Hasil untuk "African languages and literature"

Carine P. Mukamakuza, Monika Lanzenberger, George Metakides et al.

Artificial intelligence (AI) has become a transformative force across global societies, reshaping the ways we communicate, collaborate, and make decisions. Yet, as AI systems increasingly mediate interactions between humans, questions about the ability to take into account and understand culture, language, and context have taken center stage. This book explores these questions through a series of articles that try to assess AI's capacity to navigate cross-cultural, multilingual, and high-stakes policy environments, emphasizing human-centered approaches that balance technological innovation with social equity. It brings together six case studies from the First African Digital Humanism Summer School that took place in Kigali, Rwanda in July 2025.

Aishwarya Verma, Laud Ammah, Olivia Nercy Ndlovu Lucas et al.

Stereotype repositories are critical to assess generative AI model safety, but currently lack adequate global coverage. It is imperative to prioritize targeted expansion, strategically addressing existing deficits, over merely increasing data volume. This work introduces a multilingual stereotype resource covering four sub-Saharan African countries that are severely underrepresented in NLP resources: Ghana, Kenya, Nigeria, and South Africa. By utilizing socioculturally-situated, community-engaged methods, including telephonic surveys moderated in native languages, we establish a reproducible methodology that is sensitive to the region's complex linguistic diversity and traditional orality. By deliberately balancing the sample across diverse ethnic and demographic backgrounds, we ensure broad coverage, resulting in a dataset of 3,534 stereotypes in English and 3,206 stereotypes across 15 native languages.

Patrick Belibi

In each postcolonial multilingual environment, a nativised and standardising form of English is commonly used in the country's educational system, administration, and media, rather than Standard British English (SBE). Despite research indicating that teachers struggle to effectively teach SBE and learners consistently fail to acquire its features, educational authorities in these countries continue to promote SBE. This approach hinders the possibility of embracing flexible language ideologies that could lead to the development of local varieties of English and their acceptance as models for the classroom. In this article, I review the literature on attitudes towards postcolonial nativised Englishes and argue that adopting these Englishes as local standards has significant implications for the local English language teaching industries at sociolinguistic, pedagogic, economic, and policy levels.    

Rahul Porwal, Alice Rozet, Pryce Houck et al.

African American English (AAE) presents unique challenges in natural language processing (NLP). This research systematically compares the performance of available NLP models--rule-based, transformer-based, and large language models (LLMs)--capable of identifying key grammatical features of AAE, namely Habitual Be and Multiple Negation. These features were selected for their distinct grammatical complexity and frequency of occurrence. The evaluation involved sentence-level binary classification tasks, using both zero-shot and few-shot strategies. The analysis reveals that while LLMs show promise compared to the baseline, they are influenced by biases such as recency and unrelated features in the text such as formality. This study highlights the necessity for improved model training and architectural adjustments to better accommodate AAE's unique linguistic characteristics. Data and code are available.

Jack Liell-Cock, Zev Shirazi, Sam Staton

Relative monads provide a controlled view of computation. We generalise the monadic metalanguage to a relative setting and give a complete semantics with strong relative monads. Adopting this perspective, we generalise two existing program calculi from the literature. We provide a linear-non-linear language for graded monads, LNL-RMM, along with a semantic proof that it is a conservative extension of the graded monadic metalanguage. Additionally, we provide a complete semantics for the arrow calculus, showing it is a restricted relative monadic metalanguage. This motivates the introduction of ARMM, a computational lambda calculus-style language for arrows that conservatively extends the arrow calculus.

Aarush Sinha, Viraj Virk, Dipshikha Chakraborty et al.

Language Models [LMs] are now playing an increasingly large role in information generation and synthesis; the representation of scientific knowledge in these systems needs to be highly accurate. A prime challenge is hallucination; that is, generating apparently plausible but actually false information, including invented citations and nonexistent research papers. This kind of inaccuracy is dangerous in all the domains that require high levels of factual correctness, such as academia and education. This work presents a pipeline for evaluating the frequency with which language models hallucinate in generating responses in the scientific literature. We propose ArxEval, an evaluation pipeline with two tasks using ArXiv as a repository: Jumbled Titles and Mixed Titles. Our evaluation includes fifteen widely used language models and provides comparative insights into their reliability in handling scientific literature.

Rebecca Dorn, Christina Chance, Casandra Rusti et al.

Automated emotion detection is widely used in applications ranging from well-being monitoring to high-stakes domains like mental health and hiring. However, models often rely on annotations that reflect dominant cultural norms, limiting model ability to recognize emotional expression in dialects often excluded from training data distributions, such as African American Vernacular English (AAVE). This study examines emotion recognition model performance on AAVE compared to General American English (GAE). We analyze 2.7 million tweets geo-tagged within Los Angeles. Texts are scored for strength of AAVE using computational approximations of dialect features. Annotations of emotion presence and intensity are collected on a dataset of 875 tweets with both high and low AAVE densities. To assess model accuracy on a task as subjective as emotion perception, we calculate community-informed "silver" labels where AAVE-dense tweets are labeled by African American, AAVE-fluent (ingroup) annotators. On our labeled sample, GPT and BERT-based models exhibit false positive prediction rates of anger on AAVE more than double than on GAE. SpanEmo, a popular text-based emotion model, increases false positive rates of anger from 25 percent on GAE to 60 percent on AAVE. Additionally, a series of linear regressions reveals that models and non-ingroup annotations are significantly more correlated with profanity-based AAVE features than ingroup annotations. Linking Census tract demographics, we observe that neighborhoods with higher proportions of African American residents are associated with higher predictions of anger (Pearson's correlation r = 0.27) and lower joy (r = -0.10). These results find an emergent safety issue of emotion AI reinforcing racial stereotypes through biased emotion classification. We emphasize the need for culturally and dialect-informed affective computing systems.

Emily Swanson

Abraham Toluwase Owodunni, Aditya Yadavalli, Chris Chinenye Emezue et al.

Despite advancements in speech recognition, accented speech remains challenging. While previous approaches have focused on modeling techniques or creating accented speech datasets, gathering sufficient data for the multitude of accents, particularly in the African context, remains impractical due to their sheer diversity and associated budget constraints. To address these challenges, we propose AccentFold, a method that exploits spatial relationships between learned accent embeddings to improve downstream Automatic Speech Recognition (ASR). Our exploratory analysis of speech embeddings representing 100+ African accents reveals interesting spatial accent relationships highlighting geographic and genealogical similarities, capturing consistent phonological, and morphological regularities, all learned empirically from speech. Furthermore, we discover accent relationships previously uncharacterized by the Ethnologue. Through empirical evaluation, we demonstrate the effectiveness of AccentFold by showing that, for out-of-distribution (OOD) accents, sampling accent subsets for training based on AccentFold information outperforms strong baselines a relative WER improvement of 4.6%. AccentFold presents a promising approach for improving ASR performance on accented speech, particularly in the context of African accents, where data scarcity and budget constraints pose significant challenges. Our findings emphasize the potential of leveraging linguistic relationships to improve zero-shot ASR adaptation to target accents.

Lance Calvin Lim Gamboa, Mark Lee

Bias studies on multilingual models confirm the presence of gender-related stereotypes in masked models processing languages with high NLP resources. We expand on this line of research by introducing Filipino CrowS-Pairs and Filipino WinoQueer: benchmarks that assess both sexist and anti-queer biases in pretrained language models (PLMs) handling texts in Filipino, a low-resource language from the Philippines. The benchmarks consist of 7,074 new challenge pairs resulting from our cultural adaptation of English bias evaluation datasets, a process that we document in detail to guide similar forthcoming efforts. We apply the Filipino benchmarks on masked and causal multilingual models, including those pretrained on Southeast Asian data, and find that they contain considerable amounts of bias. We also find that for multilingual models, the extent of bias learned for a particular language is influenced by how much pretraining data in that language a model was exposed to. Our benchmarks and insights can serve as a foundation for future work analyzing and mitigating bias in multilingual models.

Kétévi A. Assamagan, Abdelkarim Boskri, Kenneth Cecire et al.

On April 15-19, 2024, as a part of the eighth African School of Physics, ASP2024, we organized a program for learners from selected high schools in the vicinity of Marrakesh, Morocco. In this event, within a week, we reached out to over a thousand high school students, aka learners, from many high schools in the region of Marrakesh. We present a summary report on these outreach activities.

Shawna-Leze Meiring

Martine Vanhove

Gagan Bhatia, Ife Adebara, AbdelRahim Elmadany et al.

We describe our contribution to the SemEVAl 2023 AfriSenti-SemEval shared task, where we tackle the task of sentiment analysis in 14 different African languages. We develop both monolingual and multilingual models under a full supervised setting (subtasks A and B). We also develop models for the zero-shot setting (subtask C). Our approach involves experimenting with transfer learning using six language models, including further pertaining of some of these models as well as a final finetuning stage. Our best performing models achieve an F1-score of 70.36 on development data and an F1-score of 66.13 on test data. Unsurprisingly, our results demonstrate the effectiveness of transfer learning and fine-tuning techniques for sentiment analysis across multiple languages. Our approach can be applied to other sentiment analysis tasks in different languages and domains.

D.J.M. Steijger, Chandrima Chatterjee, W. Groot et al.

Background: Cost-effectiveness is a tool to maximize health benefits and to improve efficiency in healthcare. However, efficient outcomes are not always the most equitable ones. Distributional cost-effectiveness analysis (DCEA) offers a framework for incorporating equity concerns into cost-effectiveness analysis. Objective: This systematic review aims to outline the challenges and limitations in applying DCEA in healthcare settings. Methods: We searched Medline, Scopus, BASE, APA Psych, and JSTOR databases. We also included Google Scholar. We searched for English-language peer-reviewed academic publications, while books, editorials and commentary papers were excluded. Titles and abstract screening, full-text screening, reference list reviews, and data extraction were performed by the main researcher. Another researcher checked every paper for eligibility. Details, such as study population, disease area, intervention and comparators, costs and health effects, cost-effectiveness findings, equity analysis and effects, and modelling technique, were extracted. Thematic analysis was applied, focusing on challenges, obstacles, and gaps in DCEA. Results: In total, 615 references were identified, of which 18 studies met the inclusion criteria. Most of these studies were published after 2017. DCEA studies were mainly conducted in Europe and Africa and used quality health-adjusted measurements. In the included studies, absolute inequality indices were used more frequently than relative inequality indices. Every stage of the DCEA presented challenges and/or limitations. Conclusion: This review provides an overview of the literature on the DCEA in healthcare as well as the challenges and limitations related to the different steps needed to conduct the analysis. In particular, we found problems with data availability, the relative unfamiliarity of this analysis among policymakers, and challenges in estimating differences among socioeconomic groups.

Brianne Wenning, Helen Price, Hasara Nuwangi et al.

Background More than one million people each year become infected by parasites that cause the disease cutaneous leishmaniasis (CL). This disease manifests as one or more skin lesions or ulcers that are slow to heal with variable response rates to drug treatments. Thus far, little attention has been paid to how the cultural effects of gender shape perceptions and experiences of CL. This review aims to bring together and analyse existing studies which use qualitative data to explore these differences. These studies offered insights into our specific research questions. Methods We conducted a systematic review of the literature pertaining to either CL or muco-cutaneous leishmaniasis (MCL) through EBSCO, EMBASE, Medline, Scopus and Web of Science databases. To meet inclusion criteria, articles had to be either qualitative or mixed-method with a qualitative component. They also had to include a reflection on how the gender of participants impacted the findings and addressed the lived experiences of CL. We did not exclude articles based on the language they were published in or in which country the study took place. Results From a total of 1589 potential articles, we found that thirteen met the inclusion criteria. These articles were published in English, Spanish or Portuguese and reported on studies carried out in various countries in Africa, Asia and South America. After using the principles of a meta-ethnography to analyse these studies, we generated several key themes. We found that health-seeking behaviours, treatment choices, stigma and the impact of scarring are shaped by gender in a variety of contexts. Conclusions Gender impacts on an individual’s experience of CL. In particular, women are more constricted in their health-seeking behaviours and experience more stigma both from the active lesions and from scarring than men. In many contexts, however, men are more at risk of becoming infected by the parasite that causes CL and may turn to more harmful or aggressive self-treatments. We recommend that future research on CL should consider the impact of gender as this can create very different experiences for individuals.

F. Akpa-Inyang, S. Chima

Background The Western-European concept of libertarian rights-based autonomy, which advocates respect for individual rights, may conflict with African cultural values and norms. African communitarian ethics focuses on the interests of the collective whole or community, rather than rugged individualism. Hence collective decision-making processes take precedence over individual autonomy or consent. This apparent conflict may impact informed consent practice during biomedical research in African communities and may hinder ethical principlism in African bioethics. This study explored African biomedical researchers' perspectives regarding informed consent and potential limitations to the principle of respect for autonomy in African communities. Methods We conducted a qualitative study based on in-depth interviews with 12 biomedical researchers, five females and seven males aged 34 to 74 years, currently working at an African university. Interviews lasted 35–40 min each and involved semi-structured open-ended interviews, which allowed participants to offer information about their perceptions and feelings regarding respect for autonomy and informed consent as practised in Africa. Empirical data from the interviews were recorded, transcribed, and analysed using thematic content analysis, together with an interrogation of relevant scientific literature about African communitarian ethics, making evaluations and drawing inferences consistent with the empirical bioethics approach. Results Based on these interviews and analysis of relevant literature, we found that informed consent is difficult to apply in an African context because it derives from a Western conception of libertarian rights-based autonomy. Most respondents pointed out that it was challenging to implement informed consent in the African setting. Furthermore, communalism, customary beliefs, spirituality, and relational autonomy are predominant in most African communities, as exemplified by the African moral philosophies of Ubuntu/Botho and Ukama , which emphasize communitarianism over individual rights. We also found that language, education, poverty, and cultural beliefs are barriers to obtaining proper informed consent in African communities. Conclusions We conclude that there are limitations to applying the principle of respect for autonomy and informed consent in African communities, especially in the context of human biomedical research. We recommend using a more relational approach, such as Ross’s prima facie duties, to implement informed consent in African communities.

Hanyan Zhang, M. Han

Abstract Pocket parks have the characteristics of being small in size, low in expenditure, easy to access and flexible in shapes and location. Several studies have explored the environmental and social benefits as well as the design principles of pocket parks. However, a meta-analysis of research on pocket park is lacking. There is also no previous comparison of Chinese and English language research papers on pocket parks. This paper aims to systematically review 32 key English publications and 33 key Chinese publications from 4 databases (i.e. Google Scholar, Web of Science, CNKI, Wanfang Data). The results show that the definition of pocket park is only different in size when comparing English and Chinese papers. Moreover, the review highlighted the needs of exploring pocket parks in the areas of Latin America, Africa and Southeast Asia. The summary of research topics and trends provides a useful knowledge for future studies on pocket parks.

April Monroe, Sarah Moore, Hannah Koenker et al.

BackgroundMalaria cases and deaths decreased dramatically in recent years, largely due to effective vector control interventions. Persistence of transmission after good coverage has been achieved with high-quality vector control interventions, namely insecticide-treated nets or indoor residual spraying, poses a significant challenge to malaria elimination efforts. To understand when and where remaining transmission is occurring, it is necessary to look at vector and human behaviour, and where they overlap. To date, a review of human behaviour related to residual malaria transmission has not been conducted.MethodsStudies were identified through PubMed and Google Scholar. Hand searches were conducted for all references cited in articles identified through the initial search. The review was limited to English language articles published between 2000 and 2017. Publications with primary data from a malaria endemic setting in sub-Saharan Africa and a description of night time human behaviours were included.ResultsTwenty-six publications were identified that met inclusion criteria. Study results fit into two broad categories: when and where people are exposed to malaria vectors and what people are doing at night that may increase their contact with malaria vectors. Among studies that quantified human-vector interaction, a majority of exposure occurred indoors during sleeping hours for unprotected individuals, with some variation across time, contexts, and vector species. Common night time activities across settings included household chores and entertainment during evening hours, as well as livelihood and large-scale socio-cultural events that can last throughout the night. Shifting sleeping patterns associated with travel, visitors, illness, farming practices, and outdoor sleeping, which can impact exposure and use of prevention measures, were described in some locations.ConclusionsWhile the importance of understanding human-vector interaction is well-established, relatively few studies have included human behaviour when measuring exposure to malaria vectors. Broader application of a standardized approach to measuring human-vector interaction could provide critical information on exposure across settings and over time. In-depth understanding of night time activities that occur during times when malaria vectors are active and barriers to prevention practices in different contexts should also be considered. This information is essential for targeting existing interventions and development and deployment of appropriate complementary prevention tools.

Khang Nhut Lam, Feras Al Tarouti, Jugal Kalita

This paper examines approaches to generate lexical resources for endangered languages. Our algorithms construct bilingual dictionaries and multilingual thesauruses using public Wordnets and a machine translator (MT). Since our work relies on only one bilingual dictionary between an endangered language and an "intermediate helper" language, it is applicable to languages that lack many existing resources.