposta de uma abordagem de pesquisa alternativa de ataque ao problema baseada na tradução de corpora e a (ii) disponibilização de um dataset de discursos de ódio em língua portuguesa para a comunidade. São Cristóvão, SE

Робота спрямована на виявлення мови ворожнечі в пресі та встановлення причин і наслідків її використання в друкованих ЗМІ на прикладі всеукраїнської газети «День» та обласного видання «Сумщина». З’ясовано сутність поняття «мова ворожнечі» та визначено її класифікацію; встановлено тенденції поширення мови ворожнечі та причини її використання в друкованих медіа; здійснено аналіз видань «День» та «Сумщина» з погляду наявності в них мови ворожнечі; досліджено зв'язок між використанням мови ворожнечі та суспільно-політичними настроями в суспільстві, що впливають на мову газет.

Nous présenterons une analyse de messages à la frontière entre opinion et discours de haine, basée sur un corpus de tweets de personnalités politiques belges francophones. Les tweets ont été collectés au cours de la campagne électorale de mai 2019 et durant les mois qui ont précédé cette campagne. Nous sélectionnons quatre personnalités importantes par parti. Nous montrerons que les messages du corpus analysé contiennent très peu de stratégies incitant ouvertement à la haine mais que certains messages se situent à la frontière entre opinion et discours de haine. Ceux-ci peuvent inclure des représentations polarisantes ou stéréotypées de certaines communautés, basées sur des critères liés à l’ethnicité, la religion ou l’orientation sexuelle. Nous analyserons les stratégies linguistiques utilisées pour construire ces représentations polarisantes, comme la déixis, la généralisation, le langage métaphorique, et la construction de l’agentivité et de l’intentionnalité. En outre, nous montrerons comment la combinaison de ces stratégies peut contribuer à la création d’un discours qui représente certaines communautés comme une menace. En dernier lieu, nous comparerons les résultats de cette analyse avec les productions des mêmes personnalités politiques sur Facebook.

This paper presents the German Twitter section of a large (2 billion word) bilingual Social Media corpus for Hate Speech research, discussing the compilation, pseudonymization and grammatical annotation of the corpus, as well as special linguistic features and peculiarities encountered in the data. Among other things, compounding, accidental and intentional orthographic variation, gendering and the use of emoticons/emojis are addressed in a genre-specific fashion. We present the different layers of linguistic annotation (morphosyntactic, dependencies and semantic types) and explain how a general parser (GerGram) can be made to work on Social Media data, pointing out necessary adaptations and extensions. In an evaluation run on a random cross-section of tweets, the modified parser achieved F-scores of 97% for morphology (fine-grained POS) and 92% for syntax (labeled attachment score). Predictably, performance was twice as good in tweets with standard orthography than in tweets with spelling/casing irregularities or lack of sentence separation, the effect being more marked for morphology than for syntax.

In this study, we will present an analysis of discourses on the boundary between opinion expression and hatespeech, based on a corpus of online discourse of Belgian French-speaking politicians. Our corpus consists of the Facebook and Twitter productions from leading politicians from all major parties, which were collected one month right before and one month during the 2019 electoral campaign for the regional, national and European elections on May 26th 2019. While discourses falling within the legal qualification of hatespeech are extremely rare in this corpus, we will analyze the productions that are on the boundary between opinion expression and hatespeech, engaging in polarizing and stereotyping representations of certain communities based on criteria related to ethnicity, religion and sexual orientation. We look into the linguistic strategies that are used to construe these polarizing representations, such as the use of deictics, generalization, metaphorical language use, and the construction of agentivity and intentionality (i.e. the representation of certain actions as intentional deeds, thus construing e.g. an immigrant group as willingly harming the local population). We both analyze the use of these individual strategies in our corpus and show how they are combined in creating a discourse that represents these communities as a threat. Finally, we will also discuss to which extent there are differences between discourse produced before and during the electoral campaign. In doing so, we show how a detailed discursive analysis based on a corpus of contemporary online data can contribute to a better understanding of the general public debate and of people's conceptualisations concerning these communities.