Google hits 1bn translations a day

谷歌每天翻译10亿次

2019-07-24 05:00 insight video interpreting

本文共489个字,阅读需5分钟

阅读模式 切换至中文

The use of online translations is increasing at a rapid rate, with a record one billion translations done a day through Google Translate. According to the Associated Press, the free machine translation software is now helping people decipher the equivalent of one million books’ worth of text each day. Among the 65 languages, it offers translations in are Swahili and Basque. To quickly translate an email or work out what a foreign language website says, it is invaluable. However, as Martha Mendoza, writing for the Associated Press, points out, it is no substitute for a human translator. She points to a study by the Matieland Language Centre, which compared machine and human translations of Afrikaans to English documents. There was a huge disparity between the results, with the machine translations hardly resembling the ones done by professional linguists. Of course, while any human translator will have all the necessary qualifications and experience for the job, and may also be a member of an official industry body such as the ITI, the people behind machine translations rarely speak another language professionally. Instead, machine translators work by recognising patterns in a catalogue of millions of documents to find those already used by a human translator. It then applies educated guesswork in putting together a translation. So, the more examples of a language the machine has, the better the quality of its output. Typically, Google will work out the algorithm for a particular language and not release it until it is as accurate as possible. During the test process, these are called alpha languages as the quality of translated texts they are calculated from is less reliable. Because of this, any client-facing materials such as website content, brochures or signage should always be translated by a professional, human translator. To not do so doesn’t just make you come across as unprofessional, but could also see your message getting lost. Meanwhile, Google Translate might not be the best tool to rely on for completely accurate results, but it has been given the human touch by one heart-warming story. Less than a year ago, Phillip and Niki Smith from Mississippi read about Guan Ya, an orphaned 13-year-old from China, and knew instantly she was their daughter. With only a few months before Guan Ya turned 14 and became ineligible for adoption, the Smiths relied on online translators to make sense of the legal documents they had to process as quickly as possible. It paid off and a few months later Guan Ya joined her new family in the US. Today, she and the Smiths continue to use machine translators to communicate. Not only do they speak different languages, but Guan Ya is also deaf, however, tools like Google Translate help her talk with her adoptive mum. Clearly, no matter how much a free piece of software jumbles what you want to say, there are some sentiments that are never lost in translation.
在线翻译的使用正以快速的速度增长,每天通过谷歌翻译完成的翻译达到创纪录的10亿次。 据美联社报道,免费机器翻译软件现在正在帮助人们每天解读相当于一百万本书价值的文本。在65种语言中,它提供了斯瓦希里语和巴斯克语的翻译。要迅速翻译一封电子邮件或了解一家外语网站所说的话,这是非常宝贵的。 然而,正如美联社撰稿的玛莎·门多萨( MarthaMendoza )所指出的,这并不能替代人工翻译。她提到了 Matieland 语言中心的一项研究,该中心将南非荷兰语的机器翻译和人工翻译与英文文档进行了比较。结果之间存在巨大的差异,机器翻译与专业语言学家的翻译很难相似。 当然,尽管任何翻译人员都具备从事这一工作的所有必要资格和经验,而且也可能是 ITI 等官方行业组织的成员,但机器翻译背后的人很少会专业地说另一种语言。相反,机器翻译工作的方式是在数百万份文件的目录中识别模式,以找到人类翻译已经使用的模式。然后,它运用受过教育的猜测来进行翻译.因此,机器拥有的语言示例越多,其输出的质量就越好。 通常情况下, Google 会为特定的语言制定算法,直到它尽可能精确地发布。在测试过程中,这些被称为 alpha 语言,因为它们计算出来的译文质量不那么可靠。因此,任何面向客户的材料,如网站内容、小册子或标志,都应由专业的人工翻译人员翻译。不这样做不仅会让你觉得自己不专业,还会让你的信息丢失。 与此同时, Google Translate 可能不是最好的工具,它可以完全准确地获得结果,但它已经被一个让人感动的故事赋予了人类。不到一年前,密西西比州的菲利浦和尼基史密斯读到了关亚,一个13岁的中国孤儿,立刻知道她是他们的女儿。在关亚14岁前的几个月里,史密斯夫妇没有资格被采纳,他们依靠在线翻译来理解他们必须尽快处理的法律文件。几个月后,管亚加入了她在美国的新家庭。 今天,她和史密斯夫妇继续使用机器翻译进行交流。他们说的不仅是不同的语言,但管雅也是聋子,然而,像谷歌翻译工具帮助她与她的领养妈妈交谈。显然,无论你想说的是多么免费的软件,在翻译过程中总会有一些情感。

以上中文文本为机器翻译,存在不同程度偏差和错误,请理解并参考英文原文阅读。

阅读原文