机器翻译 | 非文学文本德英语对使用 DeepL Translator 的翻译效果探究

作者:罗冰雁(北京外国语大学2019级德语口译研究生)

编辑:陈杲、余恬蕙

一. 导语

德语和英语在语言、句式等各方面存在很大差异,翻译中转换结构以准确传达信息也存在一定难度。无线电设备指令属于非文学的信息性文本,其中有着非常多的专业术语,译者在翻译之前还需要掌握一定的技术背景信息。目前为止,对于德语中的术语处理,借助谷歌翻译或是其他机器翻译得出的结果经常不够准确。

现如今,5G 通信已经是全世界都在讨论的话题。随着 5G 时代的到来,无线电技术更是推动了如物联网、智慧城市、自动驾驶等产业的蓬勃发展。2014年5月,欧盟官方公报正式公布无线电设备指令 2014/53/EG,该指令已取代无线电与电信终端设备指令(R&TTE 指令)。自2017年6月13日起,所有无线电设备必须符合新指令的要求。欧盟指令的官网上官方提供了不同语言版本的指令,以便于各个欧盟成员国以其为蓝本制定出各自国内适用的法律。本文选取了欧盟无线电设备指令作为研究对象。

本文拟借助 DeepL Translator 这一机器翻译软件对非文学文本着手:即从德语译入英文,观察该软件对于非文学文本的德英语对翻译效果。

二. DeepL Translator 简介

众所周知,大量已出版的数据库和研究论文都使用英语编写,并且迄今为止尚未翻译。相应的,许多原始的德语参考文献也没有译本。德国是一个老牌工业大国,在医学和法律、机械制造等领域都是世界上的翘楚,而非德语使用者想要了解这些领域内未经翻译的德语文献资料,需要昂贵的专业翻译服务,或者必须花费很长时间习得德语阅读的能力。因此,有学者对于通过机器翻译阅读德语文本做出了研究。

“To read these resources, either costly professional translation services are needed or a reading knowledge of German has to be acquired. A convenient approach is to utilize machine translation for reading German texts; however, there is a question of translation reliability. In this work, several different platforms that employ neural network for machine translation (NMT) were tested for translation capability of scientific German. From a preliminary survey, Google Translate and DeepL were finalized for further studies (German to English). Excerpts from German documents spanning more than a century have been carefully chosen from standard works. DeepL Translator and Google Translate were found to be reliable for converting German scientific literature into English for a wide variety of technical passages.” (Sonia Zulfiqar, M. Farooq Wahab, Muhammad Ilyas Sarwar, Ingo Lieberwirth.Is Machine Translation a Reliable Tool for Reading German Scientific Databases and Research Articles?  October 2018, Journal of Chemical Information and Modeling 58(11))

这项研究结果显示,DeepL Translator 和 Google Translate 在将德国科学文献转换为英语的各种技术文章方面是可靠的。

本文中将借助 DeepL Translator 为工具,对于德英语对的翻译进行分析。与 Google Translate 相比,DeepL Translator 的全球知名度并不高。DeepL 是一家为语言开发人工智能的深度学习公司,位于德国科隆,成立于2009年,当时名为 Linguee,并推出了第一个用于翻译的互联网搜索引擎。2017年,该公司推出了基于机器翻译系统的 DeepL Translator。在2019 年 9 月份,DeepL 获得了首届德国人工智能奖(Deutscher KI-Preis)的荣誉人工智能奖。在其官网上,写着这么一句话:DeepL 一直致力于推动人工智能技术的发展。其目标是将人工智能的前景转化为更光明、更互联的未来现实。在2020 年 3 月份时,DeepL 也增加了中文和日文的机器翻译。由此可以看出,这家人工智能公司在机器翻译方面展现出了惊人的潜力。

三. 实际操作过程

首先,我们在 memoQ 界面中新建项目,项目名称是待翻译的文本和源语和目的语的缩写。在选择语言时,考虑到是欧盟范围内的指令,此项指令于2014年发布,彼时英国还未脱欧,而英美两国在诸多用词上存在差异,因此选定英语(英国)。

翻译文本界面显示
翻译文本界面显示

在建立新项目时,笔者建立了翻译记忆库,但是没有使用术语库,因为此次操作是为了检验机器翻译的呈现效果,所以并未人为添加术语并修改译文。但是一般在使用 memoQ 时正式的翻译过程中,是需要建立术语库的。


导出双语,以便借助 GT4T 批量翻译

在 GT4T 的界面中,我们可以看到 DeepL Translator,显示的是 Pro 版本。
GT4T 待翻译界面

将文件导入 GT4T 并确定相应的源语和目标语,待 GT4T 翻译完成以后,把重新覆盖后的双语文件导入至 memoQ 的项目文档中,即可查看对照译文并在此基础上进一步修改。

GT4T 翻译完成以后的双语对比界面


提示错误

打开文章后,可以看到在第三个句段中,右侧有闪电符号提示错误:原文中的 EG 变成了 EC,然而这是正确的翻译,在英文的原版指令中,用的是 EC (European Community),而德语版本中是 EG (Europäische Gemeinschaft)。因此我们可以得出两点:

  1. DeepL Translator 在这里翻译正确,没有直接将指令名称复制粘贴,而是根据上下文及背景提供正确的译文,即便是缩写中的一个字母。
  2. memoQ 能够识别出潜在的风险,做出相对应的提示。
欧盟无线电设备指令英文版原文

我们可以粗略地对机翻译文和指令原文进行一个比对,能够看到 DeepL 翻译器呈现的德英翻译效果良好。在确认每一个句段的翻译以后,可以将译文导出。

四. 分析译文

在导出了译文以后,笔者整理出一些有特色的例子,整理成如下表格。左边栏是从德语原文通过 DeepL Translator 译成的英文文本,而右边栏则是欧盟官网上提供的英文版指令原文。

译文指令原文
“making available on the market’ means any supply of a product for distribution, consumption or use on the Union market in the course of a business, whether in return for payment or free of charge;‘making available on the market’ means any supply of radio equipment for distribution, consumption or use on the Union market in the course of a commercial activity, whether in return for payment or free of charge;
“manufacturer” means any natural or legal person who manufactures radio equipment or has radio equipment designed or manufactured and places it on the market under his name or trademark;‘manufacturer’ means any natural or legal person who manufactures radio equipment or has radio equipment designed or manufactured, and markets that equipment under his name or trade mark;
“authorised representative’ means any natural or legal person established within the Union who has been authorised by the manufacturer, in writing, to act on his behalf for specified tasks;‘authorised representative’ means any natural or legal person established within the Union who has received a written mandate from a manufacturer to act on his behalf in relation to specified tasks;
“technical specification” means a document prescribing the technical requirements to be met by a radio installation;‘technical specification’ means a document that prescribes technical requirements to be fulfilled by radio equipment;
“accreditation” means accreditation within the meaning of point 10 of Article 2 of Regulation (EC) No 765/2008;‘accreditation’ means accreditation as defined in point 10 of Article 2 of Regulation (EC) No 765/2008;
“accreditation” means accreditation within the meaning of point 11 of Article 2 of Regulation (EC) No 765/2008;‘national accreditation body’ means national accreditation body as defined in point 11 of Article 2 of Regulation (EC) No 765/2008;
“conformity assessment body” means a body that carries out conformity assessment activities;‘conformity assessment body’ means a body that performs conformity assessment activities;
“recall” means any action aimed at obtaining the return of radio equipment already provided to the end-user;‘recall’ means any measure aimed at achieving the return of radio equipment that has already been made available to the end-user;
“‘CE marking’ means a marking by which the manufacturer declares that the radio equipment is in conformity with the applicable requirements laid down in Union harmonisation legislation concerning its affixing.‘CE marking’ means a marking by which the manufacturer indicates that the radio equipment is in conformity with the applicable requirements set out in Union harmonisation legislation providing for its affixing.
the protection of the health and safety of humans and domestic animals and the protection of property, including the safety requirement objectives contained in Directive 2014/35/EU, but without the application of the voltage limit,the protection of health and safety of persons and of domestic animals and the protection of property, including the objectives with respect to safety requirements set out in Directive 2014/35/EU, but with no voltage limit applying;
an appropriate level of electromagnetic compatibility in accordance with Directive 2014/30/EU.an adequate level of electromagnetic compatibility as set out in Directive 2014/30/EU.
Depending on the specific combination of radio equipment and software, the information shall clearly identify which radio equipment and software have been assessed and shall be kept up to date.Depending on the specific combinations of radio equipment and software, the information shall precisely identify the radio equipment and the software which have been assessed, and it shall be continuously updated.
The Commission shall provide a central system for the registration of the required information by manufacturers. This system ensures adequate control of access to confidential information.The Commission shall make available a central system allowing manufacturers to register the required information. That system shall ensure appropriate control of access to information of confidential nature.
After the date of entry into force of a delegated act adopted in accordance with paragraph 2 of this Article, reports drawn up in accordance with Article 47(1) and (2) shall describe the consequences of that delegated act.Following the date of application of a delegated act adopted pursuant to paragraph 2 of this Article, the reports prepared in accordance with Article 47(1) and (2) shall evaluate its impacts.

通过对于上面表格的观察以及对比,不难发现,DeepL Translator 的译文质量很高,基本没有出现意义错误,除了表格中标黄之处出现了意义偏差,即没有细化名词意义,而是用了一个和上文一样的上层概念。除此之外,基本只是选词的不同以及句法使用区别。总体来说,欧盟指令原文用词更加精炼并且更符合法律文本的特色,用了大量的功能词组,而相对来说从德语译入的译文中,某些用词相比起来书面化程度没有那么高。比如:business & commercial activity, provided to & been made available to, provide & make available,confidential information & information of confidential nature,这些词对虽然在意义上没有太大差别,但是它们的使用语境以及书面化程度还是有所不同。

五. 结论

通过这次使用 memoQ 以及 DeepL Translator 的翻译实践,笔者收获了不少认识。首先,在处理非文学文本时,机器翻译的呈现效果基本令人满意,借助计算机辅助翻译可以大大节省译者的时间和精力。虽然此次没有用到术语库,但是可以设想,在存在大量专业术语时,对术语库的使用不仅能提高翻译准确度,而且还能有效统一文中的表达。其次,DeepL Translator 身为机器翻译的后起之秀,发挥出了意想不到却又是情理之中的效果,也许在日后,不懂德语的学者们可以借助 DeepL Translator 来广泛涉猎自己感兴趣的德语文献资料,翻译专业的同学也多了一个可以对比学习的平台和工具。

编辑注:GT4T 是一款优秀的机器翻译工具,可以直接翻译 XLIFF、PDF、WORD 等文件格式,集成了多种机器翻译引擎,调用方便。如欲购买字符额度,可使用优惠码 bigblue 获取九折优惠。

本文为作者 CAT 课程期末论文,未经作者允许,不得转载。

发表评论

邮箱地址不会被公开。 必填项已用*标注