 Your new post is loading...
|
Scooped by
Charles Tiayon
November 16, 9:55 PM
|
Faut-il parler à ChatGPT et aux IA en français ? Cette étude classe les 26 meilleures langues à utiliser Si vous avez pris l'habitude de parler à ChatGPT ou Gemini en anglais pour avoir de meilleurs résultats, vous pouvez maintenant oublier car ce n'est pas une bonne solution. Une étude vient de chambouler nos certitudes, et le français tire brillamment son épingle du jeu !
Beaucoup de personnes ont ce réflexe : quand on veut une réponse vraiment pointue de ChatGPT, de Gemini ou de n'importe quelle IA générative, on bascule sur l'anglais. On se dit, logiquement, que la langue de Shakespeare, celle qui domine internet et les bases de données d'entraînement, donnera forcément un résultat plus pertinent. Eh bien, mettez cette idée à la poubelle. Une étude très sérieuse, menée conjointement par des chercheurs de l'Université du Maryland et de Microsoft, vient de détruire cette croyance. L'anglais détrôné, le français en embuscade !
C'est le coup de tonnerre de ce rapport. Les chercheurs ont testé la performance d'une flopée de modèles (OpenAI, Google Gemini, Qwen, Llama...) sur 26 langues différentes pour voir comment ils comprenaient les demandes, appelés prompts. Le résultat est sans appel. Dans leur rapport d'étude, les auteurs ne cachent pas leur étonnement : Notre expérience a donné lieu à des résultats surprenants et peu intuitifs. Tout d'abord, l'anglais n'a pas obtenu les meilleurs résultats dans tous les modèles ; en fait, il est arrivé en sixième position sur 26 langues lors de l'évaluation de textes longs... Sixième ! L'anglais, que l'on pensait roi, se fait griller la politesse. Mais alors, qui est sur le podium ? Cocorico (ou presque), le français se hisse à une magnifique deuxième place. Mais le grand gagnant, celui auquel personne ne s'attendait, c'est le polonais.
Voici le top 10 des langues les plus efficaces pour parler à une IA : Polonais (88%)
Français (87%)
Italien (86%)
Espagnol (85%)
Russe (84%)
Anglais (83,9%)
Ukrainien (83,5%)
Portugais (82%)
Allemand (81%)
Néerlandais (80%)
Précision grammaticale > Volume de données ?
Mais alors, comment le polonais, une langue réputée pour sa complexité (oui, ses déclinaisons font passer le latin pour un jeu d'enfant) et disposant d'un corpus de données d'entraînement bien plus faible que l'anglais, peut-il être le plus efficace ? L'Office polonais des brevets, qui s'est félicité de la nouvelle sur Facebook, avance une explication : c'est la langue la plus précise pour donner des ordres à l'intelligence artificielle... Ce que l'étude semble montrer, c'est que la précision structurelle d'une langue pourrait être plus importante que le volume de données brutes. L'anglais est une langue relativement simple, qui repose beaucoup sur le contexte et l'ordre des mots. Le polonais, le français ou l'allemand, avec leurs conjugaisons plus riches, leurs genres grammaticaux et leurs déclinaisons (surtout pour le polonais et l'allemand), ne laissent que peu de place à l'ambiguïté. Un ordre en polonais est, semble-t-il, verrouillé grammaticalement. L'IA sait exactement qui fait quoi et comment. L'étude note d'ailleurs que toutes les langues du top 10 appartiennent à la famille indo-européenne. À l'inverse, le chinois, malgré un volume de données colossal, s'est classé avant-dernier, suggérant que sa structure est peut-être moins adaptée aux architectures actuelles des IA. Faut-il donc tous se mettre au polonais pour rédiger nos prompts ? Pas du tout. Ce que cette étude nous dit surtout, c'est que si vous êtes francophone, vous n'avez absolument aucune raison de passer à l'anglais pour obtenir de meilleurs résultats. Votre langue maternelle est déjà l'une des plus performantes du monde pour dialoguer avec l'IA.
https://share.google/BlCSvSmTzdKYCfiS5
#Metaglossia
#metaglossia_mundus
Researchers across Africa, Asia and the Middle East are building their own language models designed for local tongues, cultural nuance and digital independence
"In a high-stakes artificial intelligence race between the United States and China, an equally transformative movement is taking shape elsewhere. From Cape Town to Bangalore, from Cairo to Riyadh, researchers, engineers and public institutions are building homegrown AI systems, models that speak not just in local languages, but with regional insight and cultural depth.
The dominant narrative in AI, particularly since the early 2020s, has focused on a handful of US-based companies like OpenAI with GPT, Google with Gemini, Meta’s LLaMa, Anthropic’s Claude. They vie to build ever larger and more capable models. Earlier in 2025, China’s DeepSeek, a Hangzhou-based startup, added a new twist by releasing large language models (LLMs) that rival their American counterparts, with a smaller computational demand. But increasingly, researchers across the Global South are challenging the notion that technological leadership in AI is the exclusive domain of these two superpowers.
Instead, scientists and institutions in countries like India, South Africa, Egypt and Saudi Arabia are rethinking the very premise of generative AI. Their focus is not on scaling up, but on scaling right, building models that work for local users, in their languages, and within their social and economic realities.
“How do we make sure that the entire planet benefits from AI?” asks Benjamin Rosman, a professor at the University of the Witwatersrand and a lead developer of InkubaLM, a generative model trained on five African languages. “I want more and more voices to be in the conversation”.
Beyond English, beyond Silicon Valley
Large language models work by training on massive troves of online text. While the latest versions of GPT, Gemini or LLaMa boast multilingual capabilities, the overwhelming presence of English-language material and Western cultural contexts in these datasets skews their outputs. For speakers of Hindi, Arabic, Swahili, Xhosa and countless other languages, that means AI systems may not only stumble over grammar and syntax, they can also miss the point entirely.
“In Indian languages, large models trained on English data just don’t perform well,” says Janki Nawale, a linguist at AI4Bharat, a lab at the Indian Institute of Technology Madras. “There are cultural nuances, dialectal variations, and even non-standard scripts that make translation and understanding difficult.” Nawale’s team builds supervised datasets and evaluation benchmarks for what specialists call “low resource” languages, those that lack robust digital corpora for machine learning.
It’s not just a question of grammar or vocabulary. “The meaning often lies in the implication,” says Vukosi Marivate, a professor of computer science at the University of Pretoria, in South Africa. “In isiXhosa, the words are one thing but what’s being implied is what really matters.” Marivate co-leads Masakhane NLP, a pan-African collective of AI researchers that recently developed AFROBENCH, a rigorous benchmark for evaluating how well large language models perform on 64 African languages across 15 tasks. The results, published in a preprint in March, revealed major gaps in performance between English and nearly all African languages, especially with open-source models.
Similar concerns arise in the Arabic-speaking world. “If English dominates the training process, the answers will be filtered through a Western lens rather than an Arab one,” says Mekki Habib, a robotics professor at the American University in Cairo. A 2024 preprint from the Tunisian AI firm Clusterlab finds that many multilingual models fail to capture Arabic’s syntactic complexity or cultural frames of reference, particularly in dialect-rich contexts.
Governments step in
For many countries in the Global South, the stakes are geopolitical as well as linguistic. Dependence on Western or Chinese AI infrastructure could mean diminished sovereignty over information, technology, and even national narratives. In response, governments are pouring resources into creating their own models.
Saudi Arabia’s national AI authority, SDAIA, has built ‘ALLaM,’ an Arabic-first model based on Meta’s LLaMa-2, enriched with more than 540 billion Arabic tokens. The United Arab Emirates has backed several initiatives, including ‘Jais,’ an open-source Arabic-English model built by MBZUAI in collaboration with US chipmaker Cerebras Systems and the Abu Dhabi firm Inception. Another UAE-backed project, Noor, focuses on educational and Islamic applications.
In Qatar, researchers at Hamad Bin Khalifa University, and the Qatar Computing Research Institute, have developed the Fanar platform and its LLMs Fanar Star and Fanar Prime. Trained on a trillion tokens of Arabic, English, and code, Fanar’s tokenization approach is specifically engineered to reflect Arabic’s rich morphology and syntax.
India has emerged as a major hub for AI localization. In 2024, the government launched BharatGen, a public-private initiative funded with 235 crore (€26 million) initiative aimed at building foundation models attuned to India’s vast linguistic and cultural diversity. The project is led by the Indian Institute of Technology in Bombay and also involves its sister organizations in Hyderabad, Mandi, Kanpur, Indore, and Madras. The programme’s first product, e-vikrAI, can generate product descriptions and pricing suggestions from images in various Indic languages. Startups like Ola-backed Krutrim and CoRover’s BharatGPT have jumped in, while Google’s Indian lab unveiled MuRIL, a language model trained exclusively on Indian languages. The Indian governments’ AI Mission has received more than180 proposals from local researchers and startups to build national-scale AI infrastructure and large language models, and the Bengaluru-based company, AI Sarvam, has been selected to build India’s first ‘sovereign’ LLM, expected to be fluent in various Indian languages.
In Africa, much of the energy comes from the ground up. Masakhane NLP and Deep Learning Indaba, a pan-African academic movement, have created a decentralized research culture across the continent. One notable offshoot, Johannesburg-based Lelapa AI, launched InkubaLM in September 2024. It’s a ‘small language model’ (SLM) focused on five African languages with broad reach: Swahili, Hausa, Yoruba, isiZulu and isiXhosa.
“With only 0.4 billion parameters, it performs comparably to much larger models,” says Rosman. The model’s compact size and efficiency are designed to meet Africa’s infrastructure constraints while serving real-world applications. Another African model is UlizaLlama, a 7-billion parameter model developed by the Kenyan foundation Jacaranda Health, to support new and expectant mothers with AI-driven support in Swahili, Hausa, Yoruba, Xhosa, and Zulu.
India’s research scene is similarly vibrant. The AI4Bharat laboratory at IIT Madras has just released IndicTrans2, that supports translation across all 22 scheduled Indian languages. Sarvam AI, another startup, released its first LLM last year to support 10 major Indian languages. And KissanAI, co-founded by Pratik Desai, develops generative AI tools to deliver agricultural advice to farmers in their native languages.
The data dilemma
Yet building LLMs for underrepresented languages poses enormous challenges. Chief among them is data scarcity. “Even Hindi datasets are tiny compared to English,” says Tapas Kumar Mishra, a professor at the National Institute of Technology, Rourkela in eastern India. “So, training models from scratch is unlikely to match English-based models in performance.”
Rosman agrees. “The big-data paradigm doesn’t work for African languages. We simply don’t have the volume.” His team is pioneering alternative approaches like the Esethu Framework, a protocol for ethically collecting speech datasets from native speakers and redistributing revenue back to further development of AI tools for under-resourced languages. The project’s pilot used read speech from isiXhosa speakers, complete with metadata, to build voice-based applications.
In Arab nations, similar work is underway. Clusterlab’s 101 Billion Arabic Words Dataset is the largest of its kind, meticulously extracted and cleaned from the web to support Arabic-first model training.
The cost of staying local
But for all the innovation, practical obstacles remain. “The return on investment is low,” says KissanAI’s Desai. “The market for regional language models is big, but those with purchasing power still work in English.” And while Western tech companies attract the best minds globally, including many Indian and African scientists, researchers at home often face limited funding, patchy computing infrastructure, and unclear legal frameworks around data and privacy.
“There’s still a lack of sustainable funding, a shortage of specialists, and insufficient integration with educational or public systems,” warns Habib, the Cairo-based professor. “All of this has to change.”
A different vision for AI
Despite the hurdles, what’s emerging is a distinct vision for AI in the Global South – one that favours practical impact over prestige, and community ownership over corporate secrecy.
“There’s more emphasis here on solving real problems for real people,” says Nawale of AI4Bharat. Rather than chasing benchmark scores, researchers are aiming for relevance: tools for farmers, students, and small business owners.
And openness matters. “Some companies claim to be open-source, but they only release the model weights, not the data,” Marivate says. “With InkubaLM, we release both. We want others to build on what we’ve done, to do it better.”
In a global contest often measured in teraflops and tokens, these efforts may seem modest. But for the billions who speak the world’s less-resourced languages, they represent a future in which AI doesn’t just speak to them, but with them."
Sibusiso Biyela, Amr Rageh and Shakoor Rather
20 May 2025
https://www.natureasia.com/en/nmiddleeast/article/10.1038/nmiddleeast.2025.65
#metaglossia_mundus
Dans un monde du travail menacé par l'automatisation, Jeff Bezos est catégorique : il existe une qualité humaine qui résistera toujours aux algorithmes. Lors d'un événement majeur en Italie, le fondateur d'Amazon a révélé quel profil de salarié demeurera irremplaçable
L'inventivité, cette qualité irremplaçable
...Le milliardaire ne mise pas sur les diplômes ou l'expérience académique, mais sur une capacité bien plus fondamentale : l'esprit d'inventeur. Cette qualité, selon lui, constitue le rempart ultime face à l'automatisation.
Cette philosophie trouve ses racines dans les étés passés au ranch de son grand-père au Texas. Lawrence Preston Gise, son grand-père maternel, incarnait cette ingéniosité pratique que Bezos valorise tant. "Il a acheté un bulldozer pour environ 5 000 dollars parce qu'il était complètement cassé. Nous avons passé tout un été à le réparer. Pour démonter la transmission, nous avons dû construire notre propre grue", se souvient-il.
Amazon recrute des inventeurs, pas des exécutants
...Lors d'une interview au Utah Technology Council en 2012, Bezos révélait sa question d'entretien favorite : "Quand j'interviewe des candidats, je leur demande de me donner un exemple de quelque chose qu'ils ont inventé". Il précise que cette "invention" n'implique pas nécessairement un brevet, mais peut être une nouvelle métrique, un processus optimisé ou une solution inédite à un problème existant.
Cette approche n'est pas qu'un exercice rhétorique. Amazon recherche activement des personnes capables de penser en dehors des sentiers battus. Bezos explique que les candidats qui voient immédiatement les choix en termes de "soit A, soit B" ne correspondent pas au profil recherché. "La bonne question est : comment peut-on faire A ET B? Quelle invention nous permettrait de réaliser les deux?", insiste-t-il.
L'entreprise va jusqu'à préférer ne recruter personne plutôt que d'embaucher un profil inadéquat. "Quelqu'un qui arrive chez Amazon et n'aime pas être pionnier, qui n'aime pas explorer ou s'aventurer dans des impasses, ne restera pas longtemps", affirme le fondateur. Cette culture de l'expérimentation et de l'innovation constitue le cœur du modèle Amazon.
Le constat d'Andy Jassy, successeur de Bezos à la tête d'Amazon, rejoint cette analyse. Pour le PDG actuel, ce ne sont pas les connaissances qui distingueront les talents de demain, mais les attitudes et la capacité d'apprentissage...
Amazon utilise désormais l'intelligence artificielle dans "pratiquement tous les recoins" de l'entreprise : 500 000 vendeurs exploitent les outils d'IA pour générer du contenu produit, et de futurs agents intelligents autonomes sont en développement. L'automatisation progresse rapidement, mais l'inventivité demeure le domaine exclusif de l'humain.
Pour Bezos, l'innovation représente également un antidote contre une peur qui l'a toujours animé : celle des "deux jeunes dans un garage". "Ils me font plus peur que les concurrents que je connais déjà", confie-t-il. Cette référence aux origines légendaires de HP et Apple souligne que l'inventivité reste le moteur principal de la disruption, bien au-delà des capacités des algorithmes les plus sophistiqués.
https://share.google/QVeXlWhhwuHKquETC
#Metaglossia
#metaglossia_mundus
Pourquoi les IA génératives n’atteindront pas l’intelligence générale
.... Un LLM apprend en complétant des phrases ; il calcule, mot après mot, la suite statistique la plus probable depuis un corpus gigantesque de textes ingurgités (on parle ici de presque toutes les données disponibles sur Internet). Cette mécanique statistique produit un vernis de cohérence bluffant, mais elle laisse la machine aveugle au monde physique. Sans caméra ni bras pour manipuler les objets, les modèles de langage n'acquièrent ni notion de causalité, ni compréhension réelle des lois de la gravité, ni mémoire d'expériences vécues. Ils ignorent la persistance des choses ou le simple déroulement d'une action..."
https://share.google/zH8iRVoXCXCQz2L4P
#Metaglossia
#metaglossia_mundus
Un jeune normand de 23 ans crée une intelligence artificielle conversationnelle en autodidacte
Moon est la seule intelligence artificielle publique 100 % normande !...
Des intelligences artificielles (IA) conversationnelles françaises et publiques, il n’y en a pas cinquante. On retrouve d’abord Mistral, une startup fondée en 2013 et regroupant près de 200 collaborateurs, et ensuite, considérée comme la deuxième du classement, Moon, créée à Pont-Audemer (Eure) par Fabien Marie, tout seul !
Électricien de profession, le Pontaudemérien de 23 ans a toujours été passionné par l’informatique : « À 12 ans, j’apprenais les divers langages informatiques et un an plus tard, je commençais à créer des sites ». Parallèlement à ses études puis à son métier, il poursuit le développement numérique en créant un réseau social qu’il veut sécuritaire pour les plus jeunes : « L’objectif était un réseau qui empêche le cyberharcèlement et les insultes, explique-t-il. Le meilleur moyen était d’utiliser une IA qui assure la modération sur ce réseau. »
Presque cinq ans d’entraînement
Restait à disposer de sa propre intelligence artificielle… Fabien Marie a alors suivi le conseil de Winston Churchill : « Agissez comme s’il était impossible d’échouer. » Et il lui aura fallu près de cinq ans et quelques dizaines de milliers d’euros personnels investis dans du matériel ultra-performant pour créer et entraîner sa propre IA : Moon. « Pour développer Moon, il a fallu de la puissance de calcul, explique-t-il. Mais j’ai aussi élaboré une toute nouvelle méthode d’entraînement de l’IA. Elle est à présent opérationnelle et en ligne depuis plusieurs semaines, et sécuritaire également puisque les données des conversations sont cryptées et non accessibles, contrairement à certaines autres IA. »
Par ailleurs, le développeur a créé une plateforme regroupant une soixantaine d’IA internationales, dont Moon, qui a rapidement rencontré le succès, avec plusieurs centaines de connexions par jour, et plus de 60 000 inscriptions (gratuites).
Levée de fonds
Après la partie technique, le jeune eurois s’attaque à présent à la partie commerciale. Il a fondé sa société : Stellarr Studio, et a trouvé trois associés avec qui il a lancé une levée de fonds, soutenu par la CCI (Chambre de commerce et d’industrie) Seine Estuaire, et suivi par la Région Normandie. Après tout, Moon est la seule intelligence artificielle publique 100 % normande !...
https://share.google/PAoTTMM0XcakmAUoN
#Metaglossia
#metaglossia_mundus
Savez-vous que le puinave, le galibi, le dâw et le wai wai, parlés en Amérique du Sud, conjuguent au passé non seulement les verbes, mais aussi les noms ? Que le tarian (Amazonie), possède sept impératifs ? Et qu’en tchouktche (Sibérie), la manière de s’exprimer des femmes diffère nettement de celle des hommes ? Comment mieux dire que les langues ne se contentent pas de transmettre des informations, mais constituent des créations culturelles de l’humanité ? Et que toutes sont ex-tra-or-di-nai-res, et pas uniquement les plus connues d’entre elles ?..."
👇🏿👇🏿👇🏿
A la découverte de 15 langues extraordinaires de l’humanité
Sur le bout des langues. Certaines ont 120 consonnes, d’autres conjuguent les noms, d’autres encore n’ont pas de verbe "être". Bienvenue dans l’univers incroyable des langues de l’Humanité !
Par Michel Feltin-Palas
Publié le 28/10/2025 à 06:15
Sur le bout des langues – L'Express
#Metaglossia
#metaglossia_mundus
Comment le français est-il devenu une langue écrite ? - De vive(s) voix
Dans son ouvrage «Écrire le français», Gabriella Parussa retrace l'histoire de l'écrit de la langue française, et son évolution.
Comment le français est-il devenu une langue écrite ?
Le français fut tout d'abord une langue orale, une langue parlée, une langue de la communication, alors que la langue écrite, celle de la culture, de la justice, de l'administration, était le latin.
Le français a beaucoup évolué, tant au niveau de la phonétique qu'au niveau de la syntaxe et la morphologie.
Les premiers textes (rarissimes) écrits en français remontent à la deuxième partie du IXè siècle. C'est l'alphabet latin qui est utilisé pour «transférer» cette langue de l'oral à l'écrit, mais Il faut faire des ajustements, car l'alphabet latin n'est pas parfait et ne permet pas de restituer tous les sons… Il y a alors vingt-trois lettres dans l'alphabet latin.
Le passage à l'écrit s'est fait avec la volonté de conserver la mémoire. Les premiers «livres» s'écrivaient grâce à des copistes ; cela pouvait prendre des mois.
L'écriture : comme une «peinture de la voix»
En 842, les petits-fils de Charlemagne signent une alliance militaire : les serments de Strasbourg, Nithard écrit alors en ce qu'on appelle «le proto français» : un mélange entre le latin tardif et le français. C'est l'un des premiers textes conservés avec des traces de français écrit. «C'est une étape importante, on écrit dans les deux langues vernaculaires de l'époque».
Ils écrivent cette nouvelle langue en «écoutant». «On dira que l'écriture doit être comme une peinture de la voix, doit correspondre à ce que l'on dit, ce qu'on prononce».
Le français est alors très mouvant : on ne parle pas et on n'écrit pas de la même manière selon qu'on habite Arras ou Lyon. Mais les choses vont se fixer avec l'invention de l'imprimerie puis des premières grammaires écrites. Les variations vont alors s'estomper peu à peu. Les imprimeurs vont alors jouer un rôle très important dans la standardisation pour que le livre qu'ils diffusent soit lu et compris et plus grand nombre. Les graphies régionales vont alors peu à peu disparaître." https://www.rfi.fr/fr/podcasts/de-vive-s-voix/20251020-comment-le-fran%C3%A7ais-est-il-devenu-une-langue-%C3%A9crite
#Metaglossia
#metaglossia_mundus
"A sesquicentennial gift of light: Nearly 150 BYU speeches now available in French, Portuguese
By Abigail Degn, October 08, 2025
Thanks to the translation efforts of BYU Speeches, those addresses are already available in English, Spanish and Japanese as well
Thanks to the efforts of BYU Speeches, this recent devotional by Elder Ronald A. Rasband will be available to millions of people around the globe.
Every week, thousands of students gather on campus to be uplifted and inspired by remarkable devotional speakers. With the help of translators within BYU Speeches, thousands more across the globe can now access the spiritual reinforcement these speeches provide.
Since 2023, more than 100 devotionals have been translated in Spanish and Japanese. But now, in BYU’s 150th anniversary year, the speeches are now available in French and Portuguese as well, with possibly more on the way.
“As we expand our translation efforts, we hope to share the unique light of BYU and bring messages of hope to people around the world,” said Alayna Een, an editor with BYU Speeches.
Since 1972, BYU has published its weekly devotionals for public access, going online in 1996 and extending the reach of BYU devotionals to anyone with an internet connection. The Church of Jesus Christ of Latter-day Saints has grown significantly since 1972, and many members across the globe are looking for uplifting gospel content like BYU devotionals.
In an effort to translate speeches into more languages, BYU speeches formed a partnership with the BYU College of Humanities, which provides funding and special learning opportunities for students and language speakers. BYU Speeches now has a small team of students overseeing the translation of these talks.
... Peter Demars: "We're learning spiritually from what we're translating."
Machine translation produces an initial translation of each speech, the language students review and clean up the draft, and then the speech is sent off to native speakers for a final evaluation. The text is then uploaded to the website, made available for all visitors to read or listen to...
A short video about the translation efforts of BYU Speeches
The speeches are now more accessible than ever before — content is regularly uploaded to the BYU Speeches podcast, website, Facebook page, and YouTube channel, which has over 238,000 subscribers. To explore BYU Speeches, go to https://speeches.byu.edu/# and if you want to explore the speeches available in different languages, try these links:
Spanish: https://speeches.byu.edu/spa/
French: https://speeches.byu.edu/fra/
Portuguese: https://speeches.byu.edu/por/
Japanese: https://speeches.byu.edu/jpn/
Media Contact: Todd Hollingshead"
https://news.byu.edu/faith/another-gift-of-light-nearly-150-byu-speeches-now-available-in-french-and-portuguese-in-addition-to-spanish-and-japanese
#Metaglossia
#metaglossia_mundus
"WhatsApp for iPhone adds built-in message translation in 21 languages
After releasing it on Android a few weeks ago, WhatsApp is now rolling out its own translation tool on iOS. Here’s how it works.
Just a few weeks after rolling out AI-based summarization and writing suggestions, WhatsApp is now rolling out support for message translations in 21 different languages.
The feature, which is part of version 25.28.74, comes alongside the official launch of Live Photos support. Meta states that it may take a few weeks for the rollout to reach all users.
How to translate messages on WhatsApp
To see if you already have access, tap and hold any message, select “More…,” and check for the “Translate” option in the menu.
After selecting your source and target languages for the first time, you’ll be prompted to download the corresponding language packs to your device. Once downloaded, you’ll be able to translate messages even without an active connection. You can download new language packs at any time. Downloaded languages can be found and managed under Settings > Translate > Downloaded Languages.
Here is the full list of currently supported languages, according to WABetaInfo, which also notes that the “available languages depend on the iOS version installed on the device”:
Arabic
Chinese (Mandarin, Simplified)
Chinese (Mandarin, Traditional)
Dutch
English (UK)
English (US)
French
German
Hindi
Indonesian
Italian
Japanese
Korean
Polish
Portuguese (Brazil)
Russian
Spanish
Thai
Turkish
Ukrainian
Vietnamese
And here’s what Meta says about the feature’s limitations and requirements:
”You can translate messages you’ve sent or received in an individual or group chat into some languages. You can choose from the available languages to translate messages, but locations, documents, contacts, stickers, and GIFs can’t be translated. You’ll need a strong WiFi signal or cellular data, and ample storage space to translate.”
WhatsApp is available for free on the App Store"
Marcus Mendes
| Oct 7 2025 - 1:41 pm PT
https://9to5mac.com/2025/10/07/whatsapp-for-iphone-adds-built-in-message-translation-in-21-languages/
#Metaglossia
#metaglossia_mundus
"...Le devoir et la tâche d’un écrivain sont ceux d’un traducteur.
–Marcel Proust, Le Temps retrouvé
*
The duty and the task of a writer are those of a translator.
–The Past Recaptured, translated by Frederick A. Blossom
*
The function and the task of a writer are those of a translator.
–Time Regained, translated by Andreas Mayor
*
The writer’s task and duty are those of a translator.
–Finding Time Again, translated by Ian Patterson
*
In 1919, Charles Kenneth Scott Moncrieff, spurred by the challenge of Marcel Proust’s winding sentences, began privately translating Du Côté de chez Swann, the first volume of À la recherche du temps perdu, into English. Swann had come out in France in 1913, but a reissue and its sequel, À l’ombre des jeunes filles en fleurs, was published that year. Eventually Moncrieff received a contract for the work, and his version, called Swann’s Way—after “swans weg,” a phrase from Beowulf, which Moncrieff had also translated—was published in 1922.
In the one letter Proust wrote to Moncrieff, he acknowledged the “remarkable way” in which the translation had been done, but gently criticized both the title Swann’s Way and the one Moncrieff had given the larger work, Remembrance of Things Past (“À la recherche du temps perdu ne veut nullement dire cela.”) The author died a month later.
In subsequent years, amid a whirl of his own writing, travel, a robust social life, periods of illness, a stint as an intelligence agent in Fascist Italy, not to mention completing numerous other translations by the likes of Stendhal and Pirandello, Moncrieff, in a feat that looks unlikely to be repeated, translated all but the seventh and last volume of Proust’s opus before his own premature death in 1930. Moncrieff’s work was twice revised—in 1981 by Terence Kilmartin, and 1992 by D.J. Enright—to correct mistakes and smooth out prose typically described as “flowery.” It remained the standard, indeed only, English-language Proust for most of a century.
Then, in 2002, the Penguin Proust appeared, bearing a more literal title, In Search of Lost Time, which D.J. Enright had used for the Modern Library publication. Each of the seven volumes was done by a different translator.
In a sour two-part review in the New York Review of Books (followed by a spicy exchange in the letters section!), noted Proustian André Aciman pulled no punches. Peppering his take with chilling words and phrases like parallel participials, clausula, anacoluthon, and Proustian hermeneutics, Aciman argued that a “translator’s aim should be to capture both the luster of Proust’s insights and the stylistic pitch of his sentences.” At this, in his view, the Penguin project had failed. Moncrieff “had crafted a language . . . [that] despite some noticeable shortcomings and inaccuracies, was able to convey the scope and sweep of Proust’s vision.” The new translation, on the other hand, was “the product of writers who are each translating a different Proust and whose operating principles could not be more different or ill-defined.”
Indeed, it is a bummer to me that Proust occupies such a rarefied, even elitist, realm in the culture—to the extent he is present at all—because he is far more accessible than most people realize, and a lot funnier.
The reader would have to take Aciman’s word for this, since he examined only the first two volumes, comparing numerous passages from various translations to the point of tedium. Lydia Davis, who translated Swann’s Way, in her too-slavish attention to Proust’s word order, had sacrificed “psychological nuances” for “semantic tact,” resulting in a “less resonant” cadence. Sure, she could “translate a sentence by Proust; but she still doesn’t get how it works.” Davis might be a dutiful and scrupulous translator, but “Wordlingo, the automated on-line translation service that is surprisingly accurate for a machine is dutiful and scrupulous too.”
James Grieve, translator of À l’ombre des jeunes filles en fleurs, fared worse. His title, In the Shadow of Young Girls in Flower, was “gobbledygook” that approached “monstrosity.” (For good measure, Aciman also trashed Penguin’s title for the final volume, Finding Time Again, calling it “thoroughly absurd.”) Whereas Davis had adhered too strictly to Proust’s word order, Grieve altered it irresponsibly, resulting in a style “more like a cross between Anthony Trollope and Nancy Mitford.” In an act of literary “legerdemain,” the translator had essentially rewritten Proust, spewing out mere prose. “And that’s not good enough.”
Something that was “good enough,” in Aciman’s estimation, was the novel’s old title, Remembrance of Things Past, a phrase plucked by Moncrieff from a Shakespeare sonnet. Changing it had been a mistake on D.J. Enright’s part, one of few, it would seem. Otherwise, per Aciman, Enright’s final reworking of Moncrieff’s classic translation brought readers “closest to the source.” The question, then, was “not which of the four available versions must a prospective reader purchase, but why a new translation in the first place. Why ever a new Proust?”
*
Is André Aciman right? Should Penguin have left “well-enough alone”? Does the Kilmartin/Enright version of In Search of Lost Time bring us closest to Proust in French? He sure sounds authoritative. But how is a general reader to know? Is there such thing as a “general reader” of a hundred-year-old, three thousand page French novel?
And what of the work of William Carter? The eminent Proust biographer has produced an annotated version of the novel, published by Yale University Press, that dials it all the way back to Moncrieff. The Kilmartin/Enright revisions “were not always felicitous or accurate,” Carter writes in his Swann’s Way introduction. Moncrieff “seems to have been more sensitive than Kilmartin or Enright to the cadence of Proust’s sentences,” and Carter’s “sole intention . . . was to bring [his] excellent translation closer than ever to the spirit and style of Proust’s original text.” So . . . this version is closest?
Chatting about Swann’s Way in a bookstore once, the clerk said something to me like, “the Lydia Davis translation’s supposed to be the best one”—which I have heard before. But according to whom? Maybe the same etheric source that led me, when I first started to read Russian literature in 2003, to the translations of Richard Pevear and Larissa Volokhonsky. I had some notion, based on who knew what, that, well, those were the best ones. A review blurb by James Wood told me their Anna Karenina was the work of “scrupulous translators” and would allow me to grasp Tolstoy “as perhaps never before.” (Can Wood read Russian? I still do not know.)
As for Proust, whom I also read for the first time that year, Richard Howard, in a bizarrely overwritten blurb, assured readers of Modern Library’s In Search of Lost Time (meaning Moncrieff/Kilmartin/Enright) that it “affords the surest sled over the ice fields as well as the most sinuous surfboard over the breakers of Proust’s prose . . .”
*
In January 2024, as part of a radical self-improvement program, also because I am obsessed with Proust, whom I was rereading, I started to learn French. In August, although I had recently finished the entirety of the Penguin translation, I returned to Swann’s Way, determined to read it in both French and English. It took eight months. I would read the French aloud, sometimes managing only a few pages, then the same section in the original Moncrieff translation. Struggling through pages of unbroken text, in the imperfect tense, constantly consulting a French dictionary, rereading passages two, three times, I often wondered: Am I getting anything out of this? Catching the rhythm and beauty of Proust’s famously long sentences?
I confess that I may not have been sledding and surfing. But I did learn things about the language, and to my delight began to catch small instances of Moncrieff’s famous padding. As the months wore on, I found myself thinking more deeply, almost in an existential way, about the process—the art—of literary translation. And the more I thought about it, the more mysterious it became.
When I was younger I believed—well, I don’t know what I believed, because it never occurred to me. But if you had asked me what translation was, I am sure I would have mumbled something about how it was just, like, swapping out one set of words for another. In high school I read Journey to the End of the Night by Louis-Ferdinand Céline after learning the Doors had nicked the title for one of their songs. There was a second name on the cover: Ralph Manheim. Who was he? No idea. Didn’t care.
As I was finishing Du Côté de chez Swann, brimming with pride and perplexity, I learned of a new translation of In Search of Lost Time. Oxford World’s Classics, whose very name conjures for me an image of the imprint’s white spines, was doing the novel, like Penguin, using multiple translators. Volume one, The Swann Way, translated by a Brian Nelson, had been out since 2023. Volume two, in a turn that felt like fate reaching out to me—it was the book that made me fall for Proust, in whose luscious atmosphere I wish I could live—had just been published. The title was In the Shadow of Girls in Blossom. The translator was Charlotte Mandell.
*
Why ever a new Proust? According to Brian Nelson, coeditor of the Oxford Proust and translator of its first and last volumes, the answer, relayed to me in an email, is straightforward: Oxford World’s Classics—obviously a publisher of great works—simply felt it should have its own Search. Will that satisfy André Aciman? Hard to know. Maybe he has mellowed in the twenty years since his NYRB piece.
In a more generous assessment of the Penguin Proust in the London Review of Books, published the same year, Michael Wood wrote, “It seems likely, and appropriate, that people will always refer to Scott Moncrieff when they speak seriously about reading Proust in English, but I’m not sure a new translation has to ‘better’ his version.” Although he concluded a beat later that he felt Penguin had. The new version “makes Proust stranger than he seems in the earlier version—that is, almost as strange as he seems in French.”
Indeed, it is a bummer to me that Proust occupies such a rarefied, even elitist, realm in the culture—to the extent he is present at all—because he is far more accessible than most people realize, and a lot funnier. Neurotic, sickly, gay, cork-lined room, guy tastes a bit of cake and thinks up an endless book filled with interminable sentences, etc.—but you almost never hear about how funny Proust is.
In a later essay in the anthology Proust in Context, Wood wrote, “The notion of fidelity has dogged translation since antiquity . . . But it has the immense drawback of presuming the existence of a stationary original text, always faithful to itself.” Rather, Wood said, “the original text of any work that matters changes over time. It creates new readings, which in turn alter the text itself . . . Every translation has its own mode of tuning and once we are beyond gross errors, we can come to our own conclusions about the music.”
*
Around the time I was learning to ask what the weather was like in French, Charlotte Mandell was named a Chevalier dans l’Ordre des Arts et des Lettres by the French government. “Translators are used to being invisible,” she said in her remarks at the ceremony—a sentiment she echoed when I was speaking with her over Zoom.
“That’s why translators are forgotten,” she told me. “They become invisible. We disappear behind the text.”
Yet as I was reading her In the Shadow of Girls in Blossom, she was constantly on my mind. It was my fourth time through the novel—but my first with a deeper awareness of it as a translated work. The result was a kind of doubling of the text. Marcel Proust, quite obviously, had written the book. But Charlotte Mandell had written the English words I was reading, whose power and beauty left me gasping and stunned.
When did it first occur to her to translate a text?
Mandell went to Boston Latin School, where students study the dead language. She could read and speak French, but at Boston Latin translated The Aeneid into English. “I just found my notebook in the garage,” she said. “It’s actually not bad. The fact that it was a dead language helped. Because you’re making something dead come alive. You can make it live in a text. So that’s when I first got the bug for translating.”
After she graduated from Bard, Pierre Joris recommended her to Stanford University Press for her first book-length translation, Maurice Blanchot’s The Work of Fire. Since then, she has translated a range of classic and contemporary authors including Balzac, Flaubert, Genet, Valéry, Céline, Jonathan Littell, Mathias Énard, and two earlier works by Proust: The Lemoine Affair, comprising his famous pastiches, and The Mysterious Correspondent, a collection of lost stories.
I told her À l’ombre des jeunes filles en fleurs was my favorite volume of the Search. She said hers was Le Temps retrouvé.
“The way everything comes together. All the characters and all of his past experiences coalesce. Everything falls into place. You feel like you’ve been living with these people for so long. All those incredible epiphanies that he has, which are amazing. There’s just no other book like it.”
Charlotte Mandell had written the English words I was reading, whose power and beauty left me gasping and stunned.
Listening to Mandell, I recalled favorite scenes from that novel and thought: Maybe it’s my favorite too. I felt out of my depth. I was speaking to someone who had TRANSLATED PROUST! Which, to me, was akin to pure magic.
Mandell had read Moncrieff’s translation of the Search first.
“It’s really beautifully written, but there are a lot of mistakes. And he also tended to add things that weren’t there. As a translator, that’s not something that I would do. But the English is beautiful. A lot of people equate Moncrieff with Proust, but it’s very different. Moncrieff tends to overwrite a lot, and Proust doesn’t do that. He’s actually very succinct. People think that he goes on and on, but he doesn’t really. The sentences can be very long, but there’s a reason that they’re long. That’s one of the things that surprised me when I translated him—how much he got into a sentence, how much he managed to pack in.”
Moncrieff’s method, according to Chasing Lost Time, a biography by his great-great-niece Jean Findlay, “was to read a passage silently, think on it, and scribble a version in English in a jotter which he would then test by reading it aloud to a discerning ear.” Later he would have friends read the French aloud as he wrote his translations down, then read over it himself. Once he had a rhythm he liked, he would type it.
Mandell’s method for In the Shadow of Girls in Blossom involved four texts: two copies of the French that she was translating (the 1987-1989 Pléiade edition)—one on a split-screen on her computer alongside her working document, the other a hard copy—and the Moncrieff and James Grieve translations. She would read a line, translate it, check it against Moncrieff and Grieve. Grieve wanted to make Proust more idiomatic and sound up-to-date. Moncrieff was the opposite, old-worldly, not of this time. But if one or the other had a nice word or turn of phrase—more often that would be Moncrieff—she might use it.
The work was . . . slow.
“I should say that another one of my rules is that I don’t read ahead. So I’m reading the book as I translate it, which a lot of translators don’t do. Obviously I had read this book before, but in this case I was taking it sentence by sentence. And it took a long time because I had to compare, and then rework my own translation, line by line. I would only do two or three pages a day. You have a responsibility. This is such great literature that you don’t want to make a mistake.”
As with all of her books, she would read the day’s—or rather the night’s work (Mandell works through the evenings, finishing in the wee hours) aloud to her husband, the poet Robert Kelly, who would offer suggestions if something sounded off. “He calls it his bedtime story. It’s usually like two in the morning,” she laughed.
To Mandell, the so-called seams are an important aspect of Proust’s style.
Moncrieff had struggled with the book’s title, believing a literal translation would make British readers think of menstruation. He enlisted the help of no less than Joseph Conrad, who suggested “In the Shade of Young Girls in Bloom” and “In the Shade of Blossoming Youth.” Finally Moncrieff chose Within a Budding Grove, from a poem by William Allingham.
“Which loses everything, really—I mean, it loses the girls. You sort of have to have the girls in the title,” Mandell said, referring to la petite bande, the group of young women—including Albertine, a key figure in the Search—the Narrator meets in Balbec, the seaside town where much of the novel takes place.
In her Translator’s Note, Mandell stresses that, while their ages are never made clear (nor, for that matter, is the Narrator’s: he goes from playing games in the Champs-Elysées to visiting a brothel with no temporal signposts) they are still very much girls, roughly in their mid-teens. In Moncrieff’s fusty presentation, she told me, “they sound like old men. That’s one of the things I wanted to do also: get the girls to sound like girls.”
Also in her note, she brings up something James Grieve wrote in his Introduction, about Proust’s prose having “rough” seams. That description had always stuck with me. I am such an impressionable reader—and so easily cowed by scholarly argument—that I assumed it must be true without quite knowing what it meant. To Mandell, the so-called seams are an important aspect of Proust’s style.
“When Grieve says seams,” she said, “I think he means that Proust might switch tenses in an odd way. He might switch from imperfect to passé composé. Sometimes it can be jarring . . . But when you really look at it, you’ll see that there’s a reason why Proust is using one or the other. I guess some people might call that a seam. But it’s important to keep that. No one had written like that before. So you don’t want to make it sound normal. It wasn’t normal.”
And she made a suggestion that threw me.
“You should try translating something. Pick a passage that you really like and then look at what Grieve did. That’s really the way to see how someone thinks. Just do it yourself. And don’t cheat,” she laughed. “Don’t think about whether it’s good or not, just do it.”
I immediately demurred. My French is poor. Mandell insisted I give it a shot. In fact, there was an apt precedent: Marcel Proust himself spent six years translating the work of English art critic John Ruskin, despite being unable to speak or read English! He had access to something greater than a translation app: his mother. Jeanne Proust, so devoted to her son, wrote the first drafts of his Ruskin translations, which he would heavily revise and annotate, then send to a friend to polish, only to alter them yet again. According to Jean-Yves Tadié’s Marcel Proust, the writer “felt that translation was a marvellous school for style.” In constantly reworking his drafts, “[t]he structure of Ruskin’s sentences, which were long, rich in incident and imagery, supple and musical . . . impregnated his style, which since Jean Santeuil [Proust’s abandoned first try at a novel], had been groping around in search of a model.”
Fumbling for a way to wrap up, I asked, “What do readers want out of a translation? Do you think they think about it at all?”
She was shaking her head before I had even finished. “I think people are starting to pay more attention to translators. But you can see in reviews, critics will review a book as if it was written in English. They think that translation is some kind of secretarial job, like you have a book and just type it up in English.” Translators are forgotten, we disappear behind the text. “I think it’s good that readers are starting to realize there are people involved. It’s not AI.”
Speaking of—does she—do others in her profession—sense a looming threat?
Mandell said, “Not for me, because I do literary novels. But for people who translate thrillers and mass-market books, I think there is a real threat. AI translation has gotten very good. It can do a mass-market paperback pretty well—not as well as a human, but plausibly. And a lot of people might not even notice that it wasn’t translated by a person. The fact that that’s possible is kind of sad. But I’m not worried, because, I mean, no AI thing will ever be able to translate Proust.”
*
Surely not. But could I? I flipped through À l’ombre des jeunes filles en fleurs looking for a sentence that was relatively short, whose context and meaning I did not readily grasp, but that was also at least somewhat Proustian structurally. I chose the following: “Il n’est peut-être rien qui donne plus l’impression de la réalité de ce qui nous est extérieur, que le changement de la position, par rapport à nous, d’une personne même insignifiante, avant que nous l’ayons connue, et après.” (As you will see, I made at least one howler: because I did not find that spelling of connue among the conjugations of connaître [to know] in my verbs book, I used the word in a way I have heard my online French teacher use it. In my defense, I translated it both ways before checking!)
“There is perhaps nothing that gives us so strong an impression of the reality of the external world as the difference in the position, relative to ourselves, of even a quite unimportant person before we have met him and after.
–C.K. Scott Moncrieff (also CKSM/Kilmartin/Enright)
“To give us an impression of the realness of people and things external to us, even if they are insignificant, there are few comparisons more instructive than the change their disposition toward ourselves undergoes between the time before we know them and the time after.”
–James Grieve
“There is perhaps nothing that makes us see the reality of the external world more clearly than a change of location, relative to ourselves, of a person, however insignificant, before we knew that person, and after.”
–Charlotte Mandell
“There is perhaps nothing that gives a greater impression of our exterior reality, than a change of position, in relation to us, of even an insignificant person, before we were well-known, and after.”
–B.A. Charles"
BRYAN ALISTAIR CHARLES
OCTOBER 7, 2025
https://lithub.com/on-translating-proust-and-the-art-of-not-reading-ahead/
#Metaglossia
#metaglossia_mundus
The symposium responds to a new urgency to Ukrainian culture of the past, present and future both inside and outside the country.
"Writers, translators consider ‘Ukraine in Translation’ Oct. 18
By Kate Blackwood
October 7, 2025
A symposium, “Ukraine in Translation,” will explore writing and translation and cultural production and preservation in the context of Ukraine on Oct. 18 from 1-6 p.m. in the A.D. White House.
During a roundtable discussion, Oksana Maksymchuk, a Ukrainian-American poet and scholar, and Ainsley Morse and Sabrina Jaszi, both scholars and literary translators, will discuss the movement between literary practices and languages and how translation relates to political transition and rupture. They will also consider, as Ukraine continues to defend itself from Russian invasion, how war, colonization and decolonization transform culture and our ways of reading and listening.
“Russia’s full-scale invasion turned the whole world’s eyes to Ukraine and brought an unprecedented wave of translations of Ukrainian literature. English speakers can finally sample the rich variety of Ukraine’s literary movements and traditions, and there is a new urgency to Ukrainian culture of the past, present and future both inside and outside the country,” said Sophie Pinkham, professor of the practice in the Department of Comparative Literature, who organized the event. “This makes it the ideal time for a deeper exploration of Ukraine in translation.”
The roundtable will be followed by poet, playwright, and translator Sasha Dugdale’s reading of her translations from the work of acclaimed playwright Natalya Vorozhbit and a poetry reading by Maksymchuk. The program will conclude with a concert of Ukrainian folksongs by the acapella trio Zozulka.
Read the full story on the College of Arts and Sciences website."
https://news.cornell.edu/stories/2025/10/writers-translators-consider-ukraine-translation-oct-18
#Metaglossia
#metaglossia_mundus
"...The Times’ decision to dump their copy desk ignored the essential work copy editors do, instead treating them merely as disposable obstructions.
At The Daily Californian, copy editors make sure “unhoused” is used instead of “homeless,” check that “professor” is lowercase before a name while “U.S. President” isn’t and ensure there isn’t a comma before that preceding “and.”
Why does any of this matter? Well, would you trust a newspaper’s White House scoop if it doesn’t even know where to put a comma?
Perhaps more importantly, we also serve as the final line of defense against libel, ensuring the Daily Cal doesn’t publish anything untruthful that damages someone’s reputation.
It’s true that social media sites don’t have to care about these standards. Under federal law, they aren’t liable for what’s published on their sites.
Established news sources were supposed to be above the social media fray. When Mark Zuckerberg told his employees to “move fast and break things,” the editors of The New York Times weren’t supposed to take inspiration.
...The Times’ disregard for its quality and standards have thrusted it straight into the digital scrum.
Readers certainly noticed. The ensuing grammatical and stylistic errors spawned the X account “Typos of the New York Times,” which now has nearly 40,000 followers. “As a hobby, I correct typos in the Times, which no longer employs copy editors and consequently has tons of typos,” its bio reads.
The lowering of standards for traditional media couldn’t have come at a worse time. Now, with the rampant spread of misinformation on social media and the development of artificial intelligence that can produce lifelike videos, people don’t know where to turn for trustworthy news. But they certainly aren’t going to the legacy news media.
A Gallup poll last year found that a record low 31% of Americans had a “great deal or fair amount” of trust in mass media. Among those aged 18 to 29, that number was just 26%.
There are myriad reasons behind this low level of trust, from partisan polarization — just 12% of Republican respondents trusted mass media, compared to 54% of Democrats — to a greater trust in social media for its broader reach and perspectives. However, one thing is clear: Repairing trust in the news media is essential for our democracy.
Now, with most Americans neither trusting the news media nor social media, there is no objective truth to base political debates on. This has poisoned our democracy, creating dangerous polarization and allowing politicians to manipulate the truth at their own pleasure.
There’s no single solution for this crisis of trust, but fixing it may start by distinguishing news media from social media through a renewed commitment to quality and standards. And that fix starts with the copy editor."
https://www.dailycal.org/blogs/grammar/a-democracy-needs-copy-editors/article_61971953-0a65-4523-9a97-ae4bdbe71377.html
#Metaglossia
#metaglossia_mundus
Rebecca Gayle Howell, associate professor of poetry and translation in the U of A's M.F.A. Program in Creative Writing and Translation, has been awarded the 2025 Aiken Taylor Award for Modern American Poetry.
"English Professor of Poetry and Translation Wins Prestigious Literary Award
Oct. 06, 2025
Carey Neal Gough
Professor Rebecca Gayle Howell, associate professor of poetry and translation in the Department of English's M.F.A. Program in Creative Writing and Translation, has been awarded the 2025 Aiken Taylor Award for Modern American Poetry.
The award has been presented since 1987 by The Sewanee Review literary magazine. Their editors note the Aiken Taylor Award "has honored a distinguished poet in the maturity of their career. … Established by the physician and poet K.P.A. Taylor in honor of his elder brother, the modernist poet and story writer Conrad Aiken, the Aiken Taylor Award has celebrated poets such as Howard Nemerov, Gwendolyn Brooks, Wendell Berry, Louise Glück and Billy Collins."
"This award, one of many in an accruing list of notable honors, is a testament to Rebecca's prowess as a talented and multifaceted poet and translator. We are fortunate to have her as a peer and teacher in the English Department," notes Dr. Yajaira Padilla, chair of the Department of English.
Howell, who joined the faculty of the M.F.A. Program in Creative Writing and Translation in 2022, is a writer, literary translator, librettist and editor. Her books include two novels in verse — Render / An Apocalypse and American Purgatory — both of which were named Bestsellers of the Decade by Small Press Distribution. She translated Patagonia poet Claudia Prado's El Interior de la Ballena and Amal al-Jubouri's Hagar Before the Occupation / Hagar After the Occupation. Howell's work has received critical acclaim from such outlets as the Los Angeles Times, Poetry London, Publisher's Weekly and the Kenyon Review. In addition to her work at the U of A, she also serves on faculty for the University of the South's Sewanee School of Letters. From 2014 to 2024, she was the poetry editor for the Oxford American. Her sixth book, Erase Genesis, will be released this November by Project Poëtica/Bridwell Press.
"To see Rebecca's name among this list of luminary poets seems just right. She's too humble to say this award is well earned, but it is. Everyone in the program is so proud of this accomplishment, this recognition of her important work," said Toni Jensen, director of the M.F.A. Program in Creative Writing and Translation.
About the U of A M.F.A. Program in Creative Writing and Translation: For over 50 years, the University of Arkansas Program in Creative Writing and Translation has served as a preeminent training ground for some of the nation's best writers. Established in 1966, ours is one of the oldest M.F.A. programs in the nation and one of the most innovative, offering degree tracks in poetry, fiction, and literary translation."
https://news.uark.edu/articles/80147/english-professor-of-poetry-and-translation-wins-prestigious-literary-award
#Metaglossia
#metaglossia_mundus
A union is needed to ensure that this vital communications tool retains quality interpreters and better serves the Deaf community.
"Rochester has one of the largest per capita Deaf communities in the world. ASL interpreter education programs at the Rochester Institute of Technology and a Video Relay Service (VRS) call center are among essential resources.
VRS provides the most hours of interpreting services to Deaf communities nationwide, ensuring that the Deaf, Hard of Hearing, and Deaf-Blind have equal access to telecommunications services. Unfortunately, the challenges of VRS have not been addressed adequately by the companies and the FCC—and I say this as someone with 18 years of experience in working in VRS. That’s why I’m joining with my colleagues in collective organizing with the Office and Professionals Employees International Union (OPEIU). We are working to organize VRS interpreters across the United States and Puerto Rico to ensure that this vital communications tool retains quality interpreters and better serves the Deaf community.
During my years of working in VRS, I have seen an increase in concern over the deterioration in the quality of service offered. Research has explored a lack of addressing those concerns. OPEIU organizing at Sorenson and ZVRS/Purple, the two largest employers, seeks to address interpreters training and supervision to hone their ethical decision making and emotional psychological resilience. Apparently the corporate model of VRS companies renders a primary concern with the bottom line, which has from my perspective been at the expense for the well-being of interpreters and impacting VRS users.
Most VRS calls go through Z/P and Sorenson. Z/P is now owned by sprawling French multinational Teleperformance (with its own track record of misconduct), and Sorenson is owned by Ariel Investments, a Chicago-based investment firm. While it does not own a stake directly in Sorenson, the New York State Teachers Retirement System has over $340 million invested with Ariel—meaning that they should play an active role in helping us win our union. We are seeking a card check neutrality agreement, which will allow workers to organize free of management interference. This is especially important as President Trump works to dismantle the National Labor Relations Board.
I received my bachelor’s in American Sign Language Interpreter Education from RIT. My interest in supporting the development of the field led me to get a master’s in interpreting studies from Western Oregon University.
In my studies I learned that the profession has a long history of attrition due to burnout. Research-based evidence has continued to identify VRS as the setting most impacted by attrition, hurting interpreters who have a passion for the field. Initially companies hired highly experienced and skilled providers but less experienced interpreters, recent graduates and students are recruited to meet demand.
The responsibility of interpreting trauma-based calls, including but not limited to 911 calls, is one explicit example contributing to attrition. From my own experience I feel that VRS companies do not prepare interpreters for trauma-based calls. A company-hired professional to provide employee assistance assessed the situation as “unconscionable.”
Rochester can play a critical role in organizing VRS to make it better—we need as many interpreters as possible at Sorenson and Z/P to join our efforts, and we need the Deaf community to be outspoken about concerns they have with VRS. If you are an interpreter please take our survey here, and if you are a member of the Deaf community, please reach out to info@aslunion.org. I hope that by organizing to address these issues positive change can happen.
Kathleen C. Holcombe
Signed Language Interpreter
By LETTERS TO THE BEACON | September 10, 2025
https://rochesterbeacon.com/2025/09/10/vrs-interpreters-need-union-representation/
#Metaglossia
#metaglossia_mundus
"Google's powerful search tool is now available in more languages
Following its official introduction earlier this year, AI Mode in Google Search has evolved a great deal and picked up some key features over the last few updates. It's pretty clear at this point that Google wants AI Mode to be a prominent force in Search, and the company is moving forward with that ambition by adding support for five new languages.
Google announced in a blog post that AI Mode now supports five additional languages, namely Hindi, Indonesian, Japanese, Korean, and Brazilian Portuguese, with the changes also reflecting on the AI Mode support page. This takes the total number of supported languages to six (including English).
While it's a decent set of languages that covers millions of users around the world, several other languages are still missing from the list. However, it shouldn't take long for Google to remedy that in subsequent AI Mode updates...
Since Google began focusing more on AI Mode and its functionalities, there was a feeling that the company might make this the default search experience, thereby replacing Google Search. This notion gained a lot more traction over the weekend as Google's Logan Kilpatrick appeared to partially confirm that AI Mode could be the default "soon" (via Search Engine Land).
However, in response to Search Engine Land's reporting, Google VP Robby Stein suggested that it won't be the case, adding that the team is "focusing on making it easy to access AI Mode for those who want it."
This exchange occurred after Google launched the updated AI Mode URL last week (google.com/ai), offering easier access to the powerful tool. This was a minor revision, though, as AI Mode was previously available via the google.com/aimode URL.
While this episode doesn't rule out the possibility of AI Mode being the default search product in the future, it's clear there are no immediate plans for the AI-powered search experience to replace the classic Google Search. Perhaps Google can devise a way to make AI Mode the default for those who need it, while offering the traditional search experience for everyone else. But for the time being, both AI Mode and the older search experience will continue to coexist."
By Chethan Rao
https://www.androidpolice.com/google-search-ai-mode-new-languages-added/
#Metaglossia
#metaglossia_mundus
"En Veracruz se requieren intérpretes de las 15 lenguas indígenas y 31 variantes en procesos judiciales o atención médica.
Miércoles 10 de Septiembre de 2025
Para Magdalena Hernández
Hernández, titular de la Oficina de la Representación del Instituto Nacional de los Pueblos Indígenas en Veracruz (INPI), en Veracruz se requieren intérpretes de las lenguas y sus variantes que se hablan en el Estado, sobre todo en los procesos judiciales o atención médica, para que exista una buena comunicación. La funcionaria precisó que en Veracruz se hablan 15 lenguas y 31 variantes, y apenas se cuenta con un registro de 39 intérpretes que colaboran con la Fiscalía General y el Poder Judicial del Estado en los juicios, así como con la Secretaría de Salud cuando las personas reciben atención médica. Sin embargo, destacó que hacen falta muchos más para poder dar cobertura total a las necesidades que se presentan todos los días. "En el Padrón Nacional de Traductores e Intérpretes hasta el año pasado nosotros aquí en Veracruz, sólo en Veracruz, teníamos 39 intérpretes de diferentes lenguas. Lo justo es que hubiera un intérprete y un traductor para cada una de las variantes o, por lo menos de los distritos judiciales que estuvieran adscritos los intérpretes y traductores que hablan las lenguas de esos lugares", aseguró. De igual forma, destacó la necesidad de que existan defensores que conozcan la lengua y la cultura de las comunidades indígenas. Refirió que el Poder Judicial de Veracruz tiene un Padrón de Traductores e Intérpretes mismo que está inscrito en el Registro Estatal de Peritos. Concluyó que los intérpretes y traductores de los que habla son del INPI no del Poder Judicial del Estado o de la Fiscalía General del Estado."
viernes 12 de septiembre de 2025
Para Magdalena Hernández
https://www.gobernantes.com/vernota.php?id=491491
#Metaglossia
#metaglossia_mundus
"After 16 years of English-only support, the writing assistant now corrects grammar, rewrites paragraphs, and translates across Spanish, French, German, Italian, and Portuguese.
Grammarly has launched spelling and grammar correction in 5 different languages and in-line translation across 19 different languages. The writing assistant that's helped polish English prose for 16 years now supports Spanish, French, Portuguese, German, and Italian. It brings the signature red underlines and one-click rewrites to languages spoken by over a billion people globally. Ailian Gan, Director of Product Management at Grammarly, says,
"Our customers have been asking for multilingual support, and we're meeting them where they are, not just in the 500,000 apps and websites where Grammarly already works, but also in the languages that they think, learn, and communicate in daily."
Grammarly mentioned in its official blog post that nearly one in five Americans speaks multiple languages. Moreover, Grammarly's 40 million daily active users regularly switch between languages throughout their workday. Until now, that meant constant platform-hopping and second-guessing whether that email to Barcelona colleagues struck the right tone.
Luke Behnke, Grammarly's VP of enterprise product, notes that some beta users discovered the new features by accident. "The rollout took some users by surprise when the familiar red lines suddenly appeared under text written in their native language," he said to The Verge. That unexpected delight speaks volumes about how long users have been waiting for this moment.
The beta release includes three key capabilities across the five new languages:
Grammar and Spelling Corrections: Those trusted red underlines now appear whether you're writing in German or Italian, catching everything from basic typos to complex grammatical errors.
Paragraph-Level Rewrites: One click transforms casual language into the same six different languages... for professional prose, maintaining the nuances that make native speakers sound, well, native.
Real-Time Translation: The platform now translates across 19 languages directly within your workflow. It needs no more copy-pasting between Google Translate tabs.
Grammarly's expansion continues with Spanish users getting advanced clarity suggestions later this year, followed by other languages in 2026. The company, fresh off acquiring productivity startup Coda and launching new AI features, plans to integrate multilingual support directly into its AI-native writing surface.
This is a technical progress and an authentic connection across linguistic boundaries. After 16 years of English-only support, Grammarly's multilingual debut feels like a long-overdue reunion. For millions who think in one language but write in another, the message is clear: your writing assistant finally speaks your language."
Arpit Dubey
September 11, 2025
https://www.mobileappdaily.com/news/grammarly-expands-5-languages
#Metaglossia
#metaglossia_mundus
"Reflecting on Europeana’s summer translation internships
This summer, the Europeana Foundation was delighted to welcome four students from four European countries to complete internships on translation. Each student worked on translations for Europeana.eu, in total translating 65 stories of European heritage. They reflect on their experiences and what they learned.
Gábor Juhász: new insights into translation practice
My time at Europeana was one that I will never forget. It was not only a remote translation internship, but a glimpse into what my professional life might possibly look like - and that, for me, was an invaluable experience.
Next to my work translating editorial into Hungarian, during my internship I also evaluated and validated multilingual search results to support SEO. Through this work,, I realized that I was not just a translator at Europeana, but a Hungarian language expert. My skillset being so valuable allowed me to take on different tasks other than only translation, and it was an experience I am quite sure I could not get anywhere else. Despite being only a graduate student, I was treated with respect and dignity in this great amalgamation of various languages and cultures.
I also learnt that the best way for me, personally, to translate is to do blocks of four to five hours at a time, when I can truly immerse myself in the work and achieve deep focus. This was made possible by the flexible working hours provided, another benefit of this type of internship. Furthermore, I finalised my translation workflow: just write! You can always perfect text later. I also learned that if a segment proves to be too complicated, I could always skip it and come back to it later after the rest of the text is done. The same goes for editing my work: I will need the text to ‘rest’, so that when I come back to it I can take a look at it with a fresh mind.
My overall experience was extremely positive, and I am very grateful to have been able to work at Europeana, even if it was for a short period of time.
Katerina Bizirtsaki: enhancing editorial for Greek audiences
My internship was centered on translating Europeana's English articles into Greek. The themes of the texts widely varied, from architectural wonders and botanical research to literature and food myths. What made the experience even more special was the freedom to select my translation projects according to what I thought would be relevant and interesting to the Greek audience.
The primary challenge I faced was preserving the writer's original tone while ensuring the translation would feel natural and captivating to a Greek reader. Throughout this month-long internship, I not only gained valuable experience translating different themes but also explored the richness of the European cultural heritage and learned many interesting facts that I probably wouldn’t have encountered otherwise.
One of the most rewarding parts of this internship was experiencing firsthand Europeana's multilingual and multicultural work culture. Working from Greece, I felt completely connected to a team spread all over the globe and found out how harmoniously a remote system can run. Everyone was so kind and welcoming that I immediately felt comfortable exploring, asking questions, and being creative with my translations. I am especially grateful to the Audience Engagement team, who truly made us interns feel like a part of Europeana for however short that time was. By inviting us to their weekly meetings and always explaining how things worked, I felt like I was able to delve deeper into the complexity of Europeana's vision, far beyond my specific task of translation.
It was truly an honour to offer my services as a translator to make more of the collections made available through Europeana.eu accessible to a Greek audience and to experience such a supportive work environment. This internship has not only enhanced my translation skills but also taught me how to function in a remote and multilingual workplace.
Jessica Kuster: refining translation skills
During my internship at Europeana, I had the opportunity to work on a wide variety of translation tasks (from English to German) and I can honestly say that I enjoyed every moment of it. What made this experience especially rewarding was the sheer diversity of the texts I worked on. The topics ranged from historical events and cultural heritage to more contemporary themes. This variety not only kept my work fresh and engaging but also significantly expanded my historical, cultural, and general knowledge.
One of the main professional benefits of the internship was the chance to refine my translation skills. Beyond simply transferring text from one language to another, I learned to adapt content for specific audiences, making sure the tone and style matched the needs of the readers. I also had the chance to improve my proofreading abilities, which required a keen eye for detail and a focus on linguistic accuracy and consistency.
Another exciting aspect of my work was using the content platform Contentful. Before this internship, I had never worked with this tool, but I enjoyed figuring out how it worked, and I quickly saw the advantages it offered. One major benefit was being able to see my translations directly on the website once they were published. This gave me a strong sense of accomplishment and made the process even more satisfying.
A highlight of my time at Europeana was working with such an international team. Even though the internship was fully remote, I always felt part of the group. My colleagues came from various cultural and professional backgrounds, and this exchange of knowledge and perspectives added an extra layer of depth to the experience. I was also very grateful for the support I received from my supervisors.
Mario Pueblas Madueño: the nuances of translation
My main task during this internship was to translate articles from English into Spanish. While I have experimented with different types of translation during my career, the translation of art and culture was one that I was unfamiliar with. I am grateful for the opportunity to hone my translation skills and to get a glimpse of what a corporate job can be like.
The main challenge for me was adapting to all the different topics each article discussed. I translated articles about all sorts of things: history, fashion, football, literature - the list goes on. Due to this variety, I often needed to investigate a topic further before starting the translation because of my lack of prior knowledge or in order to ‘get in the zone’ after translating something completely different in theme and tone.
I also tried to keep each author’s style present in my translations, and some articles were much more difficult in that regard because of the way they were originally written. The longest translation I completed during the internship, ‘A queer tour,’ was especially challenging. It is full of metaphors, historical references, and expressions that don’t translate well into Spanish directly, so I needed more time than usual to translate and to let the text ‘marinate’ in order to ensure that the Spanish version was a pleasant read.
One of the aspects I enjoyed the most outside of translation was the multicultural work environment and how willing to help people were. Everyone comes from different professional and personal backgrounds, and I believe that added to the dynamic nature of the whole team. The weekly team meetings allowed me to see how I was contributing to Europeana’s goals and helped me feel a valuable part of the foundation. I would especially like to thank Adrian Murphy, my supervisor."
Adrian Murphy
Europeana Collections Manager , Europeana Foundation
https://pro.europeana.eu/post/reflecting-on-europeana-s-summer-translation-internships
Un nouvel ensemble de données contenant des langues africaines devrait améliorer l'accès à l'IA pour des millions de personnes sur le continent.
"Bien que l'Afrique abrite une grande partie des langues du monde (bien plus d'un quart selon certaines estimations), beaucoup d'entre elles sont absentes du développement de l'IA.
Cela s'explique à la fois par un manque d'investissement et par le manque de données facilement accessibles.
La plupart des outils d'IA utilisés aujourd'hui, tels que Chat GPT, sont entraînés en anglais ainsi que dans d'autres langues européennes et chinoises.
Ces langues disposent d'une grande quantité de textes en ligne sur lesquels s'appuyer.
L'IA en Afrique : les experts veulent combler le fossé linguistique - BBC News Afrique"
Author,Pumza Fihlani
Role,BBC News Johannesburg
8 septembre 2025 https://share.google/KoWg1Xa8VPJUqS7qa
#Metaglossia
#metaglossia_mundus
"The words we use to talk about nature are disappearing. Here’s why that matters.
We’ll need to do more than "touch grass" to revive them.
Kate Yoder
Senior Staff Writer
Sep 05, 2025
Once upon a time, the English language was full of stories with “blossoms,” “rivers,” and “moss.” But these words are disappearing from our vocabularies — and along with them, our connection to the natural world they describe.
A study published in the journal Earth earlier this summer found that the use of nature-related words declined more than 60 percent between 1800 and 2019. The study’s author, Miles Richardson, a psychology professor at the University of Derby in the United Kingdom, looked at 28 everyday terms related to nature, including “bud,” “meadow,” and “beak,” using a Google database that tracked the frequency of words in English-language books over time.
“These words reflect what people noticed, valued, and wrote about,” Richardson wrote in a blog post.
As part of the same study, Richardson developed a computer model to capture how people had lost touch with nature over time. The simulation played out across generations as cities grew and green space disappeared. When he compared the model’s projections to the nature-word data, he found that the two graphs matched extremely closely, with less than 5 percent error between them.
Nature is disappearing from our language and our lives
Since 1800, there’s been a sharp decline in nature-related words in English language books. It closely matches a simulation of nature–human interactions.
...
Experts have been raising the alarm over our growing disconnect to nature for decades, often by pointing to how our language has changed. In its 2007 edition, the Oxford Junior Dictionary, widely used in classrooms in the United Kingdom, removed dozens of entries related to the natural world, including “acorn,” “bluebell,” and “magpie,” to make room among its 10,000 entries for modern inventions like “blog,” “chatroom,” and “MP3 player.” The decision eventually drew sharp criticism from a group of authors led by Margaret Atwood. More than 200,000 people went on to sign a petition for the dictionary to reinstate the nature words that had gone missing.
But the editors didn’t budge, since a dictionary’s purpose is to describe language as people use it, not as we wish they did. Older dictionaries had lots of flower words because children lived in semirural environments, but that wasn’t the case anymore, the head of children’s dictionaries at Oxford University Press explained when the words first disappeared. But the anxiety over the loss of nature language points to a bigger question, once posed by the naturalist Robert Michael Pyle: “What happens to a species that loses touch with its habitat?”
Experts say that detachment from nature is at the root of many of the environmental problems the world faces today, from biodiversity loss to climate change. “We’ve put a lot of effort into treating the symptoms of environmental crisis, rather than looking at the root causes,” Richardson said. Just as you’re more likely to help a friend than a stranger, you’re more likely to care for nature if you have a relationship with it, he said.
This disconnection hurts people, too. Spending time in nature is good for your physical and mental health, creating opportunities for exercise as well as lowering stress levels, making our brains more creative, and even improving sleep. “There is overwhelming data that nature is good for us,” said Pelin Kesebir, a social psychologist who’s a fellow at the Center for Healthy Minds at the University of Wisconsin-Madison. There’s also research showing that spending time in nature as a kid tends to translate into environmental concern later in life.
To understand how our relationship with the natural world has shifted, many researchers have turned to studying the media we produce. In 2017, Kesebir and her sister identified a steady drop in references to nature in song lyrics, fiction books, and movie storylines since the 1950s. Another study found a marked decline in references in books to 134 common species names such as “bee,” “goose,” and “honeysuckle” over the course of the 20th century. Across children’s picture books and Disney films, portrayals of natural settings have been on the decline.
“When nature disappears from culture, then all these opportunities to evoke appreciation for nature, respect for nature, interest in nature — those disappear,” Kesebir said.
It’s not just we’re talking about nature less; the feelings behind those words have also changed, according to Robert Poole, a professor of English at the University of Alabama. “I want to know not how many times we say it,” he said, “but when we do invoke it, are we saying it’s beautiful, distant, deadly, savage?”
Poole has studied how the way Americans write about trees and forests has changed over the past 200 years. Nowadays, “we’re just less likely to use majestic words — ‘lofty,’ ‘stately,’ ‘noble’” — to describe forests, he said. That could be due to people spending less time around trees, or perhaps forests just aren’t as healthy or grand as they used to be.
As our language around forests became less reverent, Poole noted that people began using more scientific and economic terms to describe trees. In other words, people began viewing forests as something from which to extract value, not inspiration.
Richardson’s research found that the decrease in the use of nature words became particularly pronounced after 1850, around the time that industrialization and urbanization grew rapidly. When people move closer to cities, where concrete has covered over forests and meadows, it becomes harder to access green spaces. The other side of the coin is industrialization, where nature gets stripped for parts: forests into timber, meadows into farms. The United Kingdom is one of the most nature-depleted countries in Europe. As the British Empire expanded, so did its industrial model. Research has shown that places such as Australia, the United States, and Hong Kong, all shaped by British colonization, have some of the lowest levels of connectedness to nature today.
Technology may also be a factor: Kesebir’s research tied dwindling references to nature to the spread of the TV and other entertainment on screens. Whereas previous generations may have spent their leisure time playing outside, we now spend much of our free time head-down and phones up, playing video games or scrolling TikTok.
The solution isn’t as easy as encouraging people to walk outside and “touch grass,” though that small step is still good for your mental health (no, really, researchers have studied it.) Even efforts to plant more trees and expand parks, while helpful, probably won’t be enough on their own. Richardson’s study found that the most important factor in predicting what happens next is the attitude that parents pass down to their children. As one generation loses its connection with nature, their children begin life with lower levels of connection, a self-perpetuating cycle.
Jackie Morris, a British illustrator and author, has seen that problem firsthand. Inspired by the controversy with the Oxford Junior Dictionary, she had the idea to write a children’s book highlighting the missing words. The Lost Words, illustrated by Morris and authored by the nature writer Robert Macfarlane, turned into a bestseller and “cultural phenomenon” after its release in 2017. With it, Morris hoped to re-enchant kids with the plants, birds, and critters that had fallen out of their vocabularies. When the book first came out, Morris recalls, TV crews went into schools and asked kids to identify the names of these living things by their pictures. “What I said was, ‘Well, you should have taken them ’round your own office, really, because the reason kids don’t know is because the parents don’t know,’” Morris said.
For Morris, addressing our disconnect with nature starts with what she calls “rewilding our imagination.” She remembers, as a child, recognizing birds for the first time, seeing the bright light in their eyes and desperately wanting wings herself.
“Watching birds was just such a joy to me when I was young,” she said. “And it shocks me that there are many people who just don’t see them.”
Grist is the only award-winning newsroom focused on exploring equitable solutions to climate change. It’s vital reporting made entirely possible by loyal readers like you. At Grist, we don’t believe in paywalls. Instead, we rely on our readers to pitch in what they can so that we can continue bringing you our solution-based climate news."
https://grist.org/language/nature-word-language-disappear-culture/
#Metaglossia
#metaglossia_mundus
"TOWARD AN INCLUSIVE APPROACH TO LINGUA-CULTURALLY DIVERSE STUDENTS’ LANGUAGE IN SYSTEMATIC FUNCTIONAL LINGUISTICS RESEARCH
Systemic Functional Linguistics (SFL) fundamentally embraces inclusivity, as it legitimizes different language varieties as contextualized, socially contingent categories along a horizontal continuum. It also conceptualizes grammar as a set of linguistic resources for meaning-making, rather than as rigid rules that dictate grammatical “correctness.” However, much SFL-based research on lingua-culturally diverse students’ language use has tended to rely on monolingual native speaker norms. This tendency often reinforces deficit discourses by portraying students as linguistically deficient or lacking, clashing with the increasingly recognized perspective of translanguaging, which affirms the legitimacy of students’ language practices on their own terms. This article critically examines how much of SFL-based research on students’ language use reflects a monolingual orientation and then explores how insights from translanguaging can inform more inclusive approaches to studying student language within SFL scholarship. The article argues that moving away from external monolingual standards in analyzing and assessing students’ language use, and shifting the analytical focus from formal “correctness” to meaning-making, can help re-center SFL’s original commitment to inclusivity. To illustrate how such analytical and methodological adjustments can be implemented in research, it reviews an SFL study on Chinese postgraduate students’ use of grammatical metaphor that adopts the approach of treating learner language in its own right. The article also discusses the benefits of this inclusive approach in light of recent trends in applied linguistics and its implications for future SFL research in particular, and language-related research more broadly."
August 2025
Quantum Journal of Social Sciences and Humanities
6(4):56-68
DOI: 10.55197/qjssh.v6i4.739
License: CC BY 4.0
SONG HU
M. H. Chau
Nurul huda Binti Hamzah
https://www.researchgate.net/publication/395127689_TOWARD_AN_INCLUSIVE_APPROACH_TO_LINGUA-CULTURALLY_DIVERSE_STUDENTS%27_LANGUAGE_IN_SYSTEMATIC_FUNCTIONAL_LINGUISTICS_RESEARCH
A federal judge ordered a shake-up of Google's search engine in an attempt to curb the corrosive power of an illegal monopoly while rebuffing the U.S. government's attempt to break up the company and impose other restraints.
#Metaglossia
#metaglossia_mundus
"Afrique : des IA en langues locales pour réduire la fracture numérique
INTELLIGENCE ARTIFICIELLE
Des initiatives sont en cours en Afrique du Sud, au Kenya et au Nigeria pour créer plus d’outils d’IA en langues africaines. Des chercheurs ont déjà enregistré plus de 9 000 heures de discours local afin d’élargir l’accès à cette technologie révolutionnaire sur le continent.
Alors que l’utilisation des modèles de langage basés sur l’intelligence artificielle (IA) explose, avec de nombreux sites web alimentés par l’IA intégrés dans la vie quotidienne, certains en Afrique craignent que leurs populations soient exclues des logiciels principalement occidentalisés.
Malgré la richesse linguistique de l’Afrique, où plus d’un quart des langues mondiales seraient présentes selon certaines estimations, cette diversité est encore peu représentée dans le développement de l’IA.
La principale raison ? un manque d’investissement et de données accessibles. La majorité des outils d’IA, comme ChatGPT, sont entraînés sur des textes en anglais ou dans d’autres langues européennes et chinoises, qui disposent d’immenses bases de données en ligne.
Or, de nombreuses langues africaines, principalement orales, ne disposent pas de suffisamment de textes écrits pour alimenter l’apprentissage automatique, ce qui limite leur intégration dans ces technologies. Résultat : des millions de personnes sont laissées de côté.
Une avancée majeure : un jeu de données unique
Les chercheurs ont récemment publié ce qui serait le plus grand ensemble de données linguistiques pour les langues africaines. « Nous pensons en nos langues, rêvons en elles et interprétons le monde à travers elles. Si la technologie ne reflète pas cette réalité, une partie de la population sera laissée pour compte », explique le professeur Vukosi Marivate de l’Université de Pretoria, qui a pris part au projet.
L’initiative « African Next Voices » a rassemblé linguistes et informaticiens pour créer des jeux de données adaptés à l’IA dans 18 langues africaines. Bien que cela ne couvre qu’une fraction des plus de 2 000 langues parlées sur le continent, les participants espèrent étendre cette base dans le futur.
En deux ans, ils ont enregistré 9 000 heures de discours dans des contextes quotidiens liés à l’agriculture, à la santé et à l’éducation, à Kenya, au Nigeria, et en Afrique du Sud. Parmi les langues enregistrées figurent le kikuyu et le dholuo au Kenya, le haoussa et le yoruba au Nigeria, ainsi que l’isiZulu et le tshivenda en Afrique du Sud.
« Il faut une base de départ. C’est ce que fait African Next Voices. Ensuite, d’autres pourront construire dessus et innover », précise le professeur Marivate. La collecte a été rendue possible grâce à une bourse de 2,2 millions de dollars de la Fondation Gates. Les données seront en accès libre pour permettre aux développeurs de créer des outils de traduction, de transcription ou d’interaction en langues africaines.
Des innovations concrètes
Plusieurs exemples illustrent déjà l’impact potentiel de ces technologies. À Rustenburg, en Afrique du Sud, la fermière Kelebogile Mosime utilise une application appelée AI-Farmer. Elle reconnaît plusieurs langues locales, comme le sotho, l’isiZulu et l’afrikaans, pour lui venir en aide dans ses tâches agricoles.
« En tant que cultivatrice débutante, je fais face à beaucoup de défis », explique-t-elle. « Grâce à cette application, je peux poser des questions en Setswana, ma langue maternelle, et obtenir des réponses utiles. C’est précieux pour moi, surtout en zone rurale où l’accès à la technologie est limité. »
Une autre startup sud-africaine, Lelapa AI, conçoit des outils en langues africaines destinés à des banques et des télécoms. Sa PDG, Pelonomi Moiloa, déplore que les options actuelles soient très restreintes. « L’anglais est la langue de l’opportunité. Pour beaucoup, ne pas le parler peut signifier manquer des services essentiels comme la santé, la banque ou l’aide gouvernementale. On veut changer cela », affirme-t-elle.
Selon le professeur Marivate, ces initiatives ne relèvent pas seulement d’un enjeu économique ou pratique, mais touchent aussi à la justice sociale. Sans un effort concerté en faveur des langues locales, certains craignent que des inégalités linguistiques ne s’accentuent, laissant derrière eux une partie importante de la population africaine face aux évolutions numériques."
https://fr.africanews.com/2025/09/05/afrique-des-ia-en-langues-locales-pour-reduire-la-fracture-numerique/
#Metaglossia
#metaglossia_mundus
"The Art of Proofreading in the Digital Era: Challenges and Technologies That Work
Proofreading is an indispensable element of crafting any paperwork. It helps to achieve clarity and consistency and make content error-free and engaging. The impact of digitalisation, which has seeped into all areas, has caused polishing texts to evolve, with new approaches being introduced. Among other things, these include the integration of AI instruments. And as much as some of us wouldn’t like the artificial intelligence to find a place in this niche, it may come in handy.
Nowadays, technological advancements and traditional methods blend to simplify the process of proofreading to achieve top-quality texts. However, certain challenges still remain.
How Proofreading Has Changed Over The Years
Proofreading stands behind every scientific article, social media post, or magazine column one can come across. The earliest mentions of text checking date back to the ancient civilisations of Egypt and Greece. In that time, people relied on rudimentary knowledge to transcribe texts properly.
The times have changed, and so has the role of proofreading in publishing texts. During the Industrial Revolution, the importance of delivering information consistently and accurately was reconsidered. A few decades ago, making content went digital, and people started using computer software for checking errors and typos.
And nowadays, digital proofreading is having a second birth due to the integration of online checkers based on AI. These revolutionise the industry, allowing us to review texts faster to keep up with the pace of the modern world.
AI Proofreading Tools That May Come In Handy
AI instruments are already integrated into almost any sphere of human life: education, medicine, customer support and even new UK online casinos. Writing and publishing were among the pioneers in utilising software based on artificial intelligence. Here are popular tools for automatic error detection and correction that may come in handy and facilitate a proofreader’s work:
Grammarly. It’s one of the basic instruments for checking grammar, spelling, and punctuation. No repetitive word or incorrect verb form will hide from it. Additionally, it helps to boost the text clarity and accuracy. This writing assistant points out confusing or overly long sentences and gives suggestions on how to change them into simpler alternatives. Another strong side of Grammarly is adjusting text to the style one needs: casual, formal, academic, etc.
QuillBot. Except for correcting the same text aspects as Grammarly does, QuillBot is known for its paraphrasing tool. While its grammar checker is a bit weaker, QB works better for improving text flow and reducing redundancy. If you want to make your text sound more neutral or shorten long writing while keeping the key ideas, this is the way to go. Many use the above-mentioned tool jointly for better results.
Hemingway Editor. This software was inspired by the style of the famous writer. Hemingway Editor is a valuable aid in enhancing the flow and directness of written texts. It does not offer efficient grammar checks and does not paraphrase sentences. Instead, the tool helps reach good readability and make content accessible for readers. It highlights sentences that are hard to read, like confusing structures of passive voice or unnecessary adverbs. Each text gets a score: from 1 to 9 (easy level), 10 to 15 (difficult one and needs improvement), or post-graduate (too complex text or sentence). The desktop app allows people to use the editor without Wi-Fi.
This area of the market is evolving with more companies releasing AI-based solutions to simplify editing. But proofreaders must remember that all these tools are here to complement human efforts, not to replace them.
Common Challenges in Digital Proofreading
Proofreading software could seem like a solution that revolutionises the process of polishing texts and even replaces the human eye. However, there are still some challenges which prevent facilitating tools from becoming a win-win way to edit writings:
Auxiliary instruments and programs cannot understand the context properly. Hence, the suggested corrections may be inappropriate.
Accurate facts and verified data may be corrected by the digital proofreaders (inability to do the quality fact-checking).
Prolonged text reading may cause eye strain that results in missing some significant errors and typos.
Excessive emphasis on language rules leads to marking the word combinations and sentences that are actually correct.
There’s no way to get quality feedback on suggested corrections from proofreading software.
Text-checking tools may make the text lose personalisation and the unique tone of voice maintained by the writer.
Well, the key pitfalls here relate to creativity and contextual understanding, which is yet difficult for digital software.
Traditional Techniques That Still Work
The abundance of tools makes people do less work themselves. However, a range of proven methods for text-checking is still powerful enough. Some of these are:
Read out loud to locate the style inconsistencies.
Read the text backwards to find spelling errors.
Print the text on paper to notice the mistakes and typos.
Check for one type of error in a proofreading session.
The human eye can still play an important role in publishing the written work. At the same time, overlooking endless opportunities provided by AI tools is the road to nowhere. People and technologies should go hand in hand if we want the content to adhere to the highest quality standards."
https://redditchstandard.co.uk/lifestyle/proofreading-in-the-digital-era-tools-and-challenges/
#Metaglossia
#metaglossia_mundus
"Interembodiment and Language: Deaf Language Practices as Resistance
•Language is both a vehicle for interembodiment and an interembodied process.
•Signing animates bodily entanglements with others differently than speaking.
•Semiotic repertoires are co-constructed and shared among and across individuals.
•Informal interpreting helps deaf immigrants navigate the immigration system.
Abstract
This article approaches language as both a vehicle for interembodiment and as a fundamentally interembodied process. Drawing from approximately 18 months of ethnographic fieldwork with deaf immigrants in the northeast United States, this article explores how interembodiment emerges through language forms and practices; and how deaf language practices entangle interlocutors, sometimes bodily, with others. As intelligibility can only be achieved relationally, this article likewise asserts language as a fundamentally interembodied process. Meaning in communicative encounters is made among and across interactions (not individually). Given the inherent interdependence of language, I detail how deaf immigrants co-construct meaning with family members and friends through practices of informal interpreting and how, through informal interpreting, deaf immigrants can better navigate an immigration system that is ill designed to address their complex communication access needs. In this way, interdependent and embodied deaf language practices become a form of resistance to policies and procedures that traditionally marginalize or exclude those whose bodyminds are considered non-normative."
Erin Mellett
Cite
https://doi.org/10.1016/j.socscimed.2025.118544
Get rights and content
Highlights
https://www.sciencedirect.com/science/article/abs/pii/S0277953625008755
#Metaglossia
#metaglossia_mundus
"WHEN it comes to African literature, translation has mostly meant translating work from European languages into African ones. Translation from African languages into English has been long overdue.
Now it appears that a shift in the movement of stories across languages is underway. Works first written and published in African languages are increasingly being translated into English for a wider readership.
As an academic specialising in African literature and publishing, I am optimistic about the launch of a new book series African Language Literatures in Translation by the University of Georgia Press edited by US-based academics Christopher Ouma and Alexander Fyfe.
The Mad is one of the inaugural titles in the series. It’s a translation of Zimbabwean author Ignatius Mabasa’s much loved novel Mapenzi/Yhe Mad. The Mad has been jointly released in the UK and Zimbabwe by Carnelian Heart Publishing and amaBooks in July 2025.
Mapenzi is written in Shona, but the ambitious series aims to translate a range of significant African works from other languages too, including Kiswahili and Sesotho.
This helps shift the terrain of African literature. It allows English readers to encounter African novels as they were first imagined, in the rhythms, idioms, and sensibilities of African languages. It helps counter the erasure of African languages in world literature.
When Mabasa’s debut novel Mapenzi appeared in 1999, it was recognised as a landmark text in Shona literature. Shona is a Bantu language widely spoken in Zimbabwe.
Mapenzi tells the story of a disillusioned young war veteran who becomes the uncensored witness to the false promises of independence in Zimbabwe. He witnesses the collapse of social values, and the madness of a society trapped between hope and despair.
The novel’s daring style and unruly humour drew comparisons with Dambudzo Marechera, Zimbabwe’s literary provocateur. And with the stylistic innovations of Charles Mungoshi, the great craftsman of both Shona and English prose.
Mapenzi quickly won accolades and was adopted as a school text for a generation that came of age at the turn of the millennium. Since then, Mabasa has become a central figure in the promotion of indigenous African writing.
His commitment and investment was shown when he became the first academic to write and submit a PhD in Shona at Rhodes University in South Africa. Mabasa is also a translator in his own right and most recently he helped translate Zimbabwean novelist Tsitsi Dangarembga’s Nervous Conditions into Shona.
His career has been a reminder that languages such as Shona are not minor, but vibrant mediums for complex thought and artistic innovation.
Now, more than two decades later, Mapenzi can be read in English for the first time.
The Mad
The Mad was translated by J Tsitsi Mutiti and most unfortunately there is no information about her in the book - not even a cursory internet search yields little to nothing.
In works like this, that cross language, culture and geography, the translator plays a crucial role, not just in the text’s language but in shaping its tone, context and accessibility. A translator’s note would have provided insight into the challenges and decisions involved in the process. Instead, the translator and the process of translation remain invisible.
This lack of acknowledgement overlooks the labour and interpretive skill required to bring such a work to life in another language. The hope is that future editions will correct this omission. Leaving it unaddressed risks diminishing the very cross-cultural understanding that translation seeks to foster.
Translating Mapenzi is no small task. Mabasa’s prose brims with poetry, satire and linguistic play. Its cadences are deeply rooted in Shona idiom. The novel’s innovative structure includes chapters titled after characters, abstract concepts, places and song lyrics, pushing the boundaries of form and style in Shona literature.
In her translation Mutiti meets this challenge with remarkable skill, sustaining Mabasa’s lyricism and rhythm without smoothing out its texture. The result is a translation that feels alive, attentive both to the sense and the spirit of the original.
That said, there are occasional lapses into flatly literal translation, particularly in the dialogue. These moments feel more like direct transpositions from the Shona than deliberate stylistic choices in English. As a result, the translation at times struggles to assert a voice of its own.
Translation as decolonisation
Nonetheless, the publication of The Mad highlights how translation in African literature is not just a tool for accessibility. It is also a critical, interpretive and archival practice that reshapes the canon, its circulations and readerships. The Mad contributes to African literature’s global visibility and intellectual vitality.
For decades, the global image of African literature has been shaped largely by writers who chose or were compelled to write in colonial languages.
Kenyan author and academic Ngugi wa Thiong’o, who recently passed away, has been one of the most famous advocates for writing in African languages. He insisted that the struggle for decolonisation could not be separated from the struggle over language. Yet, as he himself often admitted, African-language writing has remained marginal in the circuits of publishing and translation.
A series like this aims to help change that. Other forthcoming titles include Zanzibari writer Ali Hilal Ali’s Mmeza Fupa (translated as The Swallowers of Bones by Meg Arenbeg), Kenyan novelist and sociologist Katama Mkangi’s Walenisi (translated as They Are Us by Richard Prins), Lesotho writer and teacher Ntseliseng Masechele Khaketla’s Left Behind (translated by Makafane Tsepang Ntlamelle), and Halfani Sudy’s Kirusi Kipya (translated as New Virus by Jay Boss Rubin).
In this context, The Mad is more than an English version of a Shona classic. It is part of an invitation to rethink what African literature is and where it comes from. Mabasa’s novel, in Mutiti’s supple translation, demonstrates how much vitality lies in the vernacular imagination, and how translation can open doors without erasing the local textures of language.
With The Mad, a new and vital moment begins.
Tinashe Mushakavanhu is assistant professor at the Harvard University. This article is republished from The Conversation under a Creative Commons license."
Culture / 2 September 2025
Translation as decolonisation
As African novels are being translated to English in a bold new trend, TINASHE MUSHAKAVANHU puts into wider context the publication of Ignatius Mabasa’s The Mad (Mapenzi)
TINASHE MUSHAKAVANHU
https://morningstaronline.co.uk/article/translation-decolonisation
#Metaglossia
#metaglossia_mundus
|
Faut-il parler à ChatGPT et aux IA en français ? Cette étude classe les 26 meilleures langues à utiliser
Si vous avez pris l'habitude de parler à ChatGPT ou Gemini en anglais pour avoir de meilleurs résultats, vous pouvez maintenant oublier car ce n'est pas une bonne solution. Une étude vient de chambouler nos certitudes, et le français tire brillamment son épingle du jeu !
Beaucoup de personnes ont ce réflexe : quand on veut une réponse vraiment pointue de ChatGPT, de Gemini ou de n'importe quelle IA générative, on bascule sur l'anglais. On se dit, logiquement, que la langue de Shakespeare, celle qui domine internet et les bases de données d'entraînement, donnera forcément un résultat plus pertinent. Eh bien, mettez cette idée à la poubelle.
Une étude très sérieuse, menée conjointement par des chercheurs de l'Université du Maryland et de Microsoft, vient de détruire cette croyance.
L'anglais détrôné, le français en embuscade !
C'est le coup de tonnerre de ce rapport. Les chercheurs ont testé la performance d'une flopée de modèles (OpenAI, Google Gemini, Qwen, Llama...) sur 26 langues différentes pour voir comment ils comprenaient les demandes, appelés prompts. Le résultat est sans appel.
Dans leur rapport d'étude, les auteurs ne cachent pas leur étonnement :
Notre expérience a donné lieu à des résultats surprenants et peu intuitifs. Tout d'abord, l'anglais n'a pas obtenu les meilleurs résultats dans tous les modèles ; en fait, il est arrivé en sixième position sur 26 langues lors de l'évaluation de textes longs...
Sixième ! L'anglais, que l'on pensait roi, se fait griller la politesse. Mais alors, qui est sur le podium ? Cocorico (ou presque), le français se hisse à une magnifique deuxième place. Mais le grand gagnant, celui auquel personne ne s'attendait, c'est le polonais.
Voici le top 10 des langues les plus efficaces pour parler à une IA :
Polonais (88%)
Français (87%)
Italien (86%)
Espagnol (85%)
Russe (84%)
Anglais (83,9%)
Ukrainien (83,5%)
Portugais (82%)
Allemand (81%)
Néerlandais (80%)
Précision grammaticale > Volume de données ?
Mais alors, comment le polonais, une langue réputée pour sa complexité (oui, ses déclinaisons font passer le latin pour un jeu d'enfant) et disposant d'un corpus de données d'entraînement bien plus faible que l'anglais, peut-il être le plus efficace ?
L'Office polonais des brevets, qui s'est félicité de la nouvelle sur Facebook, avance une explication : c'est la langue la plus précise pour donner des ordres à l'intelligence artificielle...
Ce que l'étude semble montrer, c'est que la précision structurelle d'une langue pourrait être plus importante que le volume de données brutes.
L'anglais est une langue relativement simple, qui repose beaucoup sur le contexte et l'ordre des mots.
Le polonais, le français ou l'allemand, avec leurs conjugaisons plus riches, leurs genres grammaticaux et leurs déclinaisons (surtout pour le polonais et l'allemand), ne laissent que peu de place à l'ambiguïté.
Un ordre en polonais est, semble-t-il, verrouillé grammaticalement. L'IA sait exactement qui fait quoi et comment. L'étude note d'ailleurs que toutes les langues du top 10 appartiennent à la famille indo-européenne.
À l'inverse, le chinois, malgré un volume de données colossal, s'est classé avant-dernier, suggérant que sa structure est peut-être moins adaptée aux architectures actuelles des IA.
Faut-il donc tous se mettre au polonais pour rédiger nos prompts ? Pas du tout. Ce que cette étude nous dit surtout, c'est que si vous êtes francophone, vous n'avez absolument aucune raison de passer à l'anglais pour obtenir de meilleurs résultats. Votre langue maternelle est déjà l'une des plus performantes du monde pour dialoguer avec l'IA.
https://share.google/BlCSvSmTzdKYCfiS5
#Metaglossia
#metaglossia_mundus