Covertly communicated hate speech: A corpus-assisted pragmatic study

Parvaresh, Vahid

doi:10.1016/j.pragma.2022.12.009

Cited by 8 publications

(6 citation statements)

References 48 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In particular, those tasks with high correlation within the offensive category are highly correlated in predicting the politeness classification task. This finding is supported by literature showing that impoliteness can fall under the umbrella of offensive language (B ączkowska, 2021) and, although key differences exist in the pragmatics of the two, the constructs are closely related (Parvaresh, 2023;Culpeper, 2021).…”

Section: Do We See Cross-task Transfer Of Socialsupporting

confidence: 70%

Do LLMs Understand Social Knowledge? Evaluating the Sociability of Large Language Models with SocKET Benchmark

Choi,

Pei,

Kumar

et al. 2023

Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing

View full text Add to dashboard Cite

Large language models (LLMs) have been shown to perform well at a variety of syntactic, discourse, and reasoning tasks. While LLMs are increasingly deployed in many forms including conversational agents that interact with humans, we lack a grounded benchmark to measure how well LLMs understand social language. Here, we introduce a new theorydriven benchmark, SOCKET, that contains 58 NLP tasks testing social knowledge which we group into five categories: humor & sarcasm, offensiveness, sentiment & emotion, trustworthiness, and other social factors. In tests on the benchmark, we demonstrate that current models attain only moderate performance but reveal significant potential for task transfer among different types and categories of tasks, which were predicted from theory. Through zero-shot evaluations, we show that pretrained models already possess some innate but limited capabilities of social language understanding and training on one category of tasks can improve zero-shot testing on others. Our benchmark provides a systematic way to analyze model performance on an important dimension of language and points to clear room for improvement to build more socially-aware LLMs. The resources are released at https://github. com/minjechoi/SOCKET.

show abstract

Section: Do We See Cross-task Transfer Of Socialsupporting

confidence: 70%

Do LLMs Understand Social Knowledge? Evaluating the Sociability of Large Language Models with SocKET Benchmark

Choi,

Pei,

Kumar

et al. 2023

Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing

View full text Add to dashboard Cite

show abstract

“…Soudeh Ghaffari [7] endeavored to analyze the hate speech targeting on female Instagram celebrities to expose online misogyny and sexism via a case study of Lena Dunham, who portrayed herself. Vahid Parvaresh [8] focused on covertly communicated hate speech on Afghan immigrants in Iran. .…”

Section: Hate Speechmentioning

confidence: 99%

"I don’t discriminate but……": Legitimation Strategies of Covert Online Hate Speech against Homosexuality in China

2024

langl

View full text Add to dashboard Cite

Internet provides an enormous space for free speech, especially for hate speech. Recent years have witnessed the increasing discriminating speech against homosexuality, which sounds righteous with the strategies of talk in China because the major opinion in online space is homophobia, which is not benefit for the equal rights and living circumstances for homosexuality. Therefore, it's necessary to probe into how these covert hate speeches are legitimized and how they create a more profound homophobic atmosphere. In this study, comments, which are hate speech to homosexuality, under 5 videos on Bilibili are collected, classified with the assistance of corpus. Then, the results are analyzed under the legitimation strategies, proposed by Van Leeuvan. Finally, it showed that three strategies are used and the incitement of covert hate speech are more powerful than overt hate speech. The study could provide a discursive angle for the problems of civil rights and equal social status of gay community.

show abstract

“…Le même problème est posé par un mot comme « élite », parfois employé par une rhétorique populiste pour désigner l’existence d’un complot mondial ourdi par les populations juives. Symétriquement, ces approches se trouvent relativement peu probantes pour mettre en évidence les discours de haine dissimulés ou « covert hate speech » (Baider et Constantinou, 2020 ; Parvaresh, 2023) et les énoncés qui emploient les formes de contournement mentionnées ci-dessus : implicite, ironie, stéréotypes...…”

Section: éTudier Automatiquement Les Discours De Haineunclassified

“…Le même problème est posé par un mot comme « élite », parfois employé par une rhétorique populiste pour désigner l'existence d'un complot mondial ourdi par les populations juives. Symétriquement, ces approches se trouvent relativement peu probantes pour mettre en évidence les discours de haine dissimulés ou « covert hate speech » (Baider et Constantinou, 2020 ;Parvaresh, 2023) et les énoncés qui emploient les formes de contournement mentionnées ci-dessus : implicite, ironie, stéréotypes... Des travaux qualitatifs plus classiques sur les discours de haine faisaient déjà état de la difficulté qu'il y a à traiter les discours de haine, qu'il s'agisse du racisme, de la xénophobie, ou de l'antisémitisme, comme une catégorie unifiée. Marc Angenot, par exemple, dans La Parole pamphl étaire, devant la variabilité de l'objet qu'il étudie, se résout à adopter une méthode idéaltypique consistant à préciser les traits minimaux, les topoı ¨sous-jacents, qui identifient le discours pamphlétaire (Angenot, 1982).…”

Section: éTudier Automatiquement Les Discours De Haineunclassified

Mesurer l’empreinte antisémite sur YouTube

Tainturier,

de Dampierre,

Cardon

2023

Bulletin of Sociological Methodology/Bulletin de Méthodologie S

View full text Add to dashboard Cite

Cet article propose une approche exploratoire visant à mesurer l’empreinte du discours antisémite sur les chaînes d’information et d’actualité de YouTube en France. Devant la complexité et la polysémie du phénomène étudié, une approche par « empreinte » a en effet été privilégiée, s’appuyant sur des méthodes de machine learning et l’entrainement d’un outil de traitement automatique du langage. Il apparaît que cet outil, imparfait, mais ouvrant de nombreuses pistes méthodologiques quant à l’étude des discours de haine, détecte un antisémitisme peu prévalent, mais surreprésenté dans les espaces numériques liés à la droite radicale.

show abstract

Covertly communicated hate speech: A corpus-assisted pragmatic study

Cited by 8 publications

References 48 publications

Do LLMs Understand Social Knowledge? Evaluating the Sociability of Large Language Models with SocKET Benchmark

Do LLMs Understand Social Knowledge? Evaluating the Sociability of Large Language Models with SocKET Benchmark

"I don’t discriminate but……": Legitimation Strategies of Covert Online Hate Speech against Homosexuality in China

Mesurer l’empreinte antisémite sur YouTube

Contact Info

Product

Resources

About