News about people using machine translation in everyday life
9 July 2021: The government of Estonia announces it will develop a translation platform that includes machine translation. It is aimed mainly at translators who work on government documents, but will also be available for the private sector.
17 April 2021: Google Explains Why App Can’t Translate Most Native American Languages. Mostly it has to do with a lack of digital linguistic resources, a problem for many of the world’s languages.
18 November 2020: ‘Hello work’ or job centre? Language experts spell trouble for Japan’s mangled English
31 December 2019: Businesses (evidently) using raw MT in contexts where they should not: The embarrassing Welsh language fails which show why you really shouldn’t rely on Google Translate. The good news is that the language authorities in Wales are doing something about it, from the article: “In a bid to prevent further blunders, the Welsh Language Commissioner offers a free proofreading service for large businesses which allows them to send text to be checked before publication.”
5 November 2019: Official information and machine translation
Links to 2 different things recently caught my eye. In the first article, a government ministry in Cambodia banned property developers from putting up signs and trademarks that used translations from Google Translate instead of human translations for signs in the Khmer language. This was prompted by officials’ concern on the effects of poor language on society and the language. The second was a requirement in the state of Maryland to translate information on websites through a method that comes “free-of-charge”, meaning that they should embed an MT feature in the websites. The purpose is to offer equal access to populations with limited proficiency in English. I see a few differences between these 2 things. The first is trying to stop the practice of using raw MT to translate into the official language of the country. Besides a valid demand that people who want to do business in the country should work with a professional level of the language of that country, it is definitely a worst practice to use raw MT for signs and advertising. The Maryland case involves an effort to provide official information into a large number of other languages to facilitate people’s understanding of official texts. This task could never be carried out by professional translators because there are too many texts and too many languages involved. The law also states that content for which a mistranslation could cause could lead to a denial of services or benefits is exempt from the requirement. A different part of the law requires that governmental organizations have to provide translations of vital documents into any language spoken by at least 3% of the population served by the organization. It doesn’t specify if those have to be human translations though. Official guidelines on the use of MT interest me. First of all, any kind of guideline supporting or forbidding the use of MT is a sign of how widespread it is. Second, it tells us something about how legitimate the use of MT is perceived to be for specific contexts, and building the knowledge on where and where not to use MT helps build what Lynne Bowker and Jairo Buitrago Ciro termed ‘MT literacy’ (@LiteracyMachine in Twitter).
Other news items
September 2019: ProPublica reveals that U.S. immigration officials use machine translation to review non-English social media posts of refugees. It is not clear how much weight social media posts are given in settlement processes, but the idea of relying entirely on MT for review leaves me with a lot of questions.
July 2019: Lynne Bowker is interviewed about a very important topic, MT literacy. Lynne is one of the top researchers of users of raw, unedited (or ‘gist’) machine translation.
June 2019: Well-known publisher Springer-Verlag published a guide book for nurses as a machine translation. The Austrian professional association for interpreters and translators, UNIVERSITAS, awarded it the most catastrophic translation of 2019. Link to the original article and the machine translation by Google. No irony intended here, using MT to read articles is a good use of MT. Using it to save costs in officially published textbooks is not.9 May 2019: Finland has seen another round of scams involving the sale of classic novels that were translated by machine, not humans. Finnish magazine Seura talked to the chair of the Finnish Association of Translators and Interpreters, Heikki Karjalainen, who listed these things to look out for when buying books online:
- A noticeably cheaper price than similar novels.
- The name of the translator is not mentioned.
- The publisher should be named and it’s helpful if it’s a publisher that is in the market of the language in question
- If ordering from Amazon, be careful with marked as ‘CreateSpace’. Those can be published by anyone. In Amazon, use the ‘Look inside’ to check out the language inside the book
1 April 2019 (but not an April Fool’s Day joke): another #MTinallthewrongplaces. This recent spotting of texts that seem to be machine translated and shouldn’t be. This time they are campaign material for the 2020 US presidential election.
25 February 2019: Another article about doctors using Google Translate in the Daily Mail, though it’s taken mostly from this article in the University of California San Francisco’s news center. Brings up many questions about when and where MT should be used.
8 November 2018: Here’s a fun one! And I’ve been meaning to start a hashtag about bad uses for MT so this gave me an excuse to start #MTinallthewrongplaces. Brought to you by Moravia: Eight of the Most Bizarre Machine Translation Fails of 2018
29 October 2018: A short time ago this article Doctors are cautioned against using Google Translate in consultations was going around Twitter. The article was published by The BMJ (formerly the British Medical Journal) and advised doctors to use extreme caution in using Google Translate in clinical practice. I dug further, to the study that eventually sparked this caution: Doctors choose Google Translate to communicate with patients because of easy access.
25 September 2018, this BBC News article is fairly down-to-earth, outlining some of the problems and also solutions in MT. It also includes an interesting finding from a British Council survey: 2/3 of 16-34 year-olds use translation apps while traveling: How translation apps iron out embarrassing gaffes
A few Microsoft articles that I found interesting:
- 17 Sept 2018: A case study of a school district that uses Microsoft Translator to aid communication between faculty and parents who speak other languages: How Microsoft Translator empowered our district leaders, students and parents
- 27 Sept 2018: Microsoft’s intelligent mobile phone keyboard SwiftKey translates as you text
TAUS’s eLearning blogs, 12 September 2018: My own blog on users of FAUT (fully automated useful translation) MT
Tech Times, 17 June 2018: Judge Says That Google Translate Can’t Be Used To Authorize Police Search
Popula: Machine Translation Work is being done, but a machine is doing it; I have a translation, but where is the translator?
BBC News, 30 July 2018: Google Translate serves up ‘scummy Welsh’ translation
edscoop: Schools find a classroom translation tool has benefits for reaching parents, too
A few different stories of MT use during the World Cup:
- The Guardian: Google Translate: the unlikely World Cup hero breaking barriers for fans
- New York Times: The Google Translate World Cup
Moravia blog on chat+MT: What Facebook Messenger Chat Translation Means for Global Brands
LinkedIn Engineering: Dynamic Machine Translation in the LinkedIn Feed
Common Sense Advisory: When Support Becomes a Multilingual Conversation
Jake Miller on using Google Translate in Excel: Translate in Google Sheets
Pingback: Gary, developer of MT systems | Machine Translation Stories
What a great list of different MT news articles! Another interesting development in MT I came across: https://blog.csoftintl.com/mt-vs-gpt-3-contrasting-profiles-in-language-ai/