![\"When](\"https:\/\/www.textform.com\/tf2019\/wp-content\/uploads\/2020\/09\/machine_translation_mistakes_blogpost.jpg\")
<\/p>\n<\/p>\n
During our recent trawling of localization news and events<\/span>\u00a0<\/span>here<\/span>\u00a0in the marketing department<\/span>\u00a0at\u00a0<\/span>text<\/span>&form<\/span>\u00a0some<\/span>\u00a0machine translation<\/span>\u00a0(MT)<\/span>\u00a0faux pas have\u00a0<\/span>really\u00a0<\/span>caught our eye<\/span>.\u00a0<\/span>With the use\u00a0<\/span>of automatic MT<\/span>\u00a0becoming more popular, easy to use and ultimately more acceptable, we wanted to take<\/span>\u00a0a look at these<\/span>\u00a0prominent<\/span>\u00a0machine translation\u00a0<\/span>errors<\/span>\u00a0and\u00a0<\/span>gauge<\/span>\u00a0just how damaging these\u00a0<\/span>mistakes can\u00a0<\/span>be<\/span>.<\/span>\u00a0<\/span>\u00a0<\/span><\/p>\n But first the funny side.\u00a0<\/span>If you have\u00a0<\/span>come across the comedy chat show from the US,\u00a0<\/span>\u2018<\/span>The Tonight Show Starring Jimmy Fallon<\/span>\u2019<\/span>, you will probably have seen<\/span>\u00a0the regular<\/span>\u00a0segment called \u2018<\/span>Google Translates Songs<\/span><\/a>\u2019. They invite a famous singer<\/span>\u00a0or actor (and we are talking big names here lik<\/span>e Camila Cabello and Idris Elba<\/span>)<\/span>\u00a0to\u00a0<\/span>sing a well-known\u00a0<\/span>pop song whose lyrics have\u00a0<\/span>been fed<\/span>\u00a0into Google Translate, translated to Russian and then back into English. The results are predictably hilarious<\/span>, if somewhat contrived<\/span>\u00a0and the long-running nature of this\u00a0<\/span>feature suggests that translation mistakes are endlessly funny<\/span>.\u00a0<\/span>Great then<\/span>,<\/span>\u00a0when laughs are the goal, b<\/span>ut what happens when machine translation<\/span>\u00a0produces inaccurate results which are not destined for a comedy sketch?<\/span>\u00a0<\/span><\/p>\n In<\/span>\u00a0July<\/span>,<\/span>\u00a0<\/span>Facebook<\/span>\u00a0Thailand<\/span>\u00a0got itself into\u00a0<\/span>serious trouble<\/span><\/a>\u00a0when it<\/span>\u00a0<\/span>published a post<\/span>\u00a0<\/span>about the Tha<\/span>i King\u2019s birthday which the US tech giant\u2019s machine translation<\/span>\u00a0had<\/span>\u00a0<\/span>automatically<\/span>\u00a0transformed<\/span>\u00a0<\/span>into Thai from English<\/span>. The\u00a0<\/span>post\u00a0<\/span>caused such offense\u00a0<\/span>to the Thai monarchy that Facebook\u00a0<\/span>immediately\u00a0<\/span>suspended its auto-translate function<\/span>\u00a0in Thailand<\/span>, promised a review of its machine translation model and issued a sincere apology.\u00a0<\/span>The Crime Technology D<\/span>ivision<\/span>\u00a0of the Thai police force<\/span>\u00a0was<\/span>\u00a0called in to investigate and the public broadcaster<\/span>,<\/span>\u00a0Thai PBS also filed a complaint against the American tech company after its viewers complained\u00a0<\/span>about\u00a0<\/span>the offensive<\/span>\u00a0subtitles on its social media pages. Qu<\/span>ite a stir then<\/span>\u00a0and one that Facebook would undoubtedly have preferred to avoid<\/span>.<\/span>\u00a0<\/span><\/p>\n Previously in\u00a0<\/span>January this year, Facebook\u00a0<\/span>als<\/span>o blamed\u00a0<\/span>technical issues<\/span><\/a>\u00a0when it<\/span>\u00a0automatic<\/span>ally<\/span>\u00a0<\/span>translated the name<\/span>\u00a0of the Chinese\u00a0<\/span>President – on an official visit to Myanmar – into an insulting and\u00a0<\/span>disrespectful English word, this time from Burmese.<\/span>\u00a0<\/span>Although perhaps amusing on the face of it, we won\u2019<\/span>t be repeating it here<\/span>\u00a0and it seems pretty certain that the Chinese weren\u2019t laughing along.<\/span>\u00a0<\/span><\/p>\n Even more worryingly,\u00a0<\/span>US immigration officials<\/span>\u00a0were last year\u00a0<\/span>advised<\/span><\/a>\u00a0to use Google Translate or other similar free, online machine translation<\/span>\u00a0tools<\/span>\u00a0to convert social\u00a0<\/span>media posts written by refugees<\/span>\u00a0into American English and use these to assess individuals\u2019 suitability for\u00a0<\/span>entry in to the US<\/span>.\u00a0<\/span>Even Google itself\u00a0<\/span>advises against<\/span><\/a>\u00a0using its MT<\/span>\u00a0for complex tasks and it would appear to be a gross misjudgement on the part of the US government to assume that this was an acceptable\u00a0<\/span>element in<\/span>\u00a0the\u00a0<\/span>vetting process.\u00a0<\/span>\u00a0<\/span><\/p>\n Social media\u2019s suitability<\/span>\u00a0for machine translation is\u00a0<\/span>to say the least, debatable. The very nature of Faceb<\/span>ook, Twitter and the like is<\/span>\u00a0<\/span>one of\u00a0<\/span>relaxe<\/span>d,<\/span>\u00a0informal\u00a0<\/span>and\u00a0<\/span>fashionable language<\/span>\u00a0<\/span>which<\/span>\u00a0is often<\/span>\u00a0colloquial and<\/span>\u00a0changes<\/span>\u00a0<\/span>constantly<\/span>. Everything that MT finds challenging.<\/span>\u00a0\u00a0<\/span>\u00a0<\/span><\/p>\n Just how quickly inaccurate translation can spread in this age of social media has been seen recently in\u00a0<\/span>Mexico<\/span>. Alt<\/span>hough whether or not this was machine translation<\/span>\u00a0remains unconfirmed, the mistranslations on the Visitmexico.com website this August were trending\u00a0<\/span>on Twitter and caused the English version of the website to be shut down completely. Not ideal during a pandemic when your country is trying to restore confidence\u00a0<\/span>and attract\u00a0<\/span>visitors<\/span>.\u00a0<\/span>\u00a0<\/span><\/p>\n But away from social media MT errors can be equally as damaging.\u00a0<\/span>I<\/span>n July this year a\u00a0<\/span>Welsh member of the UK Parliament<\/span>\u00a0was also obliged to\u00a0<\/span>apologize for\u00a0<\/span>inaccuracies<\/span><\/a>\u00a0on her website after machine translation<\/span>\u00a0(Microsoft this time)<\/span>\u00a0was\u00a0<\/span>used<\/span>\u00a0to transform English into Welsh.\u00a0<\/span>Her constituents were offered medical surgery rather than the more usual MP\u2019s surgery, which is a meeting to discuss any issues, and the site also referred to an industrial plant as a plant of the vegetable variety.<\/span>\u00a0<\/span>Hardly reassuring for the MP\u2019s Welsh speaking voters seeking reassurance in their language of choice.\u00a0<\/span>Thanks, a cabbage for us please.<\/span>\u00a0<\/span><\/p>\n Away<\/span>\u00a0from\u00a0<\/span>the pandemic where MT has been shown to p<\/span>rovide\u00a0<\/span>vital health information<\/span><\/a>\u00a0in minority languages, Doctors have<\/span>\u00a0warned of the over-reliance on automatic translation in a medical context.<\/span>\u00a0<\/span>Research carried out in 2017<\/span><\/a>\u00a0assessed the use of Google Translate for Chinese and Spanish discharge instructions<\/span>\u00a0from emergency departments<\/span>\u00a0in the US and concluded that \u2018<\/span>2% of Spanish and 8% of Chinese sentence translations\u2014had the potential to cause clinically significant harm\u2019<\/span>. No laughing\u00a0<\/span>matter then, a<\/span>gain underlining the constraints<\/span>\u00a0of these free forms of MT.<\/span>\u00a0<\/span><\/p>\n All these examples illustrate an over confidence in the capacity of machine translation\u00a0<\/span>to understand the context and complexity of human language\u00a0<\/span>and often point to\u00a0<\/span>a misreading<\/span>\u00a0of its limitations.\u00a0<\/span>The improvements and\u00a0<\/span>advances in mac<\/span>h<\/span>ine t<\/span>ranslation<\/span><\/a>\u00a0over the last five years have been wel<\/span>l documented and the rise of\u00a0<\/span>NLP<\/span>\u00a0has been nothing short of revolutionary, but there is a long way to go before machines can translate sensitive or idiomatic language like humans can. Certainly blaming the technology is just too easy. Blame the humans behind the technology and the often misguided refusal to include humanity in the translation process.<\/span>\u00a0<\/span><\/p>\n by Caroline Strappazzon<\/em>, Marketing Specialist at text&form<\/p>\n During our recent trawling of localization news and events\u00a0here\u00a0in the marketing department\u00a0at\u00a0text&form\u00a0some\u00a0machine translation\u00a0(MT)\u00a0faux pas have\u00a0really\u00a0caught our eye.\u00a0With the use\u00a0of automatic MT\u00a0becoming more popular, easy to use and ultimately more acceptable, we wanted to take\u00a0a look at these\u00a0prominent\u00a0machine translation\u00a0errors\u00a0and\u00a0gauge\u00a0just how damaging these\u00a0mistakes can\u00a0be.\u00a0\u00a0 But first the funny side.\u00a0If you have\u00a0come across the comedy chat show from […]<\/p>\n","protected":false},"author":22,"featured_media":13388,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"content-type":"","footnotes":""},"categories":[339,797,737],"tags":[382,915,388,486],"_links":{"self":[{"href":"https:\/\/www.textform.com\/en\/wp-json\/wp\/v2\/posts\/16740"}],"collection":[{"href":"https:\/\/www.textform.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.textform.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.textform.com\/en\/wp-json\/wp\/v2\/users\/22"}],"replies":[{"embeddable":true,"href":"https:\/\/www.textform.com\/en\/wp-json\/wp\/v2\/comments?post=16740"}],"version-history":[{"count":0,"href":"https:\/\/www.textform.com\/en\/wp-json\/wp\/v2\/posts\/16740\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.textform.com\/en\/wp-json\/wp\/v2\/media\/13388"}],"wp:attachment":[{"href":"https:\/\/www.textform.com\/en\/wp-json\/wp\/v2\/media?parent=16740"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.textform.com\/en\/wp-json\/wp\/v2\/categories?post=16740"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.textform.com\/en\/wp-json\/wp\/v2\/tags?post=16740"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}
\nCaroline joined text&form<\/a> in the spring of 2018 and works as a copy writer and content marketing specialist. She has a background in translation (French to English) and an interest in all things linguistic. She is based between Sheffield<\/a> in the UK and the French Alps.<\/div>\n","protected":false},"excerpt":{"rendered":"