Friday, September 2, 2022

Yasmin Moslem on Using Large Language Models to Custom Train Machine Translation - Slator - Translation

Machine Translation (MT) Researcher, Yasmin Moslem, joins SlatorPod to talk about her research on Domain-Specific Text Generation for Machine Translation — a project she conducted with Rejwanul Haque, John D. Kelleher, and Andy Way at the Adapt Center in Dublin.

Yasmin shares her experience working as a translator, discovering translation productivity (CAT) tools, and experimenting with translation memory to improve MT. She breaks down the paper’s approach to domain-specific MT training using back-translation for data augmentation.

She discusses how some LSPs are already implementing this approach in real-life, customizing it for different use cases. She explains why they used a combination of BLEU, Comet, and other quality evaluation frameworks as well as human evaluation to rate machine translation quality.

Yasmin concludes the podcast with her advice for those in the core industry looking to enter the machine translation space, from the spiral learning process to reading research papers.

First up, Florian and Esther discuss the language industry news of the week, including how a streaming platform used propriety machine dubbing technology for its film offerings in the first quarter of 2022.

Over in London, TransPerfect acquired a virtual data room (VDR) tech company to proactively address the VDR market. In transcription news, VIQ Solutions’ shares dipped by 20% despite reporting strong, half-year revenue growth of 45% year on year. Meanwhile, multilingual captioning provider Ai-Media turns EBITDA-profitable as a 2021 acquisition drives revenue growth.

Adblock test (Why?)

Wednesday, August 31, 2022

A new online dictionary is sharing the Ho-Chunk language with the public. - Literary Hub - Dictionary

Corinne Segal

August 30, 2022, 12:59pm

Years of efforts to preserve Hoocąk, the language spoken by the Ho-Chunk Nation, have yielded a new online dictionary that gives the public access to thousands of words and phrases recorded from native speakers.

Sarah Volpenhein reports for the Milwaukee Journal Sentinel that the Hoocąk-English dictionary launched last month under a team led by Henning Garvin, who studied linguistics at the University of Wisconsin-Madison and whose father is a native speaker of Hoocąk. It collects more than 11,000 words and phrases, including recordings from elders, and that number is likely to grow.

Hundreds of years of violent, discriminatory policies by the US government have left many Indigenous languages in jeopardy. Indigenous children were long abused for speaking their languages in school, from the residential school system—a project to forcibly erase the culture of Indigenous tribes by kidnapping children and incarcerating them in boarding schools, where they would be punished for displaying their culture—to the public school system.

With between 50 and 200 native speakers of Hoocąk remaining (according to various estimates), the hope is that the dictionary can serve as a tool for their descendants. “We consider it a gift from all of these people to future generations,” Adrienne Thunder, program manager of the Ho-Chunk Nation’s language division, told Volpenhein. “While we may not have them for a long time into the future, we will have this gift that they left us.”

Adblock test (Why?)

Florida schools delay dictionary donation to check for indoctrination - Palm Beach Post - Dictionary

Tuesday, August 30, 2022

Parishioners reclaim Ojibwe language through hymn translation - Catholic News Service - Translation

ST. PAUL, Minn. (CNS) — Holding a wooden flute, Larry Martin stood during a recent Mass and welcomed the congregation to join the responsorial psalm. He began: “Aw ge-chi-twaaa-wen-daa-go-zid, Gi-gi-zhe-ma-ni-doo-mi-nann.”

The language was Ojibwe, and the words translated to “Our God is one who is glorious,” taken from Psalm 19.

Martin, a 79-year-old director emeritus of American Indian Studies at the University of Wisconsin, Eau Claire, worked with another language expert to convert the English to Ojibwe, the traditional language of many of the American Indian Catholics who worship at Gichitwaa Kateri in south Minneapolis, Martin’s parish.

Most of them can’t speak their ancestors’ language, but it’s meaningful to pray in it, he said. “It helps them give voice to their Indian identity,” he said.

Gichitwaa Kateri is home of the Archdiocese of St. Paul and Minneapolis’ Office of Indian Ministry. Since 2018, Martin and fellow parishioner Rick Gresczyk have translated into Ojibwe most of the responsorial psalms used in the church’s three-year Sunday Mass cycle. Their work built on a project they began years earlier to translate popular hymns such as “Ode to Joy,” “Hail, Holy Queen” and “How Can I Keep from Singing?”

Their accomplishments caught the attention of Catholics planning Pope Francis’ visit to Canada this July. At the request of the visit’s organizers, Martin submitted a few hymns for consideration, including “Wezhitooyan Gakina Go” and “Hymn for Kateri Tekakwitha.”

The first, an Ojibwe creation song Martin and Gresczyk composed, was inspired by three sources: an Old English creation hymn, an Ojibwe creation story and a hymn attributed to Pope St. Gregory the Great.

The second hymn was created by Father Jan Michael Joncas, a well-known liturgical composer and recently retired priest of the archdiocese. In 2012, he collaborated with the Gitchitwaa Kateri community to craft a hymn to celebrate the canonization of the parish’s namesake.

Although the hymns he submitted were not ultimately used during the papal visit, Martin thinks that might be partly due to regional difference: The Ojibwe dialect spoken in Canada differs from the dialect Martin and Gresczyk use, he said. He feels it was an honor for the hymns to even be considered.

In addition to translating popular Catholic hymns and psalms, the two men have set to music Ojibwe-language prayers of Bishop Frederic Baraga, the first bishop of Marquette, Michigan.

Like elements of Pope Francis’ Canadian pilgrimage, Martin and Gresczyk’s translation initiative is tied to culture reclamation efforts underway in the U.S. and Canada, in response to the Indian boarding school era, where American Indian and Indigenous children were removed from their homes and sent to government-funded schools, some run by Catholic religious orders and dioceses, where they were often not allowed to speak their native languages or express their cultures.

“The church is responsible for damage to language, so we thought we should do something about bringing it back,” said Martin, who is Lac Courte Oreilles Ojibwe.

For the translations, Martin credits Gresczyk’s deep knowledge of Ojibwe. Martin doesn’t consider himself fluent, but says he can tweak grammar and align Gresczyk’s translations with the chosen melodies.

Gresczyk now lives in northern Minnesota, so the two mostly collaborate by phone.

Shawn Phillips, director of the archdiocese’s Office of Indian Ministry and pastoral minister at Gichitwaa Kateri, said the translations help parishioners pray and learn more about their culture and heritage. He hopes one day there will be a similar effort to translate prayers into Dakota, so both of the primary Native American cultures in Minnesota would be represented, he said.

The translation effort is important, Phillips said, because “God will speak to them in their own language.”

“That was the Pentecost message,” he said. “It wasn’t that the Gospel be in Greek or in Roman, but … all of these people could understand it. It’s that God cares about us and speaks to us in our own language and knows us intimately.”

– – –

Wiering is editor of The Catholic Spirit, newspaper of the Archdiocese of St. Paul and Minneapolis.

Adblock test (Why?)