勒芒大学

This study examines whether employee-shareholder board representatives are sufficiently salient to management to enhance the equal treatment of minority shareholders. Using data from French SBF 120 firms over the period 2002-2020, we show that this form of hybrid representation, combining labour and capital interests, is positively associated with stronger protection of minority shareholders. Importantly, the effect is substantially stronger in family-controlled firms, where concentrated ownership and socioemotional wealth considerations heighten the risk of minority shareholder expropriation and increase the relevance of hybrid employee-shareholder directors as monitors. By contrast, labour board representation shows no systematic relationship with minority shareholder protection, underscoring the distinct governance role played by employee-shareholder representatives. These findings suggest that employee-shareholder board representation functions as a responsible governance mechanism that both strengthens monitoring and signals a commitment to fair treatment of outside investors. More broadly, the results illustrate how emerging forms of worker involvement challenge traditional boundaries between labour and capital in corporate governance, highlighting the need to extend industrial relations theory to account for hybrid stakeholder roles, particularly in family firms.

Recent studies have highlighted that, beyond socio-ecological issues, environmental policies must also tackle environmental-related inequalities and injustice to be efficient. This paper examines this assumption from an empirical perspective, focusing on deforestation reduction policies in the Democratic Republic of the Congo (DRC). To this end, we mapped and analysed the United Nations deforestation and forest degradation reduction (REDD+) policy instrument in the DRC, along with the most relevant associated regulations, and policy documents issued or used in the DRC provinces of Tshopo, Kwilu, Equateur, Mongala, and Mai-Ndombe. The analysis of these materials was complemented by key-informant interviews from experts who participated in discussions and the design of REDD+ policies in the DRC at least over the past decade. The key findings of our empirical-based research indicate that the recognition principle of environmental justice received little attention during the preparation of REDD+ initiatives DRC. In a few instances, this principle was partially considered, but only bettedly, through the adoption of specific laws and ad hoc regulatory instruments, including Free, Prior and Informed Consent (FPIC) and National Social and Environmental Standards. Regarding procedural justice, the World Bank and UN-REDD effectively made the consideration of these aspects a formal condition for DRC participation in the UN REDD+ process. Finally, the principle of distributive justice remains unfair because the profit-sharing system fails to provide local entities with sufficient resources to conserve the forests under their stewardship.

This article focuses on the evolution of animal stories and fables in the literature of the Rio de la Plata between the colonial period and the 20th century. It does not examine the literary theme of transformation, or metamorphosis, as explored by Ovid and Kafka, because in the examples discussed, the animals already appear humanized and/or personified. The anthropomorphic ingredients in the protagonists of these stories take on different degrees of importance, but what is interesting is the mixed presence of elements of European or Latin American reality in the representation of these animal beings assimilated to humans. The motif is explored through two writers from the colonial era (Ruy Diaz de Guzman and Juan Cruz Varela) and two writers from the 20th century (Juana de Ibarbourou and Maria Elena Walsh). The hybridity of these characters is imbued with a distinct and gradual vision of reality: Europeanizing or indigenous.

Learning analytics researchers often analyze qualitative student data such as coded annotations or interview transcripts to understand learning processes. With the rise of generative AI, fully automated and human-AI workflows have emerged as promising methods for analysis. However, methodological standards to guide such workflows remain limited. In this study, we propose that reasoning traces generated by large language model (LLM) agents, especially within multi-agent systems, constitute a novel and rich form of process data to enhance interpretive practices in qualitative coding. We apply cosine similarity to LLM reasoning traces to systematically detect, quantify, and interpret disagreements among agents, reframing disagreement as a meaningful analytic signal. Analyzing nearly 10,000 instances of agent pairs coding human tutoring dialog segments, we show that LLM agents' semantic reasoning similarity robustly differentiates consensus from disagreement and correlates with human coding reliability. Qualitative analysis guided by this metric reveals nuanced instructional sub-functions within codes and opportunities for conceptual codebook refinement. By integrating quantitative similarity metrics with qualitative review, our method has the potential to improve and accelerate establishing inter-rater reliability during coding by surfacing interpretive ambiguity, especially when LLMs collaborate with humans. We discuss how reasoning-trace disagreements represent a valuable new class of analytic signals advancing methodological rigor and interpretive depth in educational research.

Automatic Speech Recognition (ASR) is traditionally evaluated using Word Error Rate (WER), a metric that is insensitive to meaning. Embedding-based semantic metrics are better correlated with human perception, but decoder-based Large Language Models (LLMs) remain underexplored for this task. This paper evaluates their relevance through three approaches: (1) selecting the best hypothesis between two candidates, (2) computing semantic distance using generative embeddings, and (3) qualitative classification of errors. On the HATS dataset, the best LLMs achieve 92–94% agreement with human annotators for hypothesis selection, compared to 63% for WER, also outperforming semantic metrics. Embeddings from decoder-based LLMs show performance comparable to encoder models. Finally, LLMs offer a promising direction for interpretable and semantic ASR evaluation.