 Your new post is loading...
|
Scooped by
Charles Tiayon
November 13, 2011 11:31 PM
|
Gov. Andrew M. Cuomo has been lavishing attention on minority communities amid signs of impatience and a perceived lack of attention to their concerns. In his 11-minute speech here on Friday, Mr. Cuomo cited several achievements of his first months in office: doubling the state’s goal for participation by businesses owned by members of minority groups and women in state contracting; suspending the state’s participation in a controversial federal immigration enforcement plan; closing some prisons and juvenile justice centers and directing state agencies to provide free interpretation and translation services in six foreign languages.
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
"How the EU translates
Thursday 17 July 2025
By Sam Morgan
The European Union has 24 official languages and must provide translations of all crucial legislation in those native tongues. It is a mammoth but necessary task.
There are 27 EU members and 24 official languages.
They are: Bulgarian, Croatian, Czech, Danish, Dutch, English, Estonian, Finnish, French, German, Greek, Hungarian, Irish, Italian, Latvian, Lithuanian, Maltese, Polish, Portuguese, Romanian, Slovak, Slovenian, Spanish and Swedish.
Each EU citizen is entitled by law to read official legislation in their native tongue. That means a lot of translation is needed.
Luxembourgish is not counted among the official list, as it is long-standing policy of the Grand Duchy’s government not to seek official status for the language. Austrians speak German and Cypriots speak Greek.
Interestingly, the Cypriot government has flirted with the idea of pushing for Turkish to be accepted as an official language of the EU, but that idea has always turned out to be rather impossible once exposed to the vagaries of politics.
After Brexit, there were some calls for English to be dropped as an official language, but that rather ludicrous idea forgot that it is de facto the lingua franca of the Union, not to mention that it is an official language in both Ireland and Malta.
Irish was initially under a derogation, where only certain top-level legislation was required to be translated into that language. But it has since been lifted, meaning every single one of the 24 tongues have equal status.
Official translations are handled by the various translation departments of the respective European institutions. The Commission’s does a lot of the heavy lifting, translating mostly from English into the 21 other language combinations.
Technology is playing an increasingly important role in translation, as large language models are becoming more sophisticated day by day.
Machine translation is never perfect but for some language combinations can do most of the work, simply requiring post editing by a qualified translator before it can be published.
Experts are divided about how good machines will actually get at translation. Some languages are very poor fits for one another and there is not much material for algorithms to pull from.
For example, the body of translated work between Croatian and Estonian is probably very limited indeed.
English, French and German are the working languages of the EU but every bit of official legislation has to be translated at some point.
For newer member states like Croatian and Bulgarian, that meant employing a whole team of translators to sift through all the existing legislation from the last seventy-odd years of Union activity.
There is also the interesting question of what will happen when the Union adds another member in the future. At the moment there are 552 possible language combinations, if you add one more country, that shoots up to 600.
Would new countries insist on having their language be an official tongue? Montenegrin and Serbian, for example, are extremely similar to Croatian, so would separate translation departments have to be set up?
The politics of language are fascinating and it is one aspect of multilingualism that will not be replaced anytime soon by technology."
https://www.brusselstimes.com/eu-affairs/1664808/how-the-eu-translates
#metaglossia_mundus
"Arabic translations in HK digital publishing soon
2025-07-16 HKT 17:34
Secretary for Culture, Sports and Tourism Rosanna Law said on Wednesday that Hong Kong's enhanced digital publishing initiative will soon incorporate Arabic translations in the hope of strengthening cultural ties with the Middle East.
Law made the announcement at a launch ceremony for Publishing 3.0+, where she highlighted the success of its predecessor, Publishing 3.0.
That project enabled approximately 60 local publishers to convert paper publications into 5,000 bilingual e-books and audiobooks in Cantonese, Putonghua and English, capitalising on digital transformation.
Building on this success, Publishing 3.0+ leverages artificial intelligence to further modernise the industry.
Law said the enhanced initiative will utilise large language models to significantly improve multilingual translations and content conversion capabilities.
"Publishing 3.0+ will refine the e-books and audiobooks converted over the past years and further deliver another 5,000 translated and converted copies of books," she said.
"The fruitful expected outcome of 10,000 e-books and audiobooks will be showcased in Hong Kong pavilions at various international book fairs in the coming months."
Law also confirmed plans to incorporate Arabic translations into the project.
"During my official visit to the Middle East in April this year, I was deeply impressed by the region's rich cultural and historical heritage," she said.
"I am most delighted that Publishing 3.0+ will contribute to our effort to realising closer cultural links and friendship between the Middle East and Hong Kong, aligning the shared visions of both regions."
Sharon Wong, one of the chairs at Publishing 3.0+, said the first phase saw a 75 percent improvement in e-book conversion speed and that the new initiative enhances local publishers’ multilingual capabilities.
"I hope that through this project, we can bring copyright holders, creators and AI developers together and work together in enhancing Hong Kong as a cultural centre or IP trading hub because we are under rapid AI development," she said.
"That is to promote the publishing industry and to promote our cultural product overseas.""
https://gbcode.rthk.hk/TuniS/news.rthk.hk/rthk/en/component/k2/1813620-20250716.htm?spTabChangeable=0
#metaglossia_mundus
The July 14 memo encourages agencies to determine which services would be better operated exclusively in English and to make use of AI where translations are needed.
"Justice pushes agencies to use AI-assisted translations, when offering them at all
By NATALIE ALMSJULY 16, 2025 04:46 PM ET
The July 14 memo encourages agencies to determine which services would be better operated exclusively in English and to make use of AI where translations are needed.
ARTIFICIAL INTELLIGENCE
JUSTICE
The Department of Justice is leading an effort to reduce multilingual services deemed “non-essential” across the government. When translations are needed for government products, DOJ is recommending that agencies use technology to get the job done.
The push to use artificial intelligence and machine learning is part of DOJ’s implementation of a March executive order deeming English the official language of the United States. That order rolled back another decades-old policy requiring agencies to enhance access for people that have limited English proficiency.
Although AI and machine learning could expedite translation work and potentially save the government money, the government’s own how-to resource on translation technology cautions that technology-assisted translations should still involve a human translator.
A new DOJ memo calls for “responsible” AI, but provides little detail on how agencies should address risk.
“Technological advances in translation services will permit agencies to produce cost-effective methods for bridging language barriers and reducing inefficiencies with the translation process,” the new guidance says.
“The Department encourages other agencies to follow its approach of considering responsible use of artificial intelligence and machine translation to communicate with individuals who are limited English proficient,” it continues.
The accuracy of AI-assisted translation generally worsens in more specialized contexts, like when the technology is used for legal or policy documents, according to Sonny Hashmi, the head of global public sector at Unqork and former appointee at the General Services Administration during the Biden administration.
And in these contexts, accuracy especially matters, he said, asking “what is the impact of somebody not understanding the instructions in a form … and as a result, missing out on a benefit that they're otherwise owed?”
Agencies may want to either train bespoke models for these jargony situations, or have people validate the accuracy of translations and offer end users that have questions a clear way to get answers, he said.
The government’s own guide on AI and translation tells agencies to include humans in the process, something the DOJ memo doesn’t reference at all.
Agencies “should not rely solely on automatic machine translation services or computer-aided technology,” reads a government resource on translation technology housed on digital.gov, which is run by GSA. “All translations should be checked by a competent human translator.”
“AI can expedite and save costs substantially,” said Amy Holmes, partner and principal at Holmes Consulting Group, which is focused on data, AI and digital transformation. “But you need to have some guardrails in place to ensure that it’s accurate.”
DOJ offered no comment when asked about risk mitigation.
“You don’t want to just take a generative AI translation tool and just flip it on and assume everything is perfect,” said Michael Boyce, the generative AI lead for the civic tech group U.S. Digital Response, which has collaborated on translation work using generative AI in New Jersey.
Boyce, who formerly directed the AI Corps at the Department of Homeland Security, emphasized the need for testing and user input on AI-assisted translations.
Still, when technology can potentially speed up translations and help improve access, “you also are creating other types of downside risks” when you don’t use the tool to help translate government materials, said Boyce.
Cutting multilingual services
Trump’s March executive order didn’t require agencies to remove or stop creating products in languages other than English. A fact sheet about the order touts flexibility for agencies to decide when and how to offer services in languages other than English.
But Attorney General Pam Bondi’s memo does recommend that agencies determine “which of their programs, grants and policies might serve the public at large better if operated exclusively in English,” where allowed by law.
"The Department of Justice will lead the effort to codify the President’s Executive Order and eliminate wasteful virtue-signaling policies across government agencies to promote assimilation over division,” Bondi said in a statement.
DOJ “will lead a coordinated effort to minimize non-essential multilingual services,” the memo states.
The administration’s Department of Government Efficiency has reportedly already canceled at least ten language or translation service contracts.
Civil rights law requiring language access in federally funded programs and services remains on the books, although the DOJ did rescind its guidance on the matter under the latest memo, which also set in motion new, forthcoming DOJ guidance.
Critics say that the policies will harm taxpayers’ access to government services. The Census Bureau estimates that 8.3% of people in the U.S. have limited English proficiency.
“Diversity is what makes America strong and unique,” Kica Matos, president of the National Immigration Law Center, said in a statement after Trump signed the order in March.
“[This order] is just another notch in Trump’s belt of cruelty and xenophobia that will make it harder for people to understand and access essential services, like education and health care,” she continued."
https://www.nextgov.com/digital-government/2025/07/justice-pushes-agencies-use-ai-assisted-translations-when-offering-them-all/406776/?oref=ng-homepage-river
#metaglossia_mundus
El Centro de Literatura Islandesa, como cada año, dispone una serie de becas de traducción abiertas a editoriales extranjeras que deseen publicar una obra traducida del islandés. En esta ocasión la convocatoria cierra el próximo 15 de septiembre.
"El fondo de traducción del Centro de Literatura Islandesa está abierto hasta el próximo 15 de septiembre
Publicado por Lorenzo Herrero | Jul 8, 2025 | 0
El Centro de Literatura Islandesa, como cada año, dispone una serie de becas de traducción abiertas a editoriales extranjeras que deseen publicar una obra traducida del islandés. En esta ocasión la convocatoria cierra el próximo 15 de septiembre.
El objetivo del fondo es promover la literatura islandesa en el extranjero. Se ofrecen subvenciones para literatura (prosa, poesía y teatro), no ficción de interés general (incluyendo ensayos y biografías), cómics y libros infantiles.
Las solicitudes de apoyo solo pueden presentarse una vez que el solicitante haya adquirido los derechos de la obra y firmado un contrato con el traductor. El Centro de Literatura Islandesa no concederá apoyos a las traducciones publicadas antes de la fecha límite para la presentación de la solicitud.
El formulario de solicitud está disponible...
Las solicitudes pueden realizarse por un máximo del precio de traducción acordado con el traductor. No obstante, por lo general, las ayudas cubren solo una parte del precio de traducción. La subvención a la traducción se aplica exclusivamente a los honorarios base del traductor, no a las regalías ni a una posible parte de las ganancias procedentes de audiolibros o libros electrónicos.
¿Qué debe contener la solicitud?
Se deberán adjuntar al formulario de solicitud los siguientes documentos:
El CV del traductor y los títulos traducidos y publicados previamente. El traductor debe traducir a su lengua materna.
Una copia del contrato con el titular de los derechos en islandés
Una copia del contrato con el traductor
Tres páginas de texto de muestra (traducción) y una copia del mismo texto en el idioma original
¿Cómo se evaluará su solicitud?
Al evaluar una solicitud, se valora la calidad de la obra en cuestión y la cualificación del traductor. Normalmente, se espera que el traductor sea un profesional con al menos una obra traducida, publicada o interpretada, dentro del mismo género que la obra para la que se solicita la subvención. En el caso de la no ficción, el criterio principal del comité es que la obra sea de interés general. No se conceden subvenciones para la traducción de otros tipos de no ficción, como libros de texto educativos..."
https://publishnews.es/el-fondo-de-traduccion-del-centro-de-literatura-islandesa-esta-abierto-hasta-el-proximo-15-de-septiembre/
#metaglossia_mundus
"Abstract: Based on the English-Chinese Parallel Corpus of Children’s Literature and the Corpus of Chinese Children’s Literature, this study investigates the feature of normalization in the Chinese translation of English children’s literature. Normalization refers to the adaptation of foreign features in the source text to comply with the cultural and linguistic norms of the target culture. The study analyzes both macro and micro levels of language features in translated children’s literature, comparing them with original Chinese and English texts. The findings reveal a clear trend towards normalization, evidenced by shorter sentences, increased repetition of high-frequency words, a lower frequency of hapax legomena, and a higher textual readability in translated Chinese versions. Furthermore, linguistic structures such as reduplication, modal particles, “把” (BA), and “得” (DE) constructions are found to occur at rates comparable to or significantly higher than those in the original Chinese corpus. This paper argues that normalization is a creative outcome, molded by translators aligning with reader expectations, conscientiously considering the psychological characteristics of child readers, and adapting to social, cultural, and market influences. The study contributes to understanding linguistic features of translated children’s literature, sheds light on translation universals, and underscores the dynamic interplay between normalization and translator creativity."
Published: 08 July 2025
Normalization or creation? A corpus-based study of normalization in the Chinese translation of English children’s literature
Yang Han
Humanities and Social Sciences Communications volume 12, Article number: 1051 (2025) Cite this article
https://www.nature.com/articles/s41599-025-05379-6
#metaglossia_mundus
"...Taipei, July 8 (CNA) The 2025 Taiwan-Ireland Poetry Translation Competition is now accepting entries, inviting translators worldwide to take on a poem by Taiwanese poet Dong Shu-ming (董恕明), organizers announced Tuesday.
This year's featured work is "Like a Song -- to 107-year-old Mumu on Her Journey"-- an elegy written in a combination of Chinese, Bopomofo, English, and the indigenous Puyuma language, said the National Museum of Taiwan Literature, which co-hosts the event with Trinity College Dublin's Centre for Literary and Cultural Translation.
The poem pays tribute to Dong's grandmother, a 107-year-old elder whose life embodies a century of Puyuma resilience, cultural memory, and multilingual heritage, the museum said.
"Through richly textured imagery, multilingual cadences, and intergenerational echoes, the poem traces mumu's life as a woman of the Pinaski (a Puyuma tribe) community, spanning languages from Puyuma to Japanese to Mandarin and beyond," Trinity said in a press release.
Now in its fifth year, the competition has previously featured poems in Chinese, Taiwanese and Hakka by poets including Tsao Yu-po (曹馭博), Cheng Shun-tsong (鄭順聰), Tseng Kuei-hai (曾貴海), and Temu Suyan (黃璽).
The initiative was launched in 2021 by Trinity College Dublin and the Taipei Representative Office in Ireland.
Winners will be invited to take part in an international online exchange in November 2025 to share insights on translation and creative writing. The deadline for submissions is Sept. 10, 2025.
For more information and to submit your entry, visit: https://lnkd.in/eHdXn6qT
Enditem/AW"
(By Chiu Chu-yin and Lee Hsin-Yin)
July 8, 2025
https://lnkd.in/ewDug4vT
#metaglossia_mundus
Find out why it's a great idea to study English Language and Linguistics in the most linguistically diverse city in Western Europe.
"Linguistics and English Language
Linguistics and English Language
Study
Why study here?
Explore the ways in which English dialects differ, how children learn language, how languages arise, how they change and so much more.
Did you know?
You'll be studying at one of the UK's top four institutions for linguistics.
The Complete University Guide 2020
Linguistics is the ideal subject for analytical minds. You will delve into the science of language - an everyday phenomenon which impacts our lives on an individual and a global scale. English Language covers the history of the English language and the variation between English dialects in the UK and further afield.
Linguistics and English Language at The University of Manchester is unrivalled in its exceptional breadth of subject areas and theoretical approaches.
We are ranked in The Complete University Guide 2019 top 10 universities for Linguistics in the UK.
Opportunities
Access cutting-edge resources, including one of the largest holdings of linguistics texts in the UK
Conduct research using English manuscripts held in our prestigious Special Collections
Work in close collaboration with internationally renowned experts
Conduct fieldwork and analyse linguistic corpora
Study in the most linguistically diverse city in Western Europe – home to over 150 languages
Specialist master's exploring the full breadth and depth of linguistic theory
Become the best you can be by participating in Stellify for true personal and professional growth.
Careers and employability
Study Linguistics and English Language at Manchester and you will develop a range of analytical and problem-solving skills that can be applied to many careers.
The University of Manchester is also the most targeted university in the UK for top graduate employers, according to High Fliers Research..."
https://www.humanities.manchester.ac.uk/linguistics-and-english-language/study/why-study-here/
#metaglossia_mundus
A few years ago a controversy arose at UNH over American Sign Language. For a long time the introductory course in it had two or three sections, but suddenly the sections ballooned to eight or ten.
"Speaking of Words: Is ASL a Language?
A few years ago a controversy arose at UNH over American Sign Language. For a long time the introductory course in it had two or three sections, but suddenly the sections ballooned to eight or ten. A commission was set up to look into the reasons for this surprising jump in students. Were they all really interested in communicating with deaf people? Students planning to be social workers had long been taking a year of ASL to help them in their future work, but these new students were not making such plans.
What we found out is that a lot of students had learned it was easy to get a good grade in a year-long course in ASL and thereby dispose of the foreign-language requirement for the BA. A survey showed that the average grade was A- and the amount of homework very modest; with twice as much homework students in first-year Spanish were earning an average grade of B-. This seemed unfair, somehow, but what could be done? I was appointed to the commission, so I read several books about ASL, watched videos, and listened carefully to a colleague on the commission who was a native signer as well as native speaker of English and a professor of linguistics at UNH Manchester. I did not master ASL, but I learned a lot about it.
Indepth Your Inbox
Get unbiased nonprofit watchdog news sent directly to your inbox.
Your Email
As I discussed the problem in my own department (English) and the Languages Department I found there was a good deal of ignorance about it, some of which I had shared before joining the commission. Some thought that ASL was the same as “signing English,” that is, spelling out English with manual equivalents of letters, something like Braille or Morse Code. Such a thing exists, but that is not ASL, which is in no way dependent on English or on any other spoken language. The fact that it is called American Sign Language means only that it is the sign language used by the deaf in America (and Canada outside Quebec). Some signers of ASL do not know English at all, though most have learned to read and write English, and some to lip-read it, and some to vocalize it.
ASL is a language in its own right. It is a “natural language” like English and Urdu. It is not a pidjin, which is a limited system constructed by adults to communicate across two or more languages; it is more like a creole, which is typically the product of children. ASL, and the other sign languages around the world, were created by deaf people, with at least a large input by children signing it as their first language.
Anyone studying it as a second language will be struck at how differently the words are ordered from the way they are in English. The train of signs usually begins with the topic, not what we call the grammatical subject. For example, “Do you have many children?” would begin with children: CHILDREN + HAVE + MANY? + YOU. The YOU (pointing a finger) may be left out, of course, as it may be implied by the situation, and the question may be indicated by a raised eyebrow.
Sometimes the word order reflects the actual sequence of events: NEW YORK + THERE + HOTEL + ARRIVE + FINISH + TIRED + I = “I was exhausted by the time I arrived at the hotel in New York.” FINISH is a sign that indicates the completed aspect of the preceding verb, making it similar here to the English past tense. The “I” at the end is another pointing finger, very possibly omitted.
There is a lot of pointing in ASL, of course, and it means that it can incorporate a large number of pronouns. If you are telling a long story about your family gathering at Christmas, the first time you mention your mother you sign MOTHER and point to a space on one side of the line of sight between you and the one you are signing to. When you first mention your father, your aunt, your sister, your brother, you assign them each to a unique spot on one or the other side of the line. Further mentions of the same people require just pointing to the right spot. In English “she” would be confusing, since it might refer to your mother, aunt, or sister, but in ASL they are differentiated in space. ASL has no problem either with the non-binary “they/them”: that non-binary person gets their own little spot in space waiting to be pointed at when needed.
Signing is done with the face as well as the hands. In a spoken language we can add quite a bit with hand gestures, tones of voice, and a few facial expressions as well, but ASL enlists the face in important ways, often as adverbs modifying the verb. An expression of disgust or contempt, with the tongue between the teeth like a strong th-sound, means “carelessly.” So “John does his homework” as a neutral fact would be JOHN + WRITES + HOMEWORK. But to say “John does his homework carelessly” (or “John blows off his homework”) would be JOHN + th-WRITES + HOMEWORK, where the th-gesture with the tongue is made during the manual sign for WRITE.
I have taken these examples from the books I read while serving on the commission. I did not learn enough or practice enough to talk with a deaf person, but I gained a lot of admiration for the community that created ASL and continually adds to it in brilliant and often witty ways. As for the unusual problem at UNH, it was pretty much resolved when it was discovered that each department in the College of Liberal Arts can decide if ASL counts as the language requirement for their majors. English and many other departments decided not to count it, though the linguistics majors may take it as their second foreign language requirement. Certainly, if you are studying English literature, knowing some French or Latin is more useful than knowing ASL. But students who really care about it can still study it, and if they really care about it they will grow a new and wonderful visual language in their head.
I am happy to hear from readers: mferber@unh.edu. If any of them know ASL and can correct any mistakes, I would be especially glad to learn from them.
Michael Ferber moved to New Hampshire in 1987 to join the English Department at UNH, from which he is now retired. Before that he earned his BA in Ancient Greek at Swarthmore College and his doctorate in English at Harvard, taught at Yale, and served on the staff of the Coalition for a New Foreign Policy in Washington, DC. In 1968 he stood trial in Federal Court in Boston for conspiracy to violate the draft law, with the pediatrician Benjamin Spock and three other men. He has published many books and articles on literature, and has a deep interest in linguistics. He is married to Susan Arnold; they have a daughter in San Francisco."
https://indepthnh.org/2025/07/16/speaking-of-words-is-asl-a-language/
#metaglossia_mundus
"Xi jinping likes to call the world’s dependence on advanced Chinese technologies his shashou jian. He means it’s the ace up China’s sleeve, but translated literally into English, shashou jian is “assassin’s mace”.
Translation has been an issue since the beginning of America’s formal relationship with Communist China. And as relations between the two great powers enter a less-predictable phase, linguistic sleight of hand has the potential to become more consequential..."
Jul 15th 2025
https://www.economist.com/podcasts/2025/07/15/lost-in-translation-how-language-complicates-us-china-relations
#metaglossia_mundus
Recent federal reductions in funding for language assistance and President Donald Trump’s executive order designating English as the official language of the United States have som⁵e health advocates worried that millions of people with limited English proficiency will be left without adequate support and more likely to experience medical errors.
"Lost in Translation: Interpreter Cutbacks Could Put Patient Lives on the Line
By Vanessa G. Sánchez
JULY 16, 2025
Federal law entitles patients to interpreters if they don’t have a strong grasp of English. KFF Health News correspondent Vanessa G. Sánchez appeared on WAMU’s “Health Hub” on July 9 to explain why some Trump administration policies are leaving patients fearful to ask for language services.
Patients need to communicate clearly with their health care provider. But that’s getting more difficult for those in the U.S. who don’t speak English.
Budget cuts by the Trump administration have left some providers scrambling to keep qualified medical interpreters. And an executive order designating English the official language of the United States has created confusion among providers about what services should be offered.
Patients who don’t speak English are left afraid, and perhaps at risk for medical mistakes. What happens when those who need help are too frightened to ask?
In WAMU’s July 9 “Health Hub” segment, KFF Health News correspondent Vanessa G. Sánchez explained why health advocates worry these changes could lead to worse patient outcomes."
https://kffhealthnews.org/news/article/medical-interpreter-funding-staff-cuts-patient-lives-english-language-services/
#metaglossia_mundus
"Nevada needs court interpreters. See if you qualify
LAS VEGAS (KLAS) — An urgent need for qualified court interpreters exists in Nevada as the demand for language services grows.
Nevada courts handled more than 3,100 cases in 2024 that relied on interpreters to help participants understand and engage in legal proceedings, according to a Tuesday news release.
The Supreme Court of Nevada invites bilingual individuals to explore opportunities in the justice system as certified court interpreters. Registration for the Fall 2025 Certified Court Interpreter Exam is now open. The deadline to register for the Oct. 20 exam is approaching: Sept. 22.
The most requested languages last year were Spanish, Tagalog, Cantonese and Vietnamese. Court officials said Nevada had only 79 certified Spanish court interpreters, with an urgent need for more.
“Court interpreters make the justice system accessible for thousands of Nevadans,” Amanda Walker, Language Access Program administrator, said. “We’re asking bilingual residents to be part of this essential public service.”
Interpreters licensed in other states may be eligible.
Oral exams are scheduled in Las Vegas (Nov. 3-7) and in Carson City (Nov. 10, Nov. 12). Written exams are scheduled on Nov. 7 in Las Vegas and Nov. 10 in Carson City. Find more information and a self-assessment tool online to evaluate your readiness before registering.
To become a certified court interpreter in Nevada, applicants must complete an online Written Orientation Workshop, available mid-July. This virtual workshop introduces candidates to Nevada’s judicial system, court interpreting standards, and interpreter ethics.
Applicants must also pass the written exam with a minimum score of 80 percent. The test is two hours and 15 minutes long and includes 135 multiple-choice questions that cover language proficiency, legal terminology, and professional conduct."
by: Greg Haas
Posted: Jul 15, 2025 / 11:56 AM PDT
Updated: Jul 15, 2025 / 11:56 AM PDT
https://www.8newsnow.com/news/local-news/nevada-needs-court-interpreters-see-if-you-qualify/
#metaglossia_mundus
"English-only: DOJ announces plan to phase out costly translations
WASHINGTON, D.C., (KTAL/KMSS) – The federal government will soon eliminate what it calls “wasteful virtue-signaling policies” involving multilingual services.
On Monday, President Trump’s Executive Order No. 14224 established English as the official language of the United States, and Attorney General Pam Bondi wasted no time asserting that the DOJ would follow suit and eliminate what she sees as costly translations “to promote assimilation over division.”
To comply with the president’s EO, Bondi said her department will rescind prior Limited English Proficiency (LEP) Guidance, passed during the Clinton Administration, which prohibited national origin discrimination affecting people with limited English proficiency.
The DOJ will also conduct an internal inventory of all existing non-English services and release a plan to phase out “unnecessary multilingual offerings” Department-wide. AG Bondi said the DOJ will temporarily suspend the operation of LEP.gov and other public-facing materials related to language access for individuals with LEP, including letters, internet posts, YouTube videos, and training materials, pending internal review.
Within 60 days, the DOJ will solicit input and recommendations from other federal agencies regarding programs and policies that can be implemented in an English-only format. This new guidance will help agencies prioritize English while explaining when and how to implement multilingual assistance as needed to fulfill the missions of federal agencies.
Recommended actions for communicating with those with limited English proficiency
Machine learning translations using AI are recommended as a cost-effective measure to facilitate communication with individuals who have limited English proficiency.
Consider English-only services where allowed by law, to consider programs, grants and policies would serve the public better if they were English-only.
Disclaimers announcing that English is the official language of all federal information.
Redirect funds that were used for LEP translation initiatives toward research and programs that will improve English proficiency.
President Trump’s Executive Order rescinds Executive Order No. 13,166, signed by President Clinton on August 16, 2000. Executive Order No. 13,166 directed agencies to enhance access to federal programs for persons with limited English proficiency and required tailored guidance for recipients of federal funding, straining federal resources and impeding the assimilation of new Americans, AG Bondi said in the DOJ announcement."
by: Marlo Lacen
Posted: Jul 15, 2025 / 11:13 AM CDT
Updated: Jul 15, 2025 / 03:04 PM CDT
https://www.ktalnews.com/news/u-s-world/english-only-doj-announces-plan-to-phase-out-costly-translations/
##metaglossia_mundus
"General Hamza Johari has urged the Office of the Chief Parliamentary Draftsman to fast-track the translation of all national laws into Kiswahili, especially election-related laws and regulations, to ensure public access before this year’s general election.
Mr Johari issued the directive during his visit to the Office of the Chief Parliamentary Draftsman (OCPD) pavilion at the recently concluded 49th Dar es Salaam International Trade Fair (DITF). During the visit, he also inquired about the issues and concerns raised by citizens at the pavilion to enable his office to take appropriate action.
“I believe we must continue strengthening our efforts in translating laws, especially election laws and their regulations. Doing so will significantly improve access to justice for our citizens,” said the Attorney General.
He commended the OCPD for its public awareness efforts at the exhibition and for the vital legal work it continues to undertake in fulfilling its statutory responsibilities.
Meanwhile, the Minister for Constitution and Legal Affairs, Dr Damas Ndumbaro, emphasised the crucial role of the OCPD in the legislative process, stating that no law can be enacted in Parliament without the involvement of this key institution.
Dr Ndumbaro highlighted the office’s importance in drafting, revising and translating both principal and subsidiary legislation.
“The Office of the OCPD is central to the entire process of drafting, amending and interpreting laws in the country. Without this office, no bill would be tabled in Parliament. It is also responsible for drafting amendments and subsidiary legislation. In essence, it is the legal engine room of the nation,” he said.
He further explained that the judiciary relies heavily on the OCPD to ensure accurate and up-to-date legal references.
“If this office ceased to function, courts would struggle to interpret and apply laws,” he noted, highlighting the OCPD’s role in verifying legal texts and identifying outdated statutes.
He also praised its efforts to translate laws into Kiswahili for greater public accessibility.
Earlier, OCPD Legal Officer Ms Mariam Possi briefed visitors on key legal topics, including the lawmaking process, legal amendments and the role of translation in promoting justice.
She reaffirmed the office’s commitment to legal drafting, revision and translation in line with existing regulations."
https://dailynews.co.tz/ag-fast-track-kiswahili-laws-translations/
#metaglossia_mundus
Le gouvernement fédéral compte désormais privilégier le plus bas soumissionnaire lorsque vient le temps de choisir ses interprètes pigistes.
Ottawa–Gatineau
Interprètes : pour économiser, le fédéral souhaite privilégier le plus bas soumissionnaire
L’Association internationale des interprètes de conférence (AIIC) Canada estime que ces propositions risquent de compromettre les engagements électoraux du gouvernement Carney sur le renforcement du bilinguisme officiel.
André Boisjoli (Consulter le profil)
André Boisjoli
Le gouvernement fédéral, le plus grand employeur d’interprètes au Canada, envisage de prendre le plus bas soumissionnaire lorsque vient le temps de choisir un interprète pigiste. Une option qui inquiète des voix du milieu.
Services publics et Approvisionnement Canada (SPAC) souhaite toujours choisir le meilleur interprète pour l’affectation, mais compte maintenant se tourner vers le fournisseur qui offre le prix le plus bas.
C'est en tout cas ce qu'on peut lire dans une demande d'information concernant les services d'interprétation parlementaire et de conférence de haut niveau dans les langues officielles, dont Radio-Canada a obtenu copie.
Dans le cadre de notre mandat visant à moderniser les pratiques d’approvisionnement et à assurer une gestion responsable des fonds publics, nous révisons nos processus tout en maintenant notre attachement à la qualité des services offerts, notamment en matière d’interprétation, déclare dans une réponse écrite le cabinet du ministre de la Transformation du gouvernement, des Travaux publics et de l’Approvisionnement, Joël Lightbound.
Ce changement aurait presque certainement pour effet d'écarter les pigistes les plus chevronnés d'une équipe déjà en sous-effectif. [...] La qualité ne serait plus évaluée de manière objective, ce qui ouvrirait la voie au favoritisme, au copinage et à d'autres pratiques déloyales et dangereuses, a indiqué l’AIIC Canada par voie de communiqué de presse mercredi.
L’association s’inquiète aussi de la suppression de mesures qui visent à protéger la santé auditive des interprètes, notamment en ce qui a trait à la qualité sonore."
https://ici.radio-canada.ca/nouvelle/2179948/interprete-parlement-langue-officielle
#metaglossia_mundus
"Google is rolling out new AI features for its search engine, including a more advanced language model and the ability for the AI to call local businesses on behalf of users.
Soon, users will be able to pick between two different models in AI Mode using a drop-down menu. For complex tasks involving logic, math, or programming, Gemini 2.5 Pro will provide better answers. Most everyday searches will still default to what's likely the less powerful (and cheaper) Gemini 2.5 Flash, though Google doesn't make this distinction clear in the interface.
Gemini 2.5 Pro, which launched in March 2025, is currently Google's most advanced language model. It can process up to one million tokens at once, understands multiple types of input, and handles complex reasoning. Google updates it regularly with small improvements. According to the Artificial Analysis Intelligence Index, which combines results from several standard benchmarks, only OpenAI's o3-pro and the new Grok 4 outperform it.
Deep Search for tougher questions
Google is also rolling out Deep Search, which it describes as its "most advanced search tool." Built on Gemini 2.5 Pro, Deep Search can run hundreds of queries at once, gather information from different sources, and assemble detailed, referenced reports.
Google says Deep Search is intended for more involved research, such as work projects, hobbies, or academic assignments. The company also lists potential uses like looking into real estate or analyzing finances, with the goal of reducing lengthy research to a shorter process.
The naming can be confusing. After recently sorting out its AI subscription tiers with "AI Pro" and "AI Ultra," Google now has two similarly named features: Deep Search in AI Mode and Deep Research in Gemini—both meant for advanced search tasks.
Deep Search and Gemini 2.5 Pro will debut first for AI Pro and AI Ultra subscribers in the US who join the AI Mode experiment in Google Labs. Eventually, Google plans to roll them out globally.
AI that makes phone calls for you
One of the more unusual new features lets Google's AI call local businesses to check prices and availability. Searching for "dog groomers near me," for example, will show a "Let the AI check prices" button that triggers an automated call.
The AI contacts multiple providers, then organizes the responses—like available appointments and services—into a simple summary. Google says this saves time for users while giving businesses another way to handle bookings.
This calling feature will launch first for all US users, with paid subscribers getting higher usage limits. Businesses can decide if they want to receive AI-generated calls through their business profile settings."
https://the-decoder.com/google-brings-gemini-2-5-pro-and-deep-search-to-ai-mode-and-adds-ai-phone-calling-to-search/
#metaglossia_mundus
"Abstract: This study addresses the challenges of translating culture-specific nonverbal communication in literature, focusing on the complexities of conveying rich kinetic-visual imagery across languages. By examining 18 English and French translations of a sample verse, the research aims to explore how translators handle kinesic and oculesic elements intrinsic to the source text. The comparative analysis reveals that while many translators struggle to convey cultural nuances, often simplifying or misinterpreting nonverbal cues, some successfully navigate these complexities. Factors such as chronology, interpretive choices, and domestication strategies significantly impact the translation process. The study’s findings underscore the importance of cultural awareness and sensitivity in achieving effective literary translations, highlighting the need for translators to act as cultural mediators rather than mere linguistic converters. This research contributes to the field by demonstrating how varying approaches to translating nonverbal cues can significantly impact the preservation of a text’s cultural essence, thereby advancing our understanding of the intricacies involved in cross-cultural literary translation and emphasizing the translator’s role in bridging cultural gaps..."
https://www.tandfonline.com/doi/full/10.1080/23311983.2025.2526941?src=#abstract
#metaglossia_mundus
"Abstract: Generative language modelling has surged in popularity with the emergence of services such as ChatGPT and Google Gemini. While these models have demonstrated transformative potential in productivity and communication, they overwhelmingly cater to high-resource languages like English. This has amplified concerns over linguistic inequality in natural language processing (NLP). This paper presents the first systematic review focused specifically on strategies to address data scarcity in generative language modelling for low-resource languages (LRL). Drawing from 54 studies, we identify, categorise and evaluate technical approaches, including monolingual data augmentation, back-translation, multilingual training and prompt engineering, across generative tasks. We also analyse trends in architecture choices, language family representation, and evaluation methods. Our findings highlight a strong reliance on transformer-based models, a concentration on a small subset of LRLs, and a lack of consistent evaluation across studies. We conclude with recommendations for extending these methods to a wider range of LRLs and outline open challenges in building equitable generative language systems. Ultimately, this review aims to support researchers and developers in building inclusive AI tools for underrepresented languages, a necessary step toward empowering LRL speakers and the preservation of linguistic diversity in a world increasingly shaped by large-scale language technologies..."
License: CC BY 4.0
arXiv:2505.04531v2 [cs.CL] 08 Jul 2025
Overcoming Data Scarcity in Generative Language Modelling for Low-Resource Languages: A Systematic Review
Josh McGiff
josh.mcgiff@ul.ie
Nikola S. Nikolov
Department of Computer Science and Information Systems, University of LimerickLimerickIreland https://share.google/AOO4jEsEQMTmFIjNX
#metaglossia_mundus
✨ Human + AI-Assisted Workflow Leveraging the latest state-of-the-art AI technologies and natural language processing, Colibri generates tailored codeframes, translates verbatims, and enables dynamic chat-based data exploration.
Nuance Launches Analytics Tools for Open-Ended Data
July 15 2025
Nuance, a subsidiary of Texas-based Decision Analyst with a focus on language-based analytics, has launched a verbatim coding and analytics platform called Colibri, which it says will transform how researchers and analysts explore, process, visualize, plot and interpret unstructured information.
Colibri is the Latin American word for a hummingbird, and according to the company the new platform 'brings speed and precision' to the process of coding open-end survey data and textual data of any type, in over 30 different languages.
Decision Analyst specializes in strategy research, new product development, ad testing and advanced modeling to aid marketing decisions. Nuance's expertise lies in understanding global consumers through 'the words, phrases, photos and expressions of human behavior.'
Colibri features AI-enhanced coding tools for faster, more consistent classification of open-ended data; a project dashboard for management of multiple studies in one place; built-in translation capabilities for more than 30 different languages; and collaboration and reporting tools.
'Colibri combines decades of expertise in verbatim analysis with cutting-edge software development and advanced analytics,' according to Nuance President Lisa Hazen (pictured). 'It's intuitive, collaborative, and scalable, offering researchers a smarter, more productive way to work with open-ended data.'
CEO Jerry W. Thomas comments: 'Lisa Hazen and Claire Jones are the creative geniuses behind the design of Colibri, and Cody Parker, Johnny Seay, and Justin Thomas are the software engineering savants bringing the creative ideas to life. Colibri is the future of verbatim coding and textual analytics'.
The firms are online at www.nuancecoding.com/colibri and www.decisionanalyst.com .
Earn a Master's in Translation & Interpreting from NYU. Full- or part-time study online. Comprehensive curriculum and world-class faculty. Apply today!
"The online Master's (MS) in Translation and Interpreting offered by the NYU SPS Center for Publishing, Writing, and Media prepares you for a range of fulfilling careers, whether you are interested in business translation, medical translation, legal translation, literary translation, transcreation, localization, or project management.
This comprehensive online program provides students with the best of both worlds—the convenience and flexibility of learning on your own terms and at your own pace, while benefiting from a rigorous curriculum, a variety of elective courses tailored to your interests, and the prestige of earning your master’s degree at NYU, one of the world's most respected universities.
Connect the World through Translation and Interpreting
Tuition*
$2,575 Per credit
$26,520 Per term (10-12 credits)
# of Credits
36 credits
Study Options
Online
Fall | Spring
Full-time | Part-time
Minimum Duration
15 months
*See NYU Bursar site for more info.
9% job growth in this degree area
$76,200
median national salary for this degree
95%
of our grads working, in school, or both
1,294
Global Center for Publishing, Writing and Media alumni
[1] Lightcast 2025
[2] Lightcast 2025
[3] NYU Life Beyond the Square: SPS Undergraduate Class of 2024
[4] as of 1/2024
Why This Degree?
Fully Online
This flexible, asynchronous format allows opportunities to engage with faculty members and other students at regular intervals.
A Strong Foundation
You can study translation from any language into English. Our curriculum covers both written translation and oral interpreting best practices. From here, you will have the opportunity to work in a wide range of disciplines including law, science, literary translation, social work, and more.
Tailored to Your Goals
Customize your learning experience with our elective courses and real-world translation projects.
Path Toward Credentialing
Prepare for the American Translators Association (ATA) Certification Exam, a language industry-recognized credential.
Expert Faculty
Learn translation and interpreting skills from and network with senior professionals in your field.
Designed for Your Schedule
Both full- and part-time study options are available
What You’ll Study
Core Curriculum
Core courses cover the major fields and foundations of the language professions, including translation theory and practice, editing and revision, translation technology, terminology management, and principles of interpreting.
Electives
Elective courses are designed to prepare students for multiple career paths, including legal and financial translation, literary translation, transcreation, website localization, machine translation, and project management.
Thesis Project / Capstone
In their final semester, students pursue a substantial translation or localization project, research paper, or a combination of the two."
More information 👇🏿👇🏿👇🏿
https://www.sps.nyu.edu/explore/degrees-and-programs/ms-in-translation.html
#metaglossia_mundus
IA et langues locales : le Burkina Faso veut former sa population à une tech inclusive
"(Agence Ecofin) - Face aux défis que rencontre le Burkina Faso dans le développement de solutions d’intelligence artificielle adaptées à ses réalités, une approche innovante émerge pour démocratiser l’IA : intégrer les langues locales afin d’élargir l’accès aux technologies et dynamiser l’emploi des jeunes.
En partenariat avec le Programme des Nations unies pour le développement (PNUD), le ministère de la Transition numérique a organisé, le vendredi 4 juillet, une réunion de concertation sur la promotion de l’IA dans les langues locales du pays.
Cette démarche entend rendre l’IA accessible à une majorité de Burkinabè souvent exclus des progrès numériques à cause des barrières linguistiques. Parallèlement, elle soutient la formation de jeunes techniciens à un usage responsable et éthique de ces technologies, ce qui favorise l’inclusion numérique et la création d’emplois locaux...
Avec un peu plus de 16 millions d’habitants, soit 72,4 % de la population burkinabè vivant en milieu rural selon un rapport 2022 de l’Institut national de la statistique et de la démographie (INSD), la numérisation des langues locales devient un levier clé pour faciliter l’appropriation de l’IA. Cette stratégie ouvre des perspectives concrètes pour améliorer l’accès à l’éducation numérique et aux services publics digitaux.
En formant la jeunesse à ces innovations, le Burkina Faso prépare une main-d’œuvre qualifiée prête à répondre aux besoins du secteur technologique, contribuant ainsi à la croissance économique nationale.
Le succès de cette initiative repose sur la formation des formateurs, un cadre réglementaire solide et l’adhésion des acteurs locaux. D’autres pays africains, comme le Kenya et le Rwanda, ont lancé des projets similaires avec des résultats encourageants.
Au Kenya, Simba AI développe un chatbot multilingue qui traduit l’anglais en langues locales telles que le Kikuyu, le Meru et le Kalenjin. Cette innovation facilite l’accès à l’IA pour des populations souvent exclues, tout en valorisant les langues indigènes et en créant des opportunités pour les jeunes développeurs.
Au Rwanda, l’application Mbaza a joué un rôle clé durant la pandémie de Covid-19, diffusant des informations via un chatbot multilingue, notamment en Kinyarwanda. Ce projet collecte aussi des données linguistiques essentielles pour adapter l’IA au contexte local et aux besoins des populations.
Pour le Burkina Faso, la prochaine étape consistera à généraliser l’usage de ces outils dans les écoles et à renforcer les synergies entre les secteurs public et privé afin de stimuler l’innovation. Cette stratégie pourrait positionner le pays comme un leader régional de l’IA inclusive au service d’un développement équitable.
Félicien Houindo Lokossou (stagiaire)
Edité par Sèna D. B. de Sodji"
https://www.agenceecofin.com/actualites-services/0907-129906-ia-et-langues-locales-le-burkina-faso-veut-former-sa-population-a-une-tech-inclusive
#metaglossia_mundus
RDC : des écrivains plaident pour le renforcement du kiswahili dans l’enseignement pour avoir des répercussions sur la littérature congolaise
Le 7 juillet de chaque année, le monde célèbre la Journée mondiale de la langue kiswahili. C’est une langue d'une importance capitale en Afrique de l’Est et centrale, servant de langue officielle ou nationale dans plusieurs pays. Elle est omniprésente en Tanzanie et au Kenya, est devenue langue officielle en Ouganda, et constitue l’une des quatre langues nationales de la RDC, où elle est particulièrement parlée dans l’est du pays, avec 9,1 millions de locuteurs.
Par ailleurs, cette langue demeure encore peu présente, notamment dans le domaine de la littérature en RDC. Laëtitia Malira Tembeya, écrivaine et éditrice vivant à Goma, estime que cette langue est plus présente dans la musique que dans la littérature parce qu’elle est plus facilement consommée à travers le rythme et la mélodie. En revanche, dans le système éducatif, très peu de temps est accordé à l’enseignement du swahili dans les écoles, d’autant plus que la langue varie selon les régions.
« Il ne m’est encore jamais arrivé d’écrire en swahili, et je ne l’envisage même pas. La majorité de mes lecteurs étant francophones, ils ne s’intéressent pas aux livres en swahili. Pour toucher un public plus large, je préfère écrire dans une langue avec laquelle je suis plus à l’aise. De plus, la maison d’édition dans laquelle j’évolue ne dispose pas d’une communauté de lecteurs swahiliphones. D’ailleurs, la plupart des plateformes publient principalement des livres en français », a-t-elle confié.
La perception sociale du français comme langue de prestige incite certains éditeurs à privilégier son usage pour gagner en notoriété.
« Cette faute incombe à notre système éducatif. Moins on accorde d’importance à cette langues56, moins la population sera capable de lire un livre entier en swahili, car elle n’y a pas été exposée dès le bas âge », ajoute l’écrivaine.
De son côté, Edimo Moïse, animateur culturel et défenseur des langues congolaises, souligne que le kiswahili demeure peu promu en raison de préjugés l’associant à une tribu.
« L’absence de formalisation du swahili tel qu’il est parlé au Congo, qui est assez différent de celui des pays voisins ; l’absence de son enseignement à l’école ; ainsi que la propagation de la fausse idée selon laquelle un "peuple swahili" existerait au Congo — alors que chacun des peuples qui parle swahili en RDC a sa langue propre qu’il perpétue également », a-t-il expliqué, révélant que sur l’ensemble de ses publications, une seule est en swahili.
Il précise que, selon lui, pour mieux intégrer cette langue dans la littérature, il faudrait l’enseigner aux enfants à travers des méthodologies didactiques basées sur l’éveil de la curiosité, l’utilisation quotidienne des savoirs nouvellement acquis, ainsi que la pratique de la langue au sein de la famille.
Notons que le kiswahili est l'une des langues les plus utilisées de la famille bantoue et la plus parlée en Afrique subsaharienne. Il fait partie des dix langues les plus parlées au monde, avec plus de 200 millions de locuteurs. Cette langue est l’une des principales langues véhiculaires de nombreux pays d’L'Afrique to orientale, centrale et australe, ainsi que du Moyen-Orient. Elle est également enseignée dans les principales universités et écoles supérieures à travers le monde.
Consciente de sa portée mondiale croissante, l'Assemblée générale des Nations unies a adopté la résolution A/RES/78/312, qui affirme l'importance du kiswahili dans la promotion de la solidarité, de la paix et de l’unité panafricaine.
Au-delà des frontières de la RDC, des communautés swahiliphones existent au nord du Mozambique, aux Comores et dans certaines régions de la Somalie. Son rôle est multiple et vital : il facilite la communication interethnique, est crucial pour le commerce, agit comme un puissant vecteur de culture et d’identité est-africaine, et se distingue comme la seule langue africaine reconnue langue de travail par l’Union africaine, soulignant ainsi sa portée panafricaine."
Gloria Kisenda
9 juillet 2025 - 09:42
https://actualite.cd/2025/07/09/rdc-des-ecrivains-congolais-plaident-pour-lintegration-du-kiswahili-dans-lenseignement
#metaglossia_mundus
Master the art of interpreting in a special education setting with UGA’s online Professional Interpreter in Special Education certificate course.
"This certificate course is designed for language interpreters who participate in special education meetings in K-12 settings. In such meetings, the interpreter plays a key role for officials, teachers, and parents. The interpreter’s ability to convey information accurately and completely can influence a child’s placement, support, and eligibility for special education services.
The Professional Interpreter in Special Education certificate course prepares interpreters to navigate complex regulatory, ethical, and cultural issues involved in how school systems deliver special education services.
CREDITS
3.00
DURATION
30.00
hours
NEXT OFFERING
October 20, 2025
FORMAT i
Online
COST
$
849.00
Available offerings
Professional Interpreter in Special Education [Fall 2025]
DATES
10/20/2025
–12/8/2025
FORMAT i
Online
CODE
0614-016
COST
$849.00
About this course
What you’ll learn
You will learn the educational interpreter’s roles in discussions of psychological evaluations, eligibility, Individualized Education Plans, reevaluations, and post-high school transition plans.
Through interactive discussions, practice and vocabulary-building exercises, you will also learn:
Concepts, components, and terminology of special education
The importance of special education legislation
How to apply the Interpreter’s Standards of Practice to promote trust and communication among parents and schools
How to explain special education processes—such as eligibility, Parent Rights, and Individualized Education Plans—to parents
How to remain impartial and accurate when interpreting difficult concepts
Modes of interpretation in special education settings
How to deal with the stresses and emotions of being an interpreter
Learning objectives
Module 1: Introduction to Special Education Interpretation
Understand the special skills needed to effectively and accurately interpret for special education meetings
Describe legislation pertaining to special education
Recognize the roles of special education professionals and support personnel, including interpreters
Module 2: Standards of Practice in Special Education Interpretation
Demonstrate knowledge of the interpreter’s Standards of Practice and Code of Ethics as they apply to special education interpretation
Discuss and define the appropriate application of the Standards of Practice to case scenarios
Module 3: Overview of the Special Education Process
Define the identification, referral and eligibility processes involved in special education
Describe procedural safeguards offered through the special education parental rights
Enhance knowledge of special education terminology in the areas of special education identification, referral and eligibility
Module 4: The Individualized Education Plan
Identify the main sections of an Individualized Education Plan and the role of the interpreter in the process
Describe different types of IEPs and possible support services that a student can receive
Enhance knowledge of special education terminology in the area of Individualized Education Plans
Module 5: Common Disabilities and Chronic Conditions
Identify common disabilities and chronic conditions that the interpreter may encounter in the school system
Describe assistive technology and supportive services available to help students with disabilities and chronic conditions
Evaluate appropriate ways to handle interpreter stress and emotions when dealing with families and students with disabilities and chronic conditions
Enhance knowledge of special education terminology in relation to common disabilities and chronic conditions
Module 6: Additional Terms to Know
Describe additional terms used in the explanation of IEP services
Understand the purpose of transition plans and terminology related to them
Identify aids and related services that students may be eligible for in special education
Module 7: Discipline and the Special Education Student
Identify behavior intervention plan components and the role of the interpreter in behavior intervention meetings
Examine disciplinary consequences and accommodations provided to special education students involved in disciplinary incidents
Describe Manifestation Determination meetings and possible outcomes
Enhance knowledge of special education terminology in relation to disciplinary and behavior interventions
Module 8: Course Wrap-Up, Post-Test and Final Exam
Expand your resources and continue exploring ways to build your skills as a professional interpreter in special education
Create a professional development action plan
Review your knowledge of interpretation in special education and assess your learning throughout the course
Who should attend?
K-12 educational interpreters, bilingual school staff, and contract/independent interpreters who already have a strong foundation in the basics of interpretation and seek to master the art of interpreting in a special education setting.
Continuing Education Information
Successful graduates will earn a Digital Certificate of Program Completion and 3.0 CEUs from the University of Georgia.
Requirements & policies
Schedule
Live online classroom sessions with the instructor will be held from 6:30 p.m. – 8:30 p.m. EST on the following Tuesdays:
October 21
November 4
November 18
December 2
The instructor will be available for office hours on October 28, November 11, and November 25.
Prerequisites
You must be fluent in both English and a second language to complete the certificate requirements. Since this is an advanced interpretation course, it is recommended that you take the Professional Interpreter in Education Certificate Program or that you have knowledge of the foundations of interpreting in a school setting.
This course requires the equivalent of Advanced Mid or higher on the American Council on the Teaching of Foreign Languages (ACTFL) Oral Proficiency Assessment (OPI). Please see the ACTFL website(Opens in a new window) for more information regarding proficiency guidelines. For prospective students who are unsure of their proficiency level, we highly recommend taking the Oral Proficiency Assessment (OPI) before registering for the course.
People
Instructor
Ana Soler, the course author and instructor, is the Chairperson of the National Association of Educational Translators and Interpreters of Spoken Languages (NAETISL) and Founder of SeSo, Inc., a source of qualified and trained interpreters and multicultural family engagement workshops. She completed her bachelor’s degree in Social Work at Georgia State University, her master’s degree in Public Health at Emory University, and is a Ph.D. in Special Education student at the University of Georgia. For over a decade, Ana worked with the largest school district in Georgia as the Language Services and Parent Outreach Coordinator, developing, implementing, and evaluating professional development opportunities for multilingual personnel and supporting schools with interpreter/translator quality assurance.
While working as a Multicultural Program Coordinator for Children’s Healthcare of Atlanta, Ana assisted with the creation of the hospital’s interpretation and translation department. She also coordinated interpreter training for bilingual staff, oversaw cultural competency training for physicians and staff, and provided medical translation quality assurance and guidelines. Ana has authored interpreter training curricula nationally, including the Intercultural Parent and Youth Leadership Program, the Interpretation Academy for Bilingual High School Students, the Arkansas Interpreter in Education Credential Training, a 40-hour course for medical interpreters, and other online courses for the University of Georgia, including the Professional Interpreter in Education Certificate course and the Professional Interpreter in Special Education Certificate course. She remains an active medical and educational interpreter and translator..."
https://www.georgiacenter.uga.edu/professional-programs/professional-interpreter-in-special-education-certificate-program/
#metaglossia_mundus
" Kamanthi Wickramasinghe
The OED is a historical dictionary - its job isn’t only to define a word as we use it today, but to trace its full life story
OED has always relied on people to show how language lives in the world
Be it asweddumized fields or mallung in a rice and curry meal, kiribath for a special occasion, sizzling kottu, scrumptious watalappam or a happening baila song – each one of these examples provides a glimpse of Sri Lanka in a nutshell. The aforementioned words have been used in our vocabulary for decades, highlighting the unique cultural, social and culinary diversity of Sri Lanka. This is why these words and a few others made their mark in the Oxford English Dictionary’s (OED) latest update, in appreciation of the origins and evolution of Sri Lankan English. To find out more about how these words made their entry into a world-acclaimed dictionary, the Daily Mirror spoke to Rochana Jayasinghe, the consultant for Sri Lankan English for the OED’s World Englishes update.
A journey shaped by a fascination for language
Jayasinghe’s journey into language study had begun with a deep love of reading from a young age. “I was always fascinated by words—their histories, meanings, and the patterns that connected languages across cultures. As a child, I enjoyed learning new languages and instinctively noticed similarities among them, which I later came to understand as an early curiosity in etymology and historical linguistics,” she said.
Her academic path was shaped significantly during her time at the Department of English at the University of Peradeniya, “where conversations around language use, and in particular, World Englishes, were particularly vibrant and intellectually stimulating,” as described in her own words. Jayasinghe said that the Department has long been at the forefront of interrogating language, identity, and post-colonialist. “Their scholarship has contributed much to thinking critically about the politics of English, the localisation of the language, and its ideological implications,” she added.
Though her formal training is in literature with a Master’s in World Literatures in English - it was through literary study that her interest in language itself deepened. “During my Master’s at the University of Oxford, I researched Sri Lankan-origin words in the Oxford English Dictionary (OED), published by Oxford University Press. My focus was on the correspondence between Robert Burchfield, then-editor of the OED, and Pearl Cooray, who was affiliated with the Dictionary Department of the University of Ceylon, later absorbed by the Ministry of Cultural Affairs. Their exchange, spanning from 1971 to 1981, offered a window into the collaborative history of lexicography between Sri Lanka and the OED. This research led to my selection - through a vetting process, lexicographical aptitude test, and working visit - as the consultant for Sri Lankan English for the OED’s World Englishes update,” she explained.
Understanding the context of a dictionary
Words such as asweddumize, which is to do with preparing the earth for harvesting, have made its entry into OED
kiribath and kottu (shown bottom left) were included in the Oxford English Dictionary because food is generally visible and a shared part of any culture
Apart from simplifying complex words for users, Jayasinghe opined that different dictionaries have different aims and philosophies. “So when we ask, ‘what is a dictionary, the answer depends on which one we’re talking about. I can only speak with confidence about the Oxford English Dictionary (OED), which is the one I’ve studied in depth and worked with,” she said in response to a query on what exactly is a dictionary.
“To begin with, it’s worth noting that Oxford University Press publishes two major dictionaries: the Oxford Dictionary of English, which is a current-use dictionary aimed at offering concise definitions for everyday reference; and the Oxford English Dictionary, or OED, which is quite different. The OED is a historical dictionary - its job isn’t only to define a word as we use it today, but to trace its full life story: where it came from, how its meaning has shifted over time, and how it’s been used in context, across literature, newspapers, legal documents, and more,” she explained.
“So, when people ask, ‘Why is a word in the OED’, it’s important to understand that the OED isn’t giving us permission to use these words—and never claimed to. That’s not its function. The OED’s role is to record how people are already using language, not to decide what is or isn’t “correct.” Sri Lankan English didn’t suddenly become “real” the moment it appeared in the dictionary. It’s been real for decades. We’ve spoken it, sung baila in it, written school essays in it, argued in it, and cracked jokes in it,” she added.
Latest additions to the OED include kiribath and mallung. According to Jayasinghe, these words don’t need any dictionary to validate them. “Everyday speakers of Sri Lankan English have stood by these words for generations, choosing not to translate them into “milk rice” or “sautéed greens,” but instead insisting on their continued use in English-language contexts—because those words carry more than just meaning; they carry an entire cultural world,” she opined.
Jayasinghe said that the OED has always relied on people to show how language lives in the world. “When James Murray became its first Chief Editor in 1879, he quickly realised that no one person—or even one team—could capture the entirety of the English language. So, he put out a public appeal: asking volunteers to send in examples of words they encountered in books, newspapers, speeches, or journals. These handwritten slips—often including the sentence, source, and date—poured in from around the world. Teachers, clergy, civil servants, and avid readers contributed,” she added.
“In many ways, the OED was one of the earliest examples of crowdsourced knowledge. It wasn’t simply created by scholars in Oxford, but by readers around the globe; people noticing how words were used and caring enough to document them,” she explained further.
“So, when words from Sri Lanka appear in the OED, it isn’t about the dictionary ‘granting’ recognition, but about acknowledging what speakers of Sri Lankan English have already long been doing - using these words as part of the language’s living, evolving story,” said Jayasinghe further underscorings the uniqueness of Sri Lankan English.
Process of adding words to OED
Responding to a query on the process of adding words to OED, Jayasinghe said that James Murray’s philosophy and practice still live on. “Anyone can suggest words. There is an entire process that involves carefully tracking how a word is used over time—across books, newspapers, academic texts, social media, and even spoken language—and assessing whether it is used widely and consistently enough in a way that fits the OED’s particular inclusion policy.
A dedicated team of editors, etymologists, and library researchers work together to trace the word’s history, examine spelling and pronunciation variations, and compile illustrative quotations that show how the word functions in real-world contexts,” she explained.
Speaking about her contribution to the groundwork, she said that as the consultant for Sri Lankan English she collected linguistic evidence, suggesting possible definitions, and providing sociolinguistic and cultural context for how certain Sri Lankan English words are used.
An ‘incredibly meaningful’ experience
“Incredibly meaningful,” is how she describes her overall experience as a consultant for Sri Lankan English for OED. “I have had the chance to engage with the work of Sri Lankans who have contributed to the Oxford English Dictionary over the years. In addition to Pearl Cooray whom I have mentioned, Richard Boyle, a former consultant himself, did much to raise awareness about Sri Lankan contributions to the OED, and his book Knox’s Words remains a landmark study of Sri Lankan-origin words in the OED. I also came across figures like Donald Ferguson, who wrote extensively on language in Sri Lanka, and even Dr. William Chester Minor—one of the OED’s most prolific early contributors—who, as it turns out, was born in Ceylon and lived here until his teenage years. These kinds of unexpected connections really brought home the long, complex relationship between Sri Lanka and the English language,” she further said.
For Jayasinghe, working in the OED archives has been the favorite part of this journey. “There was something quietly thrilling about holding handwritten slips, some over a hundred years old. The archivist, Beverley McCulloch, could even recognise particular people’s handwriting, just by sight; it was a level of familiarity I found so moving. It made me realise that dictionary-making was not just an intellectual process, but a profoundly intimate, human one,” she added.
She said that she has been lucky to have the support of such generous people. “Dr. Danica Salazar, Executive Editor for World Englishes, has been a kind, thoughtful, and encouraging guide throughout the process. And meeting Dr. Peter Gilliver, Associate Editor of the OED and one of its most experienced lexicographers, was a highlight. His research on the history and making of the OED had already helped me feel connected to the work I was doing, but speaking to him in person gave me a deeper sense of the continuity and dedication that underpin this massive project,” Jayasinghe said.
Sri Lankan English – a living, evolving variety
When asked about Sri Lankan English and its uniqueness, Jayasinghe began by stressing that she doesn’t consider herself capable of speaking for Sri Lankan English – nor can any one person do so. “There are many academics, writers, teachers, and everyday speakers who have shaped, studied, and reflected on Sri Lankan English for decades. My role was simply to support the OED’s broader mission of documenting language usage,” she said.
“What I can say as someone who studies language use is that Sri Lankan English is what happens when it grows roots in our soil. It is a living evolving variety that blends Sinhala, Tamil, Malay, English, and more – just like our cuisine, our music, and our everyday life. Its uniqueness lies in how it reflects Sri Lankan identities, humour, politics, and belonging. That said, it is unique in the same way all varieties of World Englishes are unique—each one grows out of its own specific context. The value of Sri Lankan English isn’t in how different it is from British or American English, but in how it may reflect our local realities while still participating in a global language,” Jayasinghe explained.
From Asweddumize to Papare
When asked how certain words such as asweddumize made its entry into OED, Jayasinghe said that interestingly, asweddumize was one of the words everyone thought was already in the dictionary. “It has certainly been part of OED discourse for nearly a century, but it missed inclusion at the time due to insufficient evidence. It resurfaced in 1971 with the intervention of Pearl Cooray, and again in the 1980s, but was set aside once more. Now, with broader access to historical and contemporary Sri Lankan sources, the editors were finally able to gather enough evidence - including a first recorded use from as early as 1857 - to support its inclusion,” she explained.
Jayasinghe further said that words like kottu roti, watalappam, mallung and kiribath were included because food is generally a visible and shared part of any culture, and that makes food-related terms especially likely to appear in English usage across communities. “These words frequently turn up in writing - menus, travel writing, cookbooks, media - and are often left untranslated because there aren’t equivalent “English” words that can be used. Other terms like baila and papare were added for similar reasons,” she added.
She further said that a ‘few words didn’t make the cut this time’, but that it doesn’t mean it won’t be added in future. “The OED is a living dictionary, so words are always being revisited as new material becomes available,” she explained further.
Bringing about social harmony through sociolinguistics
The new set of words added to the OED represent diverse cultures. Speaking about the role of sociolinguistics in bringing about social harmony Jayasinghe explained that sociolinguistics shows us how people relate to each other through language. “In a multi-ethnic, multi-diverse country like Sri Lanka, where identities, histories, and cultures intertwine, language is a living expression of who we are,” she added.
But she said that with regard to the new set of words, some media outlets and social media pages have run with the headline “Sinhala words added to the dictionary!”, which is both inaccurate and unhelpful. “Many of the words we’re talking about are nourished by a rich mix of Sinhala, Tamil, Malay, Portuguese, English, and other influences, showing us that language is not fixed or owned by any single group; it’s a living, breathing entity shaped by everyone who uses it. Recognising this alone, helps us move beyond simplistic labels and appreciate language as a shared resource that reflects the complexity of our society. When we embrace language as fluid and collective, it can become a powerful tool for connection, understanding, and social harmony in a multi-ethnic, multi-lingual country like Sri Lanka.
“Use language freely, but listen carefully”
When asked to give a message on the use of language, Jayasinghe said that for her, the key idea once again, is that language is both fluid and collective. “It shifts, blends, borrows, and adapts. In Sri Lanka, many of us may naturally move between Sinhala and Tamil, or Tamil and English - sometimes all three languages (and at times even more) - in our everyday lives. Often, it’s hard to pinpoint where one language ends and another begins. But that’s not a weakness—it’s a strength. It reflects how cultures live alongside each other, influence one another, and evolve together. And the more we accept that language doesn’t have to stay within rigid boundaries, the more space we create for creativity, mutual respect, and genuine understanding,” she added.
“One principle I’ve tried to hold on to is this, which is my message - use language freely, but listen carefully. Pay attention to how it shifts, how others use it, and what it carries for them. That kind of attentiveness opens the door to empathy. In a country as richly diverse as Sri Lanka, that’s one of the most powerful steps toward meaningful connection,” she underscored."
https://www.dailymirror.lk/news-features/How-Sri-Lankan-English-adapted-borrowed-and-blended-into-the-Oxford-English-Dictionary/131-314277
#metaglossia_mundus
Mistral's open-source speech model Voxtral can recognize multiple languages, understand spoken instructions and also offer enterprise security.
"Emilia David
@miyadavid
July 15, 2025 4:34 PM
Mistral released an open-sourced voice model today that could rival paid voice AI, such as those from ElevenLabs and Hume AI, which the company said bridges the gap between proprietary speech recognition models and the more open, yet error-prone versions.
Voxtral, which Mistral will release under an Apache 2.0 license, is available in a 24B parameter version and a 3B variant. The larger model is intended for applications at scale, while the smaller version would work for local and edge use cases.
“Voice was humanity’s first interface—long before writing or typing, it let us share ideas, coordinate work, and build relationships. As digital systems become more capable, voice is returning as our most natural form of human-computer interaction,” Mistral said in a blog post. “Yet today’s systems remain limited—unreliable, proprietary, and too brittle for real-world use. Closing this gap demands tools with exceptional transcription, deep understanding, multilingual fluency, and open, flexible deployment.”
Voxtral is available on Mistral’s API and a transcription-only endpoint on its website. The models are also accessible through Le Chat, Mistral’s chat platform.
The AI Impact Series Returns to San Francisco â August 5
The next phase of AI is here â are you ready? Join leaders from Block, GSK, and SAP for an exclusive look at how autonomous agents are reshaping enterprise workflows â from real-time decision-making to end-to-end automation.
Secure your spot now â space is limited: https://bit.ly/3GuuPLF
Mistral said that speech AI “meant choosing between two trade-offs,” pointing out that some open-source automated speech recognition models often had limited semantic understanding. Still, closed models with strong language understanding come at a high cost.
Bridging the gap
The company said Voxtral “offers state-of-the-art accuracy and native semantic understanding in the open, at less than half the price of comparable APIs.”
Voxtral, at a 32K token context, can listen to and transcribe up to 30 minutes of audio or 40 minutes of audio understanding. It offers summarization, meaning the model can answer questions based on the audio content and generate summaries without switching to a separate mode. Users can trigger functions and API calls based on spoken instructions.
The model is based on Mistral’s Mistral Small 3.1. It supports multiple languages and can automatically detect languages such as English, Spanish, French, Portuguese, Hindi, German, Italian, and Dutch.
Mistral added enterprise features to Voxtral, including private deployment, so that organizations can integrate the model into their own ecosystems. These features also include domain-specific fine-tuning and advanced context and priority access to engineering resources for customers who need help integrating Voxtral into their workflows.
Performance
Speech recognition AI is now available on many platforms today. Users can speak to ChatGPT, and the platform will process spoken instructions similarly to written prompts. Fast food chains like White Castle have deployed SoundHound to their drive-thru services, and ElevenLabs has steadily been improving its multimodal platform. The open-source space also offers powerful options. Nari Labs, a startup, released the open-source speech model Dia in April. However, some of these services can be quite expensive.
Transcription services like Otter and Read.ai can now embed themselves into Zoom meetings, recording, summarizing and even alerting users to actionable items. Many online video meeting platforms offer not just transcription, but also speech AI and agentic AI, with Google Meetings providing the option to take notes for users using Gemini. As a regular user of voice transcription services, I can say firsthand that speech recognition AI is not perfect, but it is improving.
Mistral stated that Voxtral outperformed existing voice models, including OpenAI’s Whisper, Gemini 2.5 Flash and Scribe from ElevenLabs. Voxtral presented fewer word errors compared to Whisper, which is currently considered the best automatic speech recognition model available.
In terms of audio understanding, Voxtral Small is “competitive with GPT-4o-mini and Gemini 2.5 Flash across all tasks, achieving state-of-the-art performance in Speech Translation.”
Since announcing Voxtral, social media users said they have been waiting for an open-source speech model that can match the performance of Whisper.
Mistral said Voxtral will be available through its API at $0.001 per minute. "
https://venturebeat.com/ai/mistrals-voxtral-goes-beyond-transcription-with-summarization-speech-triggered-functions/
#metaglossia_mundus
"The Indian state where people are being beaten up for speaking the wrong language
New rules for schools have ignited street protests and unrest, exposing a nationwide conflict over language and regional identity. Shahana Yasmin reports
Tuesday 15 July 2025 08:29 EDT
Maharashtra, long regarded as one of India’s most cosmopolitan states, has found itself reliving an old tension: a renewed battle over language.
In the last few months, a state government order mandating Hindi as the third language in primary schools has sparked not just political protests but also public violence. A rickshaw driver was assaulted after refusing to speak in the state language Marathi, a shopkeeper was slapped by political activists for using Hindi, and a bank employee was threatened for defending his use of Hindi and English.
At the heart of the unrest lies a 16 April education department resolution making Hindi compulsory as the third language from classes I to V in schools giving instruction in English and Marathi languages.
While state officials framed the policy as a routine implementation of the federal government’s National Education Policy 2020, which encourages education in the regional language, Hindi, and English under the “three-language formula”, critics saw it as part of a broader push by prime minister Narendra Modi’s ruling Bharatiya Janata Party (BJP) to elevate Hindi at the expense of regional languages.
Although the federal push to promote Hindi beyond the “Hindi belt” has faced opposition in many states, the backlash has been especially charged in Maharashtra.
The western state was born out of the Samyukta Maharashtra movement of the 1950s, a mass mobilisation demanding a separate province for Marathi speakers with Mumbai as its capital. The agitation, marked by mass street protests, hunger strikes and violent police crackdowns, led to the deaths of over 100 demonstrators before Maharashtra was finally carved out of the bilingual Bombay State in 1960.
Not surprisingly then, the new policy has struck a discordant note in a state where linguistic identity has long been deeply embedded in the social fabric. Many people view the introduction of compulsory Hindi not as a benign educational reform but as a renewed threat to that hard-won identity.
While the state’s BJP-led government initially downplayed the controversy, the backlash quickly escalated. Opposition parties, including the Shiv Sena UBT, the Congress, and Sharad Pawar’s Nationalist Congress Party, accused the BJP of pushing a “Hindi, Hindu, Hindustan” agenda: shorthand for the Modi government’s concerted bid to promote Hindi and Hindu nationalism at the expense of India’s pluralistic character.
Varsha Gaikwad of the Congress party, a former state education minister, accused the BJP of “working systematically to weaken the Marathi language in its own homeland,” The New Indian Express reported.
Shiv Sena UBT chief Uddhav Thackeray and Maharashtra Navnirman Sena leader Raj Thackeray, longstanding political rivals, also voiced strong opposition to the Hindi mandate. The former referred to the new policy as a “language emergency” and said he “will not allow imposition of any language”. The latter, known for his party’s past campaigns against Hindi-speaking migrants, reiterated his stance that while Maharashtrians were Hindus, they did not accept Hindi imposition.
Maharashtra deputy chief minister Ajit Pawar initially defended the policy. Its critics had “nothing else to do”, he said while arguing that Hindi and English were useful across India. But Marathi, he emphasised, would remain Maharashtra’s primary language.
Federal home minister Amit Shah said last month that Indians should eventually “feel ashamed” to speak English, promoting native tongues instead.
The central government under Narendra Modi has consistently promoted Hindi as the national lingua franca, not only through educational reforms like the three-language policy but also symbolic changes.
Public schemes, for example, now bear Hindi titles like Mera Yuva Bharat, Pradhan Mantri Ujjwala Yojana and Pradhan Mantri Jan Dhan Yojana. The colonial-era Indian Penal Code has been replaced with Bharatiya Nyaya Sanhita.
But as the Maharashtra protests grew louder and violence more visible, chief minister Devendra Fadnavis reversed the mandate and announced it would be optional. He also formed a committee to review the policy’s implementation.
The resistance to “Hindi imposition” has echoed beyond Maharashtra. In Tamil Nadu, chief minister MK Stalin rejected the National Education Policy’s three-language formula and declared that the southern state was “ready for another language war” over what he called the central government’s attempt to impose Hindi.
Deputy chief minister Udhayanidhi Stalin warned the state would “never accept the New Education Policy or the imposition of Hindi in any form” and cast the opposition to it as an “ethnic struggle” to protect Tamil culture.
Fellow southern states Karnataka and Kerala have seen similar opposition. In Bengaluru, the capital of Karnataka, activists vandalised departure boards at the main airport over the omission of Kannada, the state language, while local governments mandated Kannada-first signage in public spaces.
India is one of the most linguistically diverse countries in the world. According to the 2011 census, there are 121 languages in active use and over 1,300 distinct mother tongues, although only 22 enjoy official status under the constitution.
Hindi is the most widely spoken language, used by roughly 43.6 per cent of the population, or 528 million people. Marathi is third with about 83 million speakers and Tamil ranks fifth with about 69 million speakers.
As Maharashtra braces for municipal elections, the language row underscores how swiftly cultural identity can become a political fault line.
What began as an administrative directive on school curriculum has grown into a broader referendum on who defines the cultural contours of a diverse and multilingual India.
https://www.the-independent.com/asia/india/india-maharashtra-hindi-language-imposition-b2788722.html
#metaglossia_mundus
|
