Crowd Translation: A Beginner’s Guide

众包翻译入门

2020-03-09 20:15 RWS Insights

本文共1142个字,阅读需12分钟

阅读模式 切换至中文

As more companies recognize localization as a fundamental ingredient to reaching customers around the globe, the need for translation has risen dramatically. And the localization industry has grown accordingly. Often, traditional translation processes aren’t up to the task of producing global-ready content at the increasing speeds and volumes necessary for companies to stay competitive. This has led the industry to develop new and more flexible approaches to translation than working with single-language vendors (SLVs). You may have heard the buzzwords: crowds, communities, curated communities. They’re often mistaken for each other, but what do they really mean? How do they work? Are they better for your business than going down the traditional route of sourcing translations from in-country, specially educated linguists through agencies? For this post, RWS Moravia Group Project Manager Marina Pantcheva is helping us break it all down for you. The first thing to note: ‘crowd’ is an umbrella term for a few different methods of translation management. We’ll compare those later. Let’s start by comparing the general concept of crowd-based translation with its traditional counterpart. What’s the difference between crowd-based and traditional SLV-based translation? Crowd-based: Ad-hoc, autonomous engagement; Tens or hundreds of individual translators per language; Flexible, fluctuating capacity; 24/7/365 coverage; Transparency into resources’ work; SLV-based: Specialized, long-term engagement; One or a few vendors per language Fixed, guaranteed capacity; In-country, specialized coverage; Reduced insight into work managed internally by the vendor; From the table, you can get a quick sense of how the two models differ, but let’s unpack them. At its core, crowd-based translation allows you to break up large translation projects into smaller chunks and distribute them directly to a large global community of freelance translators to work on concurrently. These resources may or may not be professional linguists, but the assumption is that they have mastery of the language(s) they translate from and the language(s) they translate into. Under the classic single-language vendor model, on the other hand, you contract specialized services from professional translation agencies or vendors that partner with their own in-country resources who have proven to be qualified in specific language pairs. SLVs might also divide up projects to handle larger volumes, but they use their own dedicated systems for project management and resourcing. And we can separate the differences even further: Crowd translations are based on open-call work distribution Working relationships under the crowd model depend on translators picking up the work. Individuals are free to claim translation tasks if and when they choose. SLVs, on the other hand, typically have contractual obligations to deliver, guaranteeing a certain throughput per unit of time. This means they are usually long-term partners that know your brand and localization program inside-out. Crowds have hundreds of resources for each language The number of resources you work with under the SLV model, true to its name, is often commensurate to the number of languages in your localization project: usually no more than a few resources per language. One agency may not be able to cover the entire scope of your project. Under the crowd model, a large bench of resources will be available, across languages, to ensure coverage of the work. It could amount to tens, hundreds, even thousands of translators per language. Crowds provide flexible and fluctuating capacity The capacity of crowds is calculated based on the number of translators in the crowd. But there can be big discrepancies between expected and actual capacity for two reasons: crowd translators can choose not to work, and crowds typically contain lots of “lurkers”: translators who are part of the crowd but do not actively contribute. Conversely, because of its contractual nature, a vendor’s or agency’s capacity is fixed and often guaranteed. Crowds have better coverage of time and geography Flexibility of coverage is greater under the crowd model. In general, crowds are active 24/7 because they work all days of the week, both within and outside of their native countries, allowing for better coverage over weekends, holidays and time zones. SLVs typically work with translators located in-country, in their own time zone, and national holidays can pose scheduling problems. Crowds offer excellent transparency and data Transparency is lower within the SLV model. Language service providers (LSPs) send projects to vendors and get them back with reduced insight into what happens in between. In the crowd model, project managers and clients can extract data on the way each individual resource works: their average turnaround time, quality scores, the types of projects they pick up and so on. This opens up an ocean of possibilities for the smart management of these communities. And here’s where crowds can be further categorized into three distinct levels of management, from unmanaged to carefully curated. What’s the difference between crowd, community and curated community? Crowd (or ‘crowdsourced’) translation usually involves a large network of resources with certain skills. You’ll find these freelancers through online crowd translation platforms. These crowds are public groups—the entrance is free for anyone who believes they possess the skills necessary for the crowd tasks—though crowd members may still be asked to take a test to qualify for the work. Crowds are best for tasks that do not require special knowledge or training, such as translating user-generated content. Translation communities (not to be confused with volunteer translator communities, which translate things like Wikipedia articles or TED Talks for free) are built by selecting, qualifying and engaging a subgroup of the crowd to form a network of resources assigned to a given project. These people may come from the crowd and become a dedicated group for you. There is a higher level of engagement and communication among the group—who, together, have a vested interest in the project’s success—plus more solid testing and qualification of resources. Communities are great for translation of more specialized content, in which brand terminology and style guides must be adhered to. Curated (managed) communities are hierarchically structured communities. The structure comes from assigning different roles to the community members (translator, editor, reviewer, etc.) and different levels of expertise within each role (junior versus senior reviewer, standard- versus premium-quality translator, etc.). Mirroring the same level of specialization you’d get with an SLV, curated communities can deliver complex tasks as well as an SLV can. So, what’s the best model? Each has its pros and cons. There’s no “better” model per se, but different models are best suited to certain situations. For example, the flexibility of crowd models lends itself well to localization projects of large or rapidly changing scale: an increasingly common need these days as content volumes increase alongside companies’ awareness of how important it is to localize their products.
本地化是接触全球客户的基本要素,因为已经有越来越多的公司认识到了本地化的这一重要性,所以他们对翻译的需求急剧上升。 本地化产业也随之发展起来。因为公司的需求是配合自身竞争力所需快速、大量地生产出全球性的产品,所以,传统的翻译模式是无法满足公司的需求的。于是,一种比单语服务供应商(SLVs)更灵活的新模式便应运而生,即众包模式。 你可能听说过这几个热词:群体、社区、策划社区,人们经常会把它们搞混。那么,它们各自的意思是什么呢?它们分别是如何运作的?它们是否比传统模式,即通过中介从国内的语言学家那里寻找翻译,更适用于你的业务呢? 在这篇文章里,我们就会帮助你区分这三个名词。 但是,你首先要知道,“众包”是几种不同的翻译管理方法的总称,稍后我们会进行比较。但是现在,我们要先来比较一下“众包”翻译模式和传统翻译模式的区别。 众包模式和SLV模式有什么区别? 众包模式: 通过大型网络平台,译者自主参与; 每种语言有数十或数百名译者; 工作灵活、工作质量具有波动性; 译者跨国界,工作时间持续; 工作数据透明; SLV模式: 专业译者,长期参与; 每种语言有一个或几个服务供应商 工作固定,工作质量能保证; 多为国内译者,工作时间固定; 对语言服务供应商内部管理的审查少; 上表比较清晰地罗列了这两种模型的不同之处,我们会在下面逐一介绍。 在众包模式下,你可以将大型翻译项目分解成小块,然后将它们直接分发到一个大型的全球自由译者平台上,以便让他们同时工作。这些译者可能是专业的语言学家,也可能不是,但众包模式假定他们全部都精通自己的原语和译语。 而在SLV模式下,你要与专业的翻译机构或服务供应商签订专门的服务合同,然后这些翻译机构或供应商就会与它们的国内资源(这些资源已被证明具有语言服务的资质)合作来完成项目。SLV模式也可以将大型翻译项目拆分成小块,但是在项目管理和资源分配方面,翻译机构或供应商使用的是自己专用的系统。 我们可以进一步区分这些差异: 众包模式的工作分配是开放的 在众包模式中,译者自主选择工作,他们可以自由接单。而在SLV模式中,译者通常具有契约的义务。因此,SLV模式中的译者通常是了解你的品牌和本地化项目的长期合作伙伴。 众包模式中每种语言都对应有数百名译者 在SLV模式中,你能使用的资源数量通常与本地化项目中的语言数量相当——每种语言的翻译通常不超过几个,翻译机构可能无法对你的项目做到全覆盖。但是在众包模式中,你可以跨语言使用大量的资源来确保项目的全覆盖。每一种语言可能对应有几十个、几百个甚至几千个翻译。 众包模式比较灵活,翻译能力具有波动性 众包的翻译能力是根据项目拥有的译者数量来计算的。但是其实际的翻译质量可能与预期存在巨大差异,因为在众包模式中,译者可以选择不工作,而且平台上通常有许多“潜伏者”,即不积极工作的译者。相反,由于其契约性质,SLV模式里服务供应商的能力通常是有保证的,波动性不会那么大。 众包模式工作时间更连续、地理覆盖范围更广 众包模式工作时间的灵活性更大。一般来说,众包一天24小时都很活跃,因为无论是境内还是境外的译者总会有在工作的,这样就可以更好地覆盖周末、节假日和不同的时区。但SLV模式中的译者通常只局限于国内,所以,在法定假日期间SLV模式可能会遇到人员调度问题。 众包模式透明度高 SLV模式的透明度较低。语言服务提供商(LSPs)大多数情况下仅仅将项目发送给供应商,然后到结束时将其收回,对中间发生的事了解较少。但是在众包模式中,项目经理和客户可以提取每个工作版块的数据,比如译者的平均周转时间、翻译质量评估、他们挑选的项目类型等等。这为译者群体的智能管理开辟了无限可能。 在此,众包模式就可以按照管理的精心程度分为三个水平层次。 群体、社区和策划社区的区别是什么? 群体翻译通常要用到具有特定技能的大型资源网络。你可以通过在线翻译平台寻找自由译者,而这些译者都是公共群体——任何认为自己具备完成群体任务所需技能的人都可以免费进入该平台(尽管他们仍可能要做一些试译来评估翻译能力)。对于不需要特定专业知识或培训的翻译项目来说,例如用户自己私人的内容,群体翻译模式是最适合的。 翻译社区(不要与志愿翻译社区混淆,后者免费翻译维基百科文章或TED演讲之类的材料)是通过选择、筛选从群体当中建立的一个资源网络,译者负责的是一个给定的项目。这些译者可能会跳出普通群体,成为你的专用翻译团队。他们可以从项目中获取既得利益,其参与度和沟通程度也会更高,当然对于这些译者的资质筛选也更严格。翻译社区是更专业的翻译项目的不错选择,因为这些项目里的品牌术语和翻译风格必须保持一致。 策划(管理)社区是具有等级结构的。社区成员担任的角色各不相同,有译者、编辑、审稿人等,每个角色的专业水平也不同,比如初级审稿人与高级审稿人、标准与优质译者等。与SLV一样,策划社区也可以完成复杂的任务。 那么,最好的翻译模式是什么呢? 每种模式都有优点和缺点,世界上没有“最好的”模式,不同的模式适用于不同的情况。例如,鉴于其灵活性,众包模式就非常适合大型或变数比较大的本地化项目。随着工作量的增加,以及公司越来越意识到产品本地化的重要性,众包正成为一种日益普遍的需求。 译后编辑:陆遥(中山大学)

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