{"id":182,"date":"2013-05-31T17:40:43","date_gmt":"2013-05-31T17:40:43","guid":{"rendered":"http:\/\/blogs.uab.cat\/detotunamicaenllengues\/?p=182"},"modified":"2013-05-31T17:42:52","modified_gmt":"2013-05-31T17:42:52","slug":"google-translates-gender-problem","status":"publish","type":"post","link":"https:\/\/blogs.uab.cat\/detotunamicaenllengues\/2013\/05\/31\/google-translates-gender-problem\/","title":{"rendered":"Google Translate’s Gender Problem"},"content":{"rendered":"

Google Translate<\/em>‘s Gender Problem (And Bing Translate<\/em>‘s, And Systran<\/em>‘s…) Google Translate<\/em> and other popular translation platforms often provide unintentionally sexist translations where, among other things, doctors are men and teachers are women. The reason why has to do with a complex mix of algorithms, linguistics, and source materials.<\/p>\n

Google Translate is the world’s most popular web translation platform, but one Stanford University researcher says it doesn’t really understand sex and gender. Londa Schiebinger, who runs Stanford’s\u00a0Gendered Innovations<\/a>\u00a0project, says Google’s choice of source databases causes a statistical bias toward male nouns and verbs in translation. In a paper on\u00a0gender and natural language processing<\/a>, Schiebinger offers convincing evidence that the source texts used with Google’s translation algorithms lead to unintentional sexism.<\/p>\n

<\/a>Machine Translation And Gender<\/h4>\n

In a peer-reviewed case study published in 2013, Schiebinger illustrated that Google Translate has a tendency to turn gender-neutral English words (such as\u00a0the,<\/em>\u00a0or occupational names such asprofessor<\/em>\u00a0and\u00a0doctor<\/em>) into the male form in other languages once the word is translated. However, certain gender-neutral English words are translated into the female form . . . but only when they comply with certain gender stereotypes. For instance, the gender-neutral English terms\u00a0a defendant<\/em>\u00a0and\u00a0a nurse<\/em>\u00a0translate into the German as\u00a0ein Angeklagter<\/em>\u00a0and\u00a0eine Krankenschwester.<\/em>\u00a0Defendant<\/em>translates as male, but\u00a0nurse<\/em>\u00a0auto-translates as female.<\/p>\n

Where Google Translate really trips up, Schiebinger claims, is in the lack of context for gender-neutral words in other languages when translated into English. Schiebinger ran an article about her work in the Spanish-language newspaper\u00a0El Pais<\/em>\u00a0into English through Google Translate and rival platform Systran. Both Google Translate and Systran translated the gender-neutral Spanish words \u201csuyo\u201d and \u201cdice\u201d as \u201chis\u201d and \u201che said,\u201d despite the fact that Schiebinger is female.<\/p>\n

These sorts of words bring up specific issues in Bing Translate, Google Translate, Systran, and other popular machine translation platforms. Google engineers working on Translate told Co.Labs that translation of all words, including gendered ones, is primarily weighed by statistical patterns in translated document pairs found online. Because \u201cdice\u201d can translate as either \u201che said\u201d or \u201cshe said,\u201d Translate’s algorithms look at combinations of \u201cdice\u201d in conjunction with neighboring words to see what the most frequent translations of those combinations are. If \u201cdice\u201d renders more often in the translations Google obtains as \u201che says,\u201d then Translate will usually render it male rather than female. In addition, Google Translate’s team added that their platform only uses individual sentences for context. Gendered nouns or verbs in neighboring sentences aren’t weighed in terms of establishing context.<\/p>\n

<\/a>Source Material, Cultural Context, And Gender<\/h4>\n

Schiebinger told Co.Labs that the project evolved out of a paper written by a student who was working on natural language-processing issues. In July 2012, a workshop was held at Stanford University with outside researchers that was turned, post-peer review, into the machine translation paper.<\/p>\n

Google Translate, which faces the near-impossible goal of accurately translating the world’s languages in real time, has faced gender issues for years. To Google’s credit, Mountain View regularly tweaks Google Translate’s algorithms to fix translation inaccuracies. Language translation algorithms are infamously tricky. Engineers at Google, Bing, Systran, and other firms don’t only have to take grammar into account–they have to take into account context, subtext, implied meanings, cultural quirks, and a million other subjective factors . . . and then turn them into code.<\/p>\n

But, nonetheless, those inaccuracies exist–especially for gender. In one instance last year, users discovered that translating \u201cMen are men, and men should clean the kitchen\u201d\u00a0into German<\/a>\u00a0became \u201cM\u00e4nner sind M\u00e4nner, und Frauen sollten die K\u00fcche sauber\u201d–which means \u201cMen are men and\u00a0women<\/em>\u00a0should clean the kitchen.\u201d Another German-language Google Translate user found\u00a0job bias in multiple languages<\/a>–the gender-netural English language termsFrench teacher, nursery teacher,<\/em>\u00a0and\u00a0cooking teacher<\/em>\u00a0all showed up in Google Translate’s French and German editions in the feminine form, while engineer, doctor, journalist, and president were translated into the male form.<\/p>\n

Nataly Kelly, author of\u00a0Found In Translation: How Languages Shape Our Lives And Transform The World,<\/em><\/a>\u00a0whose firm\u00a0offers language-technology products, told Co.Labs that a male bias in machine translating is extremely common. \u201cIf you’re using a statistical approach to produce the translation, the system will mine all past translations and will serve up the most likely candidate for a “correct” translation based on frequency.\u00a0Given that male pronouns have been over-represented throughout history in most languages and cultures, machine translation tends to reflect this historical gender bias,\u201d Kelly said.<\/p>\n

\u201cThe results can be highly confusing, even inaccurate. For example, in Google Translate, if you translate\u00a0engineer<\/em>\u00a0into Spanish, it comes out as the masculine\u00a0ingeniero,<\/em>\u00a0but if you put in\u00a0female engineer,<\/em>you get\u00a0ingeniero de sexo feminino,<\/em>\u00a0which means something like\u00a0a male engineer of the feminine sex.<\/em>\u00a0This sounds pretty strange in Spanish, to say the least! If you type\u00a0female engineer<\/em>\u00a0into Bing Translate, you get\u00a0ingeniera,<\/em>\u00a0which is technically correct. But still, you have to specify\u00a0female<\/em>\u00a0in order to produce a feminine result. You don’t have to specify\u00a0male engineer<\/em>\u00a0to get\u00a0ingeniero.<\/em>\u00a0You only need to type in\u00a0engineer.<\/em>\u00a0[There is] an inherent gender bias in most machine translation systems.”<\/p>\n

<\/a>The Statistical Nature Of The Corpus<\/h4>\n

The reason why this happens is statistical. In every language that Google Translate operates in, algorithms process meaning, grammar, and context through a massive number of previously uploaded documents. These documents, which vary from language to language, determine how Google Translate actually works. If source material used for translations has an aggregated bias in terms of one gender being preferred over another, that will be reflected in translations received by users.<\/p>\n

When a user on Google Groups questioned\u00a0male gender bias in Hebrew translations<\/a>\u00a0in 2010, Google’s Xi Cheng noted that \u201cGoogle Translate is fully automated by machine; no one is explicitly imposing any rules; the translation is generated according to the statistical nature of the\u00a0corpus we have.\u201d<\/p>\n

According to Schiebinger, machine translation systems such as Google Translate use two separate kinds of corpuses. A \u201cparallel corpus\u201d with text in one language that is used to compare a translation in another language, while a large monolingual corpus in the target language being translated to is used to determine grammar and word placement. If masculine or feminine forms of words are systematically favored in the corpus used, it leads the algorithm to translate in favor of that particular gender.<\/p>\n

Machine translation ultimately depends on translators and linguists giving context to both algorithms and the source material they use. Google Translate, Bing Translate, and Systran all do a stunning job of providing instant translations in a staggering array of languages. The challenge for translation platform developers is how to further refine their product and increase accuracy–something we’ll see more of in the future.<\/p>\n

http:\/\/www.fastcolabs.com\/3010223\/google-translates-gender-problem-and-bing-translates-and-systrans<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"

Google Translate‘s Gender Problem (And Bing Translate‘s, And Systran‘s…) Google Translate and other popular translation platforms often provide unintentionally sexist translations where, among other things, doctors are men and teachers are women. The reason why has to do with a complex mix of algorithms, linguistics, and source materials. Google Translate is the world’s most popular […]<\/p>\n","protected":false},"author":594,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":[],"categories":[3],"tags":[],"_links":{"self":[{"href":"https:\/\/blogs.uab.cat\/detotunamicaenllengues\/wp-json\/wp\/v2\/posts\/182"}],"collection":[{"href":"https:\/\/blogs.uab.cat\/detotunamicaenllengues\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/blogs.uab.cat\/detotunamicaenllengues\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/blogs.uab.cat\/detotunamicaenllengues\/wp-json\/wp\/v2\/users\/594"}],"replies":[{"embeddable":true,"href":"https:\/\/blogs.uab.cat\/detotunamicaenllengues\/wp-json\/wp\/v2\/comments?post=182"}],"version-history":[{"count":2,"href":"https:\/\/blogs.uab.cat\/detotunamicaenllengues\/wp-json\/wp\/v2\/posts\/182\/revisions"}],"predecessor-version":[{"id":194,"href":"https:\/\/blogs.uab.cat\/detotunamicaenllengues\/wp-json\/wp\/v2\/posts\/182\/revisions\/194"}],"wp:attachment":[{"href":"https:\/\/blogs.uab.cat\/detotunamicaenllengues\/wp-json\/wp\/v2\/media?parent=182"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/blogs.uab.cat\/detotunamicaenllengues\/wp-json\/wp\/v2\/categories?post=182"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/blogs.uab.cat\/detotunamicaenllengues\/wp-json\/wp\/v2\/tags?post=182"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}