[unable to retrieve full-text content]
Valley Grange and local schools celebrate Dictionary Day observer-me.comMonday, November 15, 2021
Misinterpreted: The underappreciated value of translations - The Fulcrum - Translation
Human translators, as opposed to translation software, typically have a wealth of subject knowledge relevant to the translation assignment
The term “lost in translation” is rather overused. It’s a hot topic that refers to the true essence of what is lost when a text is translated from one language to another, because some of the original meaning, intentions, or plays on words can only be understood in the context of the piece’s original language.
Unfortunately, some might say translation is a dying art form and career due to translation softwares — the most obvious benefits of digital translation are its efficiency and low cost.
While human translators can translate roughly 2,000 words per day (and charge a reasonable fee for this), translation softwares are nearly instantaneous and largely available at no cost. The internet has revolutionized the ability to access, translate and understand texts and documents from all over the world.
The vast majority of people use Google Translate or DeepL to rapidly translate a few words or sentences into another language. However, if they do not speak the language or know the appropriate method to express a given term or spot-check the software’s work, they might wind up with a statement that makes no sense.
While machine translation is frequently a good solution for translating individual words or short phrases, such as when reading a road sign or communicating with a taxi driver in a foreign country, it struggles with complex sentences, and is frequently incapable of translating idioms and colloquialisms, among with other more difficult concepts.
Human translators, as opposed to translation software, typically have a wealth of subject knowledge relevant to the translation assignment they are given, as well as the ability to conduct extensive research around a specific term or topic in order to ensure it is translated using exactly the right terminology in the target language.
Julián Zapata Rojas, a part-time professor and internships coordinator at the U of O’s school of translation and interpretation, has had the privilege to teach 12 different courses, including general and specialized translation from and into English, Spanish and French. As well, he’s taught “non-transfer” courses, such as professional aspects of translation, lexicology and documentation, terminology and translation technologies (at the undergraduate and graduate levels).
Rojas claims that he uses human resources, such as field experts, and material resources, such as dictionaries and computer tools, when teaching his students the most effective strategies to use when translating literature from one language to another.
“Human translators have always used technologies as aids — since the invention of writing systems and writing tools,” said Rojas in an interview.
“The idea of using computers to translate (automatically) or as a tool that aids translators is as old as the idea of computers itself. But other types of technologies also help in different ways: translation memories, electronic parallel corpora, and voice recognition systems are among the many technologies that help translators, say, to speed up the process, to produce better-quality texts.”
When translating, Jordan Gagné (a fourth-year French-English-Spanish translation student)’s professors advise her to use TERMIUM Plus because it’s an official government database, as well as Le grand dictionnaire terminologique, Linguee, DeepL, and both bilingual and monolingual online dictionaries. “[We] are never to use Google Translate. Technology is an amazing tool. It helps fill in the gaps, helps us justify our choices and find better ways to say things idiomatically. Machine translation isn’t 100 per cent accurate, but that doesn’t make it any less useful a tool,” she said.
Rojas argued that when properly learned and used, no technological application should impede the translator from understanding a text in one language and translating it into another.
Furthermore, Rojas elucidates the ways in which the meaning of a word changes when it is translated. “Words that appear to be “the same” in two closely related languages may have different meanings — also known as false cognates or faux amis. Sometimes they mean the opposite. Sometimes they also mean the same, but one of them also means something else. There is inevitably, almost always, some[thing] loss in translation.”
Moreover, some individuals rely heavily on subtitles from their favourite shows to learn the language; Rojas explains how this is an acceptable approach to learning the language.
“Reading closed captions as one [watches] a [program] in the same language has been proposed as a good way to get familiar with that language, learn vocabulary and get acquainted with a particular accent. Now, subtitles in another language, should be used cautiously in foreign language learning.”
Because globalization is a continual process, new content is always being translated. This must be done both linguistically and in conformity with the cultural expectations of the target audience.
Salah Basalamah, an associate professor and chair of the University of Ottawa’s school of translation and interpretation, believes that “to translate is almost always an approximation, especially if one is keen to preserve the idiomaticity of the target language, that is not to imitate the form of the source language.”
“Certainly, emotions are as difficult to transfer as ideas are because in the process feelings and meanings get slightly transformed.”
Basalamah also believes that there’s no other way of translating aside from understanding the cultures and languages in contact — as both a source and a target — as a prior condition.
“You don’t really know its value or its challenges until you put your hands in it and confront the difficulties of going through the process,” said Basalamah.
“It’s also a matter of perception from the outside, but also a matter of presentation and representation of what translation is from the perspective of translators. I believe that despite their ability to use the arcanes of the languages they master, their ability to communicate about their practice and its value has still to be improved and perfected.”
Rojas expresses his disappointment for the undervaluation of translations. “Translators have enabled communications between humans from different tribes, communities, countries and continents for millennia. And yet, even still today, translation is only “visible” when there is a translation error here and a funny translation there,” he said.
Sarah May, a fourth-year French-English-Spanish translation student chose this program because she likes how there are so many different fields in translation, which allows her to deepen her knowledge of the world and understand different perspectives.
“It’s a career where you get to learn something new every day by translating texts on different topics. I also value translation because there are so many translation errors in our everyday lives that may have consequences in people’s lives. These errors should be avoided in the future with educated translators who pay attention to small details,” she said.
May plans to be a subtitle translator after she graduates. She also explains the importance of understanding the culture of the source language in order to understand idioms and nuances, and to be able to find equivalents or similar expressions in the target language. “If the differences between the cultures of target audiences are not taken into account, the translations may not be as idiomatic or geared toward certain aspects of the right culture.”
“I personally watch Spanish telenovelas to improve my Spanish. I listen to the characters speak in Spanish while using subtitles in Spanish (not English) so I can also read it. It has helped me, but I cannot only rely on tv shows since subtitles are not always accurate. I try to read, write, and speak in Spanish too.”
However, May claims that technology can impede us from properly learning a language because people rely too heavily on it without learning basic grammatical rules, as well as acquiring different communication skills.
“True meanings of words get lost in translation because there are several interpretations of the “true” meaning of a single word,” said May. “One translator may have a different interpretation of a text than another depending on their knowledge, background, culture, gender, education, age, sexual orientation, experience, etc.”
For example, sometimes people who need to fill out immigration papers to flee their country don’t speak English. Therefore, it is crucial to provide accurate translations to everyone in order to encourage diversity and equality all over the world.”
Gagné chose to study translation because she’s always loved languages and dialects and how they shape our cultures and communities.
“I think a big challenge that we face in the world today is miscommunication and misunderstanding, not just across languages but among people who speak the same language as well,” she said. “That idea of misunderstanding has always interested me, and translation really delves into learning how to express what you want to communicate clearly.”
Gagné believes that culture plays a huge role in translation. “If you don’t understand the culture of the audience you’re translating for, then that audience might not be able to fully relate to or understand the text they’re reading. Cultural elements like idioms, religions, etiquette among other things might be completely missed in translation if the translator doesn’t have adequate knowledge about the target audience’s culture,” she said.
“If a translation is done incorrectly, you could at best confuse people and at worst put them in danger, such as mistranslating instruction manuals for heavy machinery. The translation is often inseparable from the source text, thus the translator is inseparable from the author, at least in the eyes of most readers.”
“I think that translation is more of a silent hero because people can’t know what they don’t know. If information isn’t available to a certain language community because there is no translation for it, then they have no opportunity to appreciate the translation [itself],” Gagné said.
Sunday, November 14, 2021
Valley Grange and local schools celebrate Dictionary Day - observer-me.com - Dictionary
[unable to retrieve full-text content]
Valley Grange and local schools celebrate Dictionary Day observer-me.comOxford English Dictionary Makes "Vax" 2021's Word of the Year - The Great Courses Daily News - Dictionary
By Jonny Lupsha, Current Events Writer
The most impactful word in the English language this year is “vax”—no surprise there. “Vax,” a shortening of “vaccination,” has been named the Word of the Year by the publishers of the Oxford English Dictionary. It comes from “clipping” a word.
Whether an individual is for or against the COVID-19 vaccination, there’s no doubt that the subject has permeated the everyday lexicon. In all likelihood, any one person has discussed the coronavirus vaccine as the pandemic has continued throughout 2021. Additionally, there’s a good chance we’ve used the shortened word “vax” instead of “vaccination.”
Due to its prominence in language for the year, it’s no surprise that the publishers of the Oxford English Dictionary have made “vax” their Word of the Year for 2021. In her video series The Secret Life of Words: English Words and Their Origins, Dr. Anne Curzan, Arthur F. Thurnau Professor of English at the University of Michigan, explained different types of shortened words.
So That’s What That’s Called
Initialisms are one common way of shortening terms, such as saying “laser” instead of lightwave amplification by stimulated emission of radar. Taking the first letter from each word and making it an intelligible acronym saves time. Another way of abbreviating words is clipping.
“Clipping is just what it sounds like: We’re clipping part of the word off; that could be the front or that could be the back,” Dr. Curzan said. “Off the front, we’d get something like ‘rents’ for parents or ‘do’ from hairdo. ‘Blog’ is a clipping from weblog.”
On the other hand, if we clip the back end of the word off, we get shortened words like “limo” from limousine, “rehab” from rehabilitation, and even “mob.” Surprisingly, mob is a clip of the Latin phrase “mobile vulgus,” meaning “a fickle crowd.” Additionally, sometimes we take the middle of words and leave off the front and the back, such as “fridge” from refrigerator.
“Some of these clippings probably feel slangy to you because they are, but some clippings are now entrenched in the language and no longer feel slangy or even like clippings,” Dr. Curzan said. “‘Flu’ is a clipping from influenza; ‘phone’ from telephone; deli, lab, dorm. These, to us, now feel, I think, well established in the lexicon.”
Back Formation
There are also what are called minor word formation processes. One kind is blending, also known as making a portmanteau word. For example, “smog” is a blend of the words smoke and fog. “Netiquette” is a portmanteau of “Net”—itself a clipping of Internet—and etiquette. Another minor word formation process is back formation.
“This is where we take a word in the language and we reanalyze the parts, and from that we create a new word,” Dr. Curzan said. “We borrowed into English the word ‘beggar’ in the 13th century. You can even see from the spelling, which ends with ‘-ar,’ that it’s not a verb plus a suffix.”
However, the word “beggar” sounds a lot like “baker.” So if a baker bakes, and a taker takes, then a beggar must beg. This line of thinking caused people to backform the word “beg” from “beggar.” Surprisingly, words like this are everywhere.
“English had the word ‘diagnosis’ before we had the verb ‘diagnose’—that was backformed from diagnosis. ‘Television’ gives us the verb ‘televise;’ ‘lazy’ gives us the verb ‘laze,'” Dr. Curzan said. “If you’re lazy, what are you doing? You must be lazing around.”
“Vax” and its derivatives are popular enough to have become Oxford English Dictionary’s Word of the Year 2021, following previous winners like selfie, unfriend, and toxic. Which word will top the charts in 2022?
Edited by Angela Shoemaker, The Great Courses Daily
Google Document Translation Now Generally Available - InfoQ.com - Translation
Google Cloud recently announced the general availability of Document Translation, a new feature of Translation API Advanced that allows formatting of documents to be retained throughout the translation process.
Until now the translation of documents required that text was separated from the layout attributes, with the document’s structure either lost or recreated after the text translation. Sarah Weldon, product manager at Google, explains:
One of the biggest differentiators for Translation API Advanced’s document translation capabilities is the ability to do real-time, synchronous processing for a single file. For example, if you are translating a business document such as HR documentation, online translation provides flexibility for smaller files and provides faster results (...) Meanwhile, batch translation allows customers to translate multiple files into multiple languages in a single request.
The new service lets customers translate documents in over 100 languages and supports formats such as Docx, PPTx, XLSx, and PDF while preserving document formatting. The GA adds right to left language support for PDFs, preservation of font size, font color, font style, and hyperlinks for native PDFs and introduces configurable endpoints to store machine translation processing in the European Union.
To improve the accuracy of the results, Document Translation now supports four different translation approaches: customers can rely on Google’s SOTA translation models, import glossaries for specific terms and phrases defining preferred translations, choose a pre-trained model or build custom translation models with AutoML.
In a separate article, Tristan Li, customer engineer at Google, and Wayne Davis, customer engineering manager at Google, highlight the best practices for translating websites with Translation API. Google is not the only cloud provider offering API for document translation. As recently reported on InfoQ, Microsoft Translator now supports over 100 languages and dialects, covering languages natively spoken by 72% of the world population. AWS offers Amazon Translate to localize websites and applications or translate large volumes of text for analysis.
Rafael Quevedo questions the accuracy of the new API:
The cloud projects are at mercy of the diversity team that designed them. Google Translator can claim that it can translate languages from all types using the existing literature, but can it deal with old style TV phrases? Or slang?
Cloud Translation charges customers by the amount of text processed by the service, starting at 20 USD per million characters. Additional charges apply for the Advanced API calls detectLanguage, translateText, batchTranslateText, translateDocument, and batchTranslateDocument. For example, TranslateDocument costs 0.08 USD for every page processed.
Saturday, November 13, 2021
Google Document Translation Now Generally Available - InfoQ.com - Translation
Google Cloud recently announced the general availability of Document Translation, a new feature of Translation API Advanced that allows formatting of documents to be retained throughout the translation process.
Until now the translation of documents required that text was separated from the layout attributes, with the document’s structure either lost or recreated after the text translation. Sarah Weldon, product manager at Google, explains:
One of the biggest differentiators for Translation API Advanced’s document translation capabilities is the ability to do real-time, synchronous processing for a single file. For example, if you are translating a business document such as HR documentation, online translation provides flexibility for smaller files and provides faster results (...) Meanwhile, batch translation allows customers to translate multiple files into multiple languages in a single request.
The new service lets customers translate documents in over 100 languages and supports formats such as Docx, PPTx, XLSx, and PDF while preserving document formatting. The GA adds right to left language support for PDFs, preservation of font size, font color, font style, and hyperlinks for native PDFs and introduces configurable endpoints to store machine translation processing in the European Union.
To improve the accuracy of the results, Document Translation now supports four different translation approaches: customers can rely on Google’s SOTA translation models, import glossaries for specific terms and phrases defining preferred translations, choose a pre-trained model or build custom translation models with AutoML.
In a separate article, Tristan Li, customer engineer at Google, and Wayne Davis, customer engineering manager at Google, highlight the best practices for translating websites with Translation API. Google is not the only cloud provider offering API for document translation. As recently reported on InfoQ, Microsoft Translator now supports over 100 languages and dialects, covering languages natively spoken by 72% of the world population. AWS offers Amazon Translate to localize websites and applications or translate large volumes of text for analysis.
Rafael Quevedo questions the accuracy of the new API:
The cloud projects are at mercy of the diversity team that designed them. Google Translator can claim that it can translate languages from all types using the existing literature, but can it deal with old style TV phrases? Or slang?
Cloud Translation charges customers by the amount of text processed by the service, starting at 20 USD per million characters. Additional charges apply for the Advanced API calls detectLanguage, translateText, batchTranslateText, translateDocument, and batchTranslateDocument. For example, TranslateDocument costs 0.08 USD for every page processed.
Meta AI Puts A Step Towards Building Universal Translation System - Analytics India Magazine - Translation
What does the curve arrow in the logo of Amazon signify? It simply portrays that one can get A to Z products from a single platform, making your task easy, right? The same will be the case when it comes to the translation system (production of text in one language from another).
To that end, Meta AI announced a new breakthrough and introduced a new multilingual model, outperforming present state-of-the-art bilingual models across 10 out of 14 language pairs, winning the Conference on Machine Translation (WMT) – a prestigious MT competition. The model thus introduced is a step towards building a universal translation system.
The Bottleneck
The ultimate goal of the machine translation (MT) field is to create a universal translation system that will allow everyone to access information and communicate more effectively. However, some of the existing fundamental limitations need to be resolved for that vision to be a reality in the future.
Presently, a lot of modern MT systems rely on bilingual models, which often necessitate a large number of labelled examples for each language pair and task. Unfortunately, there are many languages with limited training data, say, for example, Icelandic and Hausa. The shortcomings make the present approaches redundant. Also, the tremendous complexity makes it difficult for a platform like Facebook to scale present modes to practical applications, where billions of users post every day in hundreds of languages.
Meta to rescue
As per the team at Meta, the MT field needs a shift from bilingual models towards multilingual translation, to be precise, where a single model can translate many language pairs at once. Further, the step to introduce a better multilingual model stands to benefit both – low and high resource languages as they are simple, scalable and efficient.
Image: Meta AI
Last year, Facebook AI (now Meta) introduced M2M-100 as the first multilingual model to translate any pair of 100 languages without relying on English-centric data. The team deployed different mining strategies to prepare a dataset with 7.5 billion sentences for 100 languages as translation data. The researchers employed a variety of scaling strategies to create a global model with 15 billion parameters that include data from related languages and reflect a more diversified script of languages and morphology. The model proves efficient for low resource languages. However, it loses high performance when it comes to high resource languages.
Building on this previous model, the team made three new advancements for:
- large-scale data mining
- scaling model capacity
- more efficient infrastructure
The team built two multilingual systems to train WMT 2021 model — any other language to English and English to any. They utilised parallel data mining techniques such as CCMatrix, which the company claims to be the largest dataset of web-based, high-quality bitexts for training translation models. CCMatrix dataset is more than 50 times larger than the WikiMatrix corpus Facebook provided earlier, with over 4.5 billion parallel phrases in 576 language pairs extracted from the CommonCrawl public dataset snapshots.
Additionally, the model capacity has been raised from 15 million parameters to 52 million. The large scale training was made five times faster than the previous models by adding a GPU memory-saving tool – Fully Sharded Data-Parallel from Meta itself. Further, it is important to note that scaling model size often results in high computational costs. To overcome, the team claims to have used a Transformer architecture with the FeedForward block in every alternate Transformer layer, which is then replaced with a Sparsely Gated Mixture-of-Experts layer with top-2 gating in the encoder and decoder. As a result of the same, only a subset of all the model’s parameters is used per input sequence.
Machine translation has made significant progress in breaking down barriers, but most of it has focused on a small number of commonly spoken languages. Low-resource translation remains MT’s “last mile” dilemma and the subfield’s biggest open challenge today.
Subscribe to our Newsletter
Get the latest updates and relevant offers by sharing your email.Join our Telegram Group. Be part of an engaging community
Subscribe to:
Posts (Atom)