Online Speech Research Articles

Drivers of hate speech in political conversations on Twitter: the case of the 2022 Italian general election

We study the prevalence and characteristics of toxic speech on Twitter in the run-up to the 2022 Italian General election. We analyzed over 8.5 M tweets shared by 450k unique users and employed a machine learning classifier to estimate the toxicity score of their messages. We found that supporters of different political coalitions exhibit varying levels of toxic speech, sending between 6-8% of toxic messages overall. Notably, Centre-Left politicians received more toxic messages on average, with the largest target of hate receiving over 15,000 toxic replies. We employed Generalized Linear Models to study factors that drive hate speech to political targets, finding that, importantly, politicians employing more abusive and harmful language are also more likely to attract more inflammatory speech. Our findings underscore the critical need for targeted interventions to address hate speech online, fostering healthier dialogue and safeguarding democratic discourse.

The effect of different visual feedback interfaces of music training games on speech rehabilitation in hearing-impaired children: An fNIRS study

Singing plays a critical role in enhancing musicality, sound discrimination, and attention, and proves advantageous for speech rehabilitation in children with hearing impairments. Computer-based training games are well-suited to the learning behaviors of children, with substantial evidence suggesting that music training augments speech training capabilities in this demographic. Despite this, there is a lack of detailed exploration into the design of interactive online music training interfaces tailored for these needs. This study investigates brain activation changes using two visual feedback singing games, analyzed through functional near-infrared spectroscopy: a serious game (SG) and an entertainment game (EG) with visually enhanced feedback. It also assesses the efficacy of home-based music training software for speech rehabilitation.Methods involved recording oxygenated hemoglobin concentration (Delta [HbO]) signals from the prefrontal cortex, motor cortex, occipital lobe, and temporal lobe in 21 children (average age: 9.3 ± 1.9 years) during two singing interface experiments. Subjects also completed the Intrinsic Motivation Inventory (IMI) questionnaire post-experiment.Main results showed that brain regions, particularly the temporal lobe, exhibited stronger and more pronounced activation signals with the SG interface compared to the EG, suggesting that SG is more effective for speech system rehabilitation. The Intrinsic Motivation Scale results revealed higher acceptability for SG than for EG. This study provides insights into designing online speech rehabilitation products for children with hearing impairment, advocating for better interactive training methods from a neuroscience perspective.

Reduced on-line speech gesture integration during multimodal language processing in adults with moderate-severe traumatic brain injury: Evidence from eye-tracking

BackgroundLanguage is multimodal and situated in rich visual contexts. Language is also incremental, unfolding moment-to-moment in real time, yet few studies have examined how spoken language interacts with gesture and visual context during multimodal language processing. Gesture is a rich communication cue that is integrally related to speech and often depicts concrete referents from the visual world. Using eye-tracking in an adapted visual world paradigm, we examined how participants with and without moderate-severe traumatic brain injury (TBI) use gesture to resolve temporary referential ambiguity. MethodsParticipants viewed a screen with four objects and one video. The speaker in the video produced sentences (e.g., “The girl will eat the very good sandwich”), paired with either a meaningful gesture (e.g., sandwich-holding gesture) or a meaningless grooming movement (e.g., arm scratch) at the verb “will eat.” We measured participants’ gaze to the target object (e.g., sandwich), a semantic competitor (e.g., apple), and two unrelated distractors (e.g., piano, guitar) during the critical window between movement onset in the gesture modality and onset of the spoken referent in speech. ResultsBoth participants with and without TBI were more likely to fixate the target when the speaker produced a gesture compared to a grooming movement; however, relative to non-injured participants, the effect was significantly attenuated in the TBI group. DiscussionWe demonstrated evidence of reduced speech-gesture integration in participants with TBI relative to non-injured peers. This study advances our understanding of the communicative abilities of adults with TBI and could lead to a more mechanistic account of the communication difficulties adults with TBI experience in rich communication contexts that require the processing and integration of multiple co-occurring cues. This work has the potential to increase the ecological validity of language assessment and provide insights into the cognitive and neural mechanisms that support multimodal language processing.

Does pre-speech auditory modulation reflect processes related to feedback monitoring or speech movement planning?

Previous studies have revealed that auditory processing is modulated during the planning phase immediately prior to speech onset. To date, the functional relevance of this pre-speech auditory modulation (PSAM) remains unknown. Here, we investigated whether PSAM reflects neuronal processes that are associated with preparing auditory cortex for optimized feedback monitoring as reflected in online speech corrections. Combining electroencephalographic PSAM data from a previous data set with new acoustic measures of the same participants’ speech, we asked whether individual speakers’ extent of PSAM is correlated with the implementation of within-vowel articulatory adjustments during /b/-vowel-/d/ word productions. Online articulatory adjustments were quantified as the extent of change in inter-trial formant variability from vowel onset to vowel midpoint (a phenomenon known as centering). This approach allowed us to also consider inter-trial variability in formant production, and its possible relation to PSAM, at vowel onset and midpoint separately. Results showed that inter-trial formant variability was significantly smaller at vowel midpoint than at vowel onset. PSAM was not significantly correlated with this amount of change in variability as an index of within-vowel adjustments. Surprisingly, PSAM was negatively correlated with inter-trial formant variability not only in the middle but also at the very onset of the vowels. Thus, speakers with more PSAM produced formants that were already less variable at vowel onset. Findings suggest that PSAM may reflect processes that influence speech acoustics as early as vowel onset and, thus, that are directly involved in motor command preparation (feedforward control) rather than output monitoring (feedback control).

Decentralized social networks and the future of free speech online

Decentralized social networks like Mastodon and BlueSky are trending topics that have drawn much attention and discussion in recent years. By devolving powers from the central node to the end users, decentralized social networks aim to cure existing pathologies on the centralized platforms and have been viewed by many as the future of the Internet. This article critically and systematically assesses the decentralization project's prospect for communications online. It uses normative theories of free speech to examine whether and how the decentralization design could facilitate users’ freedom of expression online. The analysis shows that both promises and pitfalls exist, highlighting the importance of value-based design in this area. Two most salient issues for the design of the decentralized networks are: how to balance the decentralization ideal with constant needs of centralization on the network, and how to empower users to make them truly capable of exercising their control. The article then uses some design examples, such as the shared blocklist and the opt-in search function, to illustrate the value considerations underlying the design choices. Some tentative proposals for law and policy interventions are offered to better facilitate the design of the new network. Rather than providing clear answers, the article seeks to map the value implications of the design choices, highlight the stakes, and point directions for future research.

Researching hate speech online: Exploring the potential and limitations of Facebook as a survey tool in Africa

The reasons driving users to flag/report online hate and social media mis/disinformation in Africa—or, conversely, why they choose not to. We conducted surveys with social media users in several African countries to gauge their definitional understandings of online hate and mis/disinformation; their awareness of content moderation processes, including the role of artificial intelligence (AI); and their engagement with flagging/reporting online content. The challenge, however, was how to conduct survey research in difficult-to-reach places, some of which are affected by violent conflict, and within a limited budget. Given that our aim was to reach social media users, we used Facebook to advertise our survey, drawing on the breadth of research on using the platform as a research tool and the fact it has a large user base in the targeted countries. In this article, we explore the practicalities, challenges, and limitations of Facebook as a tool for survey research in African countries.

Navigating hate speech and content moderation under the DSA: insights from ECtHR case law

ABSTRACT The dissemination of hate speech online necessitates forceful content moderation to protect individuals and democratic values, without undue infringement of freedom of expression and the right to access information. In Europe, recent regulatory measures like the Digital Services Act (DSA) address the amplification of harmful content on social media, and place responsibilities on Very Large Online Platforms to counter societal risks, such as hate speech. The DSA mandates that platforms balance commercial interests with protecting user rights and safety, necessitating nuanced moderation strategies. However, both automated and human moderation face challenges in accurately identifying and countering hate speech. The European Convention on Human Rights (ECHR) and its interpretation by the European Court of Human Rights (ECtHR) provide essential guidance for navigating how to counter hate speech and also protect freedom of expression. This article aims to contribute to the understanding the emerging landscape of platform regulation and how the demands in the DSA can be better understood and operationalized regarding moderation of hate speech, in light of ECtHR case law.

Ambiguous Threats

In a recent case, a Facebook user in Iran posted “death to Khamenei”, which the platform removed as a violation of its policy against threats and incitement. Facebook ultimately overturned the decision on the grounds that the speech, while contravening its rules, was newsworthy. Yet the company’s Oversight Board offered a distinct rationale for allowing the post: “death to Khamenei” wasn’t a threat or an incitement at all, but rather a rhetorical expression of criticism, disdain, or disgust. Who was right? Drawing on longstanding distinctions in the philosophy of language, we argue that the answer depends on whether platform rules should primarily target an utterance’s illocutionary force or instead its perlocutionary effects. Using this case study to trace broader lessons for the proper governance of online speech, we argue that platforms’ rules should primarily target illocutionary force, only considering perlocutionary effects in cases where illocutionary force is ambiguous.

Hate Speech and Human Rights: A Social Justice Perspective in Pakistan

This paper explores hate speech issues through the lens of human rights and social justice, focusing particularly on its impact on minority communities in Pakistan. It examines how hate speech undermines their human rights and leads to a denial of social justice, eroding the social fabric and perpetuating discrimination. The paper begins with a discussion of the current global context and then delves into the socio-political landscape of Pakistan, presenting relevant background information. It highlights that Pakistan, marked by religious, cultural, ethnic, and linguistic diversity, has struggled to fully embrace this diversity. Within this context, the article specifically studies the issue of online hate speech directed at minority communities and its detrimental effects on their well-being and societal integration. The paper concludes with comprehensive recommendations to tackle hate speech both online and offline, aiming to promote a more inclusive and just society for all, where every individual’s rights are respected and protected.

Examining the Use of Address Terms among Postgraduate Students in an Online Speech Community and Their Implication

This paper examined the terms of address as an important part of verbal behavior through which the behavior, norms and practices of a society can be identified. As a classic study in sociolinguistics, Brown and Gilman (1960) pronominal address system highlighted the semantic power and solidarity in relation to address terms. Since then, several studies have emerged on address terms in various social domains notably academia, religion, politics and many more. However, most of these studies did not give much consideration to online speech communities. This study therefore focuses on sociolinguistic analysis of address terms among postgraduate students in an online speech community using ethnographic approach. The study selected three (3) postgraduate online groups from the University of Cape Coast. One hundred and fifty chats (150) were collected from each group and studied to reveal the key address terms used by the students to address each and also the factors influencing the choice of the address terms. The study revealed that the key address terms used among postgraduate students are personal names, descriptive phrases and titles. Again, the study revealed the following as the factors influencing the choice of address terms: situational and socio-cultural factors, level of formality, innovativeness and playfulness, gender and age variations.

Exploring the Mechanism of Adult Users' Cyber-Aggression Against Adolescents: The Roles of Online Communication, Age Group Identity, and Online Moral Disengagement.

The increasing prevalence of adults' prejudice against adolescents on the Internet could cause opposing attitudes and even online aggression against teenagers. Yet, such age discrimination is less challenged compared to other social biases in cyberspace. Employing a social identity approach, this study aims to explore how features of online communication, teen-related personal experiences, and the dual identity of adult aggressors influence online moral disengagement (OMD), motives, and behaviors for cyber-aggression. We conducted an online survey of 767 Chinese adults in May 2023. The results demonstrated that perceived controllability of online speech, negative teen-related online experiences, and generation-based bias were positively associated with OMD. Next, adult respondents' self-serving motives positively predicted moderate and severe cyber-aggression, while educational motives only increased moderate cyber-aggression. Mediation analysis revealed that generation-based bias was the sole significant factor that amplified OMD, self-serving motives, educational motives, and cyber-aggression behaviors. In contrast, the effects of lifestage-based bias were insignificant. Our findings yield insights into individuals' moral transgressions in digital environments and shed light on the dynamics of the identity of adult aggressors in age-based discrimination. This study suggests that empathy from adults and age-appropriate prevention by online platforms are crucial to address online aggression against young generations.

A mobile application and system architecture for online speech training in Portuguese: design, development, and evaluation of SofiaFala

A mobile application and system architecture for online speech training in Portuguese: design, development, and evaluation of SofiaFala

Does pre-speech auditory modulation reflect processes related to feedback monitoring or speech movement planning?

Previous studies have revealed that auditory processing is modulated during the planning phase immediately prior to speech onset. To date, the functional relevance of this pre-speech auditory modulation (PSAM) remains unknown. Here, we investigated whether PSAM reflects neuronal processes that are associated with preparing auditory cortex for optimized feedback monitoring as reflected in online speech corrections. Combining electroencephalographic PSAM data from a previous data set with new acoustic measures of the same participants' speech, we asked whether individual speakers' extent of PSAM is correlated with the implementation of within-vowel articulatory adjustments during /b/-vowel-/d/ word productions. Online articulatory adjustments were quantified as the extent of change in inter-trial formant variability from vowel onset to vowel midpoint (a phenomenon known as centering). This approach allowed us to also consider inter-trial variability in formant production and its possible relation to PSAM at vowel onset and midpoint separately. Results showed that inter-trial formant variability was significantly smaller at vowel midpoint than at vowel onset. PSAM was not significantly correlated with this amount of change in variability as an index of within-vowel adjustments. Surprisingly, PSAM was negatively correlated with inter-trial formant variability not only in the middle but also at the very onset of the vowels. Thus, speakers with more PSAM produced formants that were already less variable at vowel onset. Findings suggest that PSAM may reflect processes that influence speech acoustics as early as vowel onset and, thus, that are directly involved in motor command preparation (feedforward control) rather than output monitoring (feedback control).

Neural semantic effects of tone accents.

This study investigated whether the brain utilizes morphologically induced tones for semantic processing during online speech perception. An auditory comprehension task was conducted while measuring event-related potentials (ERPs). The study tested whether a discrepancy between contextual expectations and the tonal realizations of the target word would yield an N400 effect, indicative of semantic processing difficulty. An N400 effect was observed, reflecting integration difficulty due to semantic anomalies caused by incongruent tones. Additionally, the ERPs in the congruent conditions were modulated by the cohort entropy of the target word indicating lexical competition. The late negativity observed in this study encompasses both the N400 and preactivation negativity. This overlap underscores the brain's potential for rapidly connecting form and meaning from different sources within the word, relying on statistically based prediction in semantic processing.

The Bypass Strategy: platforms, the Online Safety Act and future of online speech

ABSTRACT In this paper, we argue that the Online Safety Act 2023 and Ofcom’s guidance incentivise online platforms to adopt a ‘Bypass Strategy’, where they create and enforce content moderation rules that are broader than existing criminal law to bypass judgements of illegal content. This strategy aims to avoid complex legal interpretations of criminal intent and potential defences but would be unfeasible considering the volume of content on social media platforms and incompatible with automated moderation tools. We argue, however, that the Bypass Strategy, driven by the Act’s focus on illegal content and by the lack of clarity in Ofcom’s proposed guidance, poses a significant threat to users’ freedom of expression and incentivises overremoval of legitimate speech. We offer insights that could help Ofcom to improve its guidance on how platforms should interpret such duties on moderating content and might mitigate this risk within the constraints of the Act.

“View and Hide Definitions” of Racist Hate Speech: Ethnophaulisms in Google’s English Dictionary

This paper aims to foster debate about the language of racist hate speech in online English lexicography. For this purpose, it presents a study on the treatment of ethnophaulisms, or ethnic slurs, in “powered by Oxford Languages” Google’s English dictionary. The focus is indeed on the perspective of the general user of the Internet, in light of the connection between two facets of this digital age. The first one is the strong and growing tendency among Internet users to ‘google’ their language issues. The second one is the alarming increase in cases of hate speech online, most of which are based on ethnicity and nationality, according to reports by the United Nations. Consequently, the free and pervasive content of Google’s English dictionary represents a case in point to investigate whether and how online users are warned against the power of these hate words. A selected sample of 285 English ethnic slurs have been looked up in the dictionary and, if recorded, their entries have been scrutinised to identify lexicographic data regarding their semantic relevance and offensiveness. Findings show that the majority are included, they mostly present ethnicity-related senses, but less than half of the total are treated as ethnophaulisms. In this respect, the major dictionary markers indicating offensiveness are effect labels, predominantly alone or combined with definitions. Relative to their size, thus, ethnophaulisms in Google’s English dictionary are clearly described as offensive or derogatory expressions, thus making online users aware of their hurtful nature.

On the Influence and Political Leaning of Overlap between Propaganda Communities

Social media offers increasingly diverse mechanisms for the distribution of motivated information, with multiple propaganda communities exhibiting overlaps with respect to user, content, and network characteristics. This has particularly been an issue in the Global South, where recent work has shown various forms of strife related to polarizing speech online. It has also emerged that propagandist information, including fringe positions on issues, can find its way into the mainstream when sufficiently reinforced in tone and frequency, some of which often requires sophisticated organizing and information manipulation. In this study, we analyze the overlap between three events with varying degrees of propagandist messaging by analyzing the content and network characteristics of users leading to overlap between their users and discourse. We find that a significant fraction of users leading to overlap between the three event communities are influential in information spread across the three event networks, and political leaning is one of the factors that helps explain what brings the communities together. Our work sheds light on the importance of network characteristics of users, which can prove to be instrumental in establishing the role of political leaning on overlap between multiple propaganda communities.

Online speech synthesis using a chronically implanted brain–computer interface in an individual with ALS

Brain–computer interfaces (BCIs) that reconstruct and synthesize speech using brain activity recorded with intracranial electrodes may pave the way toward novel communication interfaces for people who have lost their ability to speak, or who are at high risk of losing this ability, due to neurological disorders. Here, we report online synthesis of intelligible words using a chronically implanted brain-computer interface (BCI) in a man with impaired articulation due to ALS, participating in a clinical trial (ClinicalTrials.gov, NCT03567213) exploring different strategies for BCI communication. The 3-stage approach reported here relies on recurrent neural networks to identify, decode and synthesize speech from electrocorticographic (ECoG) signals acquired across motor, premotor and somatosensory cortices. We demonstrate a reliable BCI that synthesizes commands freely chosen and spoken by the participant from a vocabulary of 6 keywords previously used for decoding commands to control a communication board. Evaluation of the intelligibility of the synthesized speech indicates that 80% of the words can be correctly recognized by human listeners. Our results show that a speech-impaired individual with ALS can use a chronically implanted BCI to reliably produce synthesized words while preserving the participant’s voice profile, and provide further evidence for the stability of ECoG for speech-based BCIs.

Borderline content and platformised speech governance: Mapping TikTok's moderation controversies in South and Southeast Asia

AbstractContent moderation comes with trade‐offs and moral dilemmas, particularly for transnational platforms governing borderline content where the boundaries of acceptability are subject to debate. While extensive research has explored the legality and legitimacy of platformised speech governance in established democracies, few address the complexities of less‐than‐democratic developing nations. Through sociolegal analysis and controversy mapping of TikTok's localised moderation in South and Southeast Asia, the study examines how major actors negotiate the shifting boundaries of online speech. The analysis reveals that neither the platform nor regional states are well‐positioned to achieve sound governance of borderline content. Primarily, TikTok localises its moderation based on pragmatic necessity rather than moral obligations, deliberately sidestepping contentious political controversies. Governments demonstrate determined efforts to control online discourse, leveraging legal uncertainty to further political agendas. Thus, local content governance often relies on vague rationales around securitisation and morality. The contradictory motives of (de)politicising borderline moderation seemingly counterbalance each other, yet in practice lead to an accountability vacuum where legitimate interests are sidelined. Given the lack of normative common ground, ensuring procedural justice and encouraging civic participation are essential to counteract rhetoric that rationalises imposition of certain speech norms hinging on imbalanced political power.

Who cares? How personal political characteristics are related to online counteractions against hate speech

Abstract Although many Internet users have encountered hate speech online, interventions against those incidents are unlikely. To learn more about online counteractions against hate speech, which we conceptualize as a form of online civic participation, we conducted an online survey representing adult German Internet users (N = 2,691). We investigated what distinguishes users who intervene against hate speech, focusing on associations between citizens’ prior counteractions and their personal political characteristics, their attributed responsibility for intervening against hate speech to social and political actors, and their institutional trust. Results indicate that especially Internet users’ personal political characteristics (e.g., internal political efficacy, political opinion leadership) as well as trust in regulative political institutions are positively associated with counteracting against online hate speech. While social media use and previous experience with counteractions increased the likelihood of users counteracting against hate speech, authoritarian attitudes and attributing responsibility for intervening to platform operators decreased it.