The traditional use of the generic masculine in the Italian legal language has been criticised for being sexist. Therefore, the paper examines the possibilities and limitations of drafting inclusive definitions of legal agentives in Italian. In particular, it outlines the challenges encountered when rewriting definitions of legal agentives within a terminology database with the help of four undergraduates. The challenges include the need to balance inclusion of all genders with preserving the correct legal meaning, while maintaining comprehensibility and clarity. The abstract nature of legal agentives – which may refer to both natural persons and legal entities – raises questions about some feminine forms. Legally defined agentives may prove difficult to modify to achieve inclusive forms due to the risk of losing intertextual references that ensure legal certainty and of hindering a consistent interpretation of legal concepts. Inclusive definitions of legal agentives in Italian are feasible, but going beyond binary inclusion may be complicated. In-depth legal knowledge is required to propose inclusive and legally accurate alternatives to definitions in the generic masculine.
With the rise of artificial intelligence, the use of machine translation (MT) has become commonplace. However, concern has been voiced with regard to MT output, from the perspective of both the quality and a range of biases that are evident in texts translated using this technology. In this study, we focus on the phenomenon of gender bias, specifically in the case of translation from languages that have no explicit grammatical gender, such as Basque, to languages that do, such as Spanish. In the study, we collected samples from three corpora created from texts drawn from different fields (literature, science, and journalism) and examined the translations proposed by an MT system (Elia) with regard to use of stereotypical masculine and feminine adjectives and occupations. Our findings suggest that when no explicit gender is given in Basque, the MT system primarily selects the masculine option in Spanish. Nevertheless, in certain occupations, we observed that the use of certain translation methods can contribute to producing less stereotypical target texts.
Previous research into the use of gender-fair noun references in German-speaking newspapers has found an overall increase in the use of gender-fair language (GFL) over the last decade. This trend also extends to the non-binary asterisk and colon. The increased integration of these controversial forms of GFL may suggest that newspaper authors and editors have been placing more importance on gender-related issues. It is therefore unsurprising that, particularly in recent years, more editorial offices have published official guidelines or statements regarding their use of GFL. While some previous studies on the use of GFL in German-speaking media have touched on language guidelines and official statements, a systematic overview that contextualises the use of GFL and published guidelines is lacking. Thus, the aim of this paper is to address this research gap by examining recent official language guidelines and statements from selected German-speaking newspapers and contextualising them with actual GFL usage frequency. To achieve this, newspapers both supporting and opposing the use of special characters were chosen, and the frequency of selected forms of GFL was analysed through a corpus-based study and subsequently compared. The results indicate that newspapers with official guidelines supporting non-binary forms tend to use these forms more frequently. However, the use of such forms is in decline in the selected newspapers, while it is still increasing – albeit at lower frequency levels – in newspapers whose guidelines do not support non-binary forms. Hence, the findings suggest that the relationship between guidelines and GFL usage in the media is far from being straightforward.
The objective of this paper is to present the research carried out as part of the HCAI4GEND project, which focuses on the development of a linguistic model aimed at creating an artificial intelligence-based tool to detect, interpret and generate linguistic solutions related to sensitive language in institutional texts in Italian, English and German. The designed tool ensures effective user engagement through a feature-rich text editor, real-time highlighting and interactive suggestion features. The continuous learning mechanism built into the artificial intelligence back-end enables the system to adapt to evolving language norms and user preferences, ensuring long-term effectiveness and cultural relevance. To achieve these goals, the use of high-quality language resources is critical. These resources are developed following hybrid approaches that integrate contrastive typological studies with varietistic analyses, ensuring a nuanced understanding of grammatical and social usage of gender.
This section provides an overview of the collection of selected papers from the LaGendA Conference, held at Ca’ Foscari University of Venice on October 3–4, 2024. The papers cover four areas of research i) guidelines, communication strategies, and adaptations; ii) cultural stereotypes and grammatical gender in specific languages; iii) interpreting and translating into languages with grammatical gender; iv) gender issues in Artificial Intelligence, Large Language Models, and Chatbots. The volume provides tools for the interaction of linguists, translators and interpreters, social scientists, computationalists, and content makers.
Gender bias is deeply embedded in language and influences perceptions of social roles. Prior research suggests that the Foreign Language Effect (FLE) – the phenomenon where using a foreign language affects judgment and reasoning – can reduce implicit biases, including gender stereotypes, when individuals process information in a second language (L2). This effect has been attributed to increased cognitive effort and emotional detachment, leading to more deliberate and less automatic judgments. However, existing studies primarily examine FLE in linguistically distant languages, leaving open the question of whether the effect persists when both L1 and L2 share grammatical gender systems. This study investigates gender bias in adjective evaluations among native Italian speakers and Italian speakers assessing adjectives in Spanish as an L2. Participants rated adjectives categorized under Power, Weakness, Warmth, and Coldness, which have established gender connotations. The results replicated prior findings, showing that Power and Coldness adjectives are predominantly associated with masculinity, while Weakness and Warmth adjectives are linked to femininity. Additionally, female participants tended to assign stronger gender-stereotyped scores than male participants – a pattern consistent with previous research. However, no significant differences were found between evaluations in Italian (L1) and Spanish (L2), suggesting that linguistic similarity might have weakened or neutralized the FLE.
Achieving linguistic inclusivity in artificial intelligence systems cannot be accomplished by technological advancements alone. This paper underscores the necessity of a multidisciplinary and collaborative approach involving linguists, sociologists, software developers, and policymakers to ensure that AI-driven language technologies are not only technically robust but also culturally sensitive and ethically sound. Equally crucial is the active engagement of end-users in the design and implementation of inclusive AI systems, fostering a sense of empowerment rather than imposition. Promoting inclusive language through AI is therefore not merely a technical endeavor, but a shared social responsibility. Language that reflects and respects human diversity enhances communication and contributes to a more just and equitable society. The future of AI in the domain of language will depend on our collective capacity to design systems that are transparent, accountable, and responsive to the evolving needs of a pluralistic global community.
This article offers a contrastive examination of gender-inclusive language guidelines issued by universities in Germany, Austria, Switzerland, Italy, and the United Kingdom. Drawing on a corpus of 287 institutional documents – 122 in German, 34 in Italian, and 133 in English – the study identifies and quantifies specification and neutralization strategies recommended for administrative and pedagogical discourse. After manual annotation, strategies were coded into a database, enabling statistical comparison across languages and nations. Results show that German-speaking institutions favor gender-marked neographies and pair forms, whereas Italian guidelines privilege binary splitting and feminine derivation, with neographies largely proscribed. English documents overwhelmingly promote lexical and pronominal neutralization, including singular they and neopronouns, and eschew specification entirely. Legal framing – especially the recognition of a non-binary civil status in Germany – emerges as a key predictor of neography uptake. Despite divergent prescriptions, all guidelines condemn the generic masculine and seek inclusive representational equity. The paper concludes that cross-linguistic variation in inclusive strategies reflects structural properties of the languages, local legal contexts, and institutional ideologies, and argues for heightened intercultural dialogue in developing future policies. These findings contribute to sociolinguistic theory by evidencing how macro-social variables interact with grammatical typology to shape emergent language planning norms and practices.
This research aims to examine the use of gender-fair language in job advertisements within the contemporary Italian job market. The study is based on a sample of 240 job announcements collected online in 2024, with results compared to those of previous research and analyzed using corpus linguistics tools. The findings will confirm the predominance of neutralizing strategies and masculine forms, while revealing a slight increase in the use of split forms and the emergence of non-binary language, such as neomorphemes. This trend underscores the persistence of linguistic asymmetries in job announcements: problematic gender-related linguistic strategies may contribute to maintaining access barriers to specific segments of the labor market for women and to the overall persistence of social and cultural stereotypes.
This study investigates the use of non-binary inclusive language in a recent German novel, as well as in a French translation of this novel. While inclusive language in the German novel mainly targets nominal forms using the gender star, as well as the indefinite pronoun, in the French translation, inclusive language forms are more widespread and also present through agreement on adjectives, determiners, and a wider range of pronouns.
Within studies on the complex relationship between language and gender, there are proposals that indicate an effect of stereotypical mental representations on language processing. Likewise, other proposals indicate that morphological gender marking, typical of languages with grammatical gender, can also bias interpretation. Within the MultiLingualGender project, we are developing a broad research on language and gender in Romance languages. In this paper we present two preliminary studies that also function as normative stage for future psycholinguistic tasks. On the one hand, we developed a study of Gender association judgments, with the objective of verifying the association of role names (professions) with gender stereotypes, but without explicitly involving the lexical form (Study 1). On the other hand, we conducted a study of Acceptability judgments of NP, to analyze the degree of acceptability of NPs that explicitly contained the lexical form (Study 2). In both cases, these are tasks that involve conscious and open judgments. Our results show that there are differences by linguistic community, but there are also some common patterns: for example, the acceptability and possibility of representing men in roles typically associated with women is greater than vice versa. Furthermore, data show that in languages with grammatical gender, the interaction between stereotypicality biases and grammatical gender marking plays a crucial role in understanding the relationship between gender and language in its complexity.
This article examines the development and characteristics of guidelines for non-sexist/inclusive language, a genre of texts that has evolved since the 1970s to promote gender equality in language use. These guidelines, which vary in format and scope, aim to address the representation of individuals in language and advocate for gender balance through practical recommendations. The article defines these guidelines and explores typical elements found in them. It also discusses the variance in content and format among guidelines, which can range from brief recommendations to comprehensive books and may address one or multiple languages. The concept of inclusion in such guidelines has expanded beyond gender to encompass various groups of people, aspects of communication, and broader contexts. The article traces the development of the term “inclusive” in guidelines and discusses how different guidelines define and approach inclusivity. It concludes by emphasizing the evolving role of these guidelines in adapting to changing social, political, and linguistic contexts, and their potential to address multiple dimensions of inclusion in the future.
As has been shown in various studies considering different languages, in professional contexts women tend to be referred to differently than men. While men are typically referred to by their surname (e. g., Fermi), women are more often referenced with their full name (e. g., Samantha Cristoforetti) or first name alone (e. g., Samantha). The present study proposes an empirical case study investigating whether this gender-indexing bias is also present in texts generated by large language models (LLMs). Based on the analysis of a self-assembled data collection comprising 420 biographies produced by GPT-4 on 140 eminent Italian and French female and male personalities, our study reveals that the synthetic texts investigated not only reflect the gender biases found in human-authored texts but, in some cases, even amplify them.
Integrating gender-fair language in spoken multilingual settings presents communication challenges such as cross-linguistic constraints. In simultaneous interpreting, the fast-paced and cognitively demanding nature of the task may limit compliance with institutional toolkits for gender fairness and inclusivity. Additionally, when translating into grammatical gender languages, such as Italian and French, more recent non-binary language innovations may encounter resistance in both attitudes and usage. This paper discusses gender-fair translation strategies in conference interpreting through an experimental study on two language pairs (EN>IT; EN>FR) and in-depth retrospective interviews with EU interpreters. After conceptualising gender-fair interpreting as a task involving three overlapping challenges, the study explores key factors –cognitive load, strategic decision-making, ethical dilemmas – that contribute to the gap between interpreters’ positive attitudes and a moderate use of gender-fair language strategies in simultaneous interpreting in EU institutional settings.
This study presents a sociolinguistic survey that investigates how the reclamation of LGBT+ slurs is perceived in Italian. Reclamation refers to the practice of repurposing historically derogatory terms into expressions of identity, pride, solidarity, activism. Despite significant advancements in AI in the automatic detection of the abusive use of derogatory terms, the reclamation of slurs remains largely overlooked in NLP research – especially in Italian. Failure to adequately consider the context may lead automated moderation systems to erroneously classify reclaimed slurs as abusive, with the risk of censoring the voices of activists and marginalized communities and limiting their freedom of expression. The survey is based on a web questionnaire designed with input from a preliminary focus group, and it collected both linguistic perceptions and sociodemographic data from 279 respondents. Results show that LGBT+ individuals and younger participants are more familiar with and accepting of reclaimed language. While most respondents agree that ingroup members have greater legitimacy in using reclaimed slurs, opinions vary regarding acceptable contexts and speakers. This underscores the importance of context, intention, and positionality in language interpretation, something that should be taken into account to develop NLP tools that are sensitive enough to deal with reclamatory uses of slurs in the future. Ultimately, this work lays the groundwork for ethical dataset creation and annotation practices that respect marginalized voices, since it supports a community-centered, socially-aware approach to future NLP tools.
Recent research has increasingly exposed the meaning-making potential of AI systems across a range of AI-coded platforms, particularly highlighting their tendency to reproduce deep-rooted gender- and sex-discrimination (cf. Bolukbasi et al. 2016; Noble 2018). While much of this work has focused on marginalised groups, the present study shifts attention to the normalised subjects of coupledom, i. e., cis-heteronormative couples, to examine how dominant ide-ologies are naturalised in the outputs of AI-based search engines. Adopting a Multimodal Critical Discourse Analysis approach, this study investigates how the visual and verbal modes combine in Google Images when querying the linguistically unmarked terms couples and coppie in the British and Italian digital landscapes respectively. Findings reveal that Google Images’ multimodal representation of cis-heteronormative couples aligns with socio-cultural norms entrenched in what is the Western basic default norm of conceiving coupledom. While specific patterns emerge in each digital landscape, both datasets reproduce neoliberal imaginaries of coupledom where idyllic relational success, self-optimisation and aestheticised desirability marginalise alternative couple forms and reinforce the monolithicity of cis-heteronormative coupledom. By comparing data from two distinct cultural contexts, this study demonstrates the relevance of MCDA for uncovering how AI-based search engines visually and verbally reinforce dominant norms. While data reflect a specific socio-cultural and temporal snapshot of each digital landscape, the findings illustrate how Google Images systematically reproduces normative couple models across both the Italian and British contexts.
This study examines Cardinal Rex Lawson’s (1969) “So Ala Temem” as a multidimensional communicative event in which language, music, and sociocultural meaning intersect. Highlife music provides a lens to analyze how melodic, rhythmic, and performative elements convey both referential and affective content. The study draws on a high-fidelity recording, researcher-generated transcription, and consultations with native Kalabari, Ibani, Izon, and Okrika speakers. An interdisciplinary approach combines ethnographic linguistics, musicological analysis, and sociocultural interpretation, guided by Highlife scholarship, musilanguage theory, and Hymes’ (1974) SPEAKING model. Findings show that “So Ala Temem” goes beyond entertainment, integrating multilingual elements and playful vocalisations (ludic or vocable-based sound) with instrumental and vocal techniques to create a musilinguistic synthesis. The interplay of language and music reinforces social cohesion, moral guidance, and communal identity. Melodic phrasing, rhythmic patterns, and expressive vocalizations encode cultural knowledge and affective meaning, while repetition, interjections, and chorused refrains structure the song and enhance audience engagement. Lawson’s performance exemplifies the communicative and cultural significance of Nigerian Highlife, demonstrating how musical and linguistic strategies convey ethical instruction, social norms, and collective memory.
Slogans play a central role in branding: they contribute to brand image and are primary touching points between brands and consumers. Slogan generation is not easy because slogans need not only to be in touch with the brand, but they also need to be memorable, catchy, and overall appealing. New technologies, however, such as Generative Pre-trained Transformers (GPT), by different AI platforms, may help in slogan generation. Although AI is researched in many contexts, more research is needed on specific marketing-connected tasks. The current study analyzes and evaluates AI-generated slogans: General and specific slogans connected to sustainable brands are created in English and Hungarian using different AI platforms. The linguistic characteristics of the generated slogans are analyzed, and university students rate Hungarian slogans in five dimensions. Results show that a given software performs differently depending on the criteria used for evaluation, language, the prompt used, and the time of slogan generation; thus, results are hardly generalizable. Due to the results and ethical considerations, AI platforms are recommended for use for brainstorming during slogan generation.
This explorative study discusses a hitherto underdescribed Source of motion expression type that is based on sequencing a Place-denoting component with a Goal-denoting component. Nine languages that show this bipartite coding strategy are discussed in detail, and are compared with regard to eventhood criteria, such as shared TAM marking. This type of Source expression shows that some languages make do without dedicated Source verbs and without grammatical Source marking. Instead, they are headed by Goal verbs which are iconically ordered with Place-denoting forms. Together, they express two subevents (staying at X, going to Y), the combination of which results in a Source of motion event (leaving [from] X). The bipartite and iconically ordered Source expression type is a crosslinguistically attested strategy which shows some formal variety across languages. The bipartite strategy is further contrasted with Goal/ Source expressions where Path is inferred from lexical semantic contrast.
Abstract This article presents the results from an intervention study on process-oriented writing in L3 German among 21 Swedish students in an upper secondary school. Our study investigates the relationship between process-oriented writing and text quality concerning accuracy, content, text length, and syntactic complexity. We also consider how individual variables, such as cognitive load (CL), learner engagement, and achievement goals, relate to the learning outcome in process-oriented writing. Our data consist of students’ mind maps, student texts, and questionnaires. Regarding accuracy, results show a slight improvement in basic morphology and syntax. As for text quality, there is a correlation between a well-prepared mind map and a text rich in content, whereas only a weak relationship between mind map and text length was found. In contrast to previous studies (cf. Pon/Varga 2017), our analysis shows only a weak relationship between text length and syntactic complexity. Concerning cognitive load and engagement, the students experienced problems when giving and receiving feedback. Even so, most students perceived the work with process-oriented writing as engaging. Overall, they express a clear desire to develop their writing and appreciate the opportunity for planning and time for reflection.
