Translation Crowdsourcing - Why and How?

1. Introduction

This portfolio aims to discuss the strength and importance of crowdsourced localization, alongside necessary implementations and considerations for using this method. Crowdsourcing essentially tries to gather the work and intelligence of many people and put them together to perform a quick and solid localization project. Several approaches to guarantee both quality and quantity are provided in Professional Presentation, while a practical use case in a proposal for Notion.so is used as an example. Notion.so is in need of a Chinese localization, and is a perfect fitting scenario given its massive native user base. The portfolio also demonstrates my understanding of using crowdsourced localization and its workflow, QA process, merits, and practical applications.

2. Quality and Quantity Control

With the option of crowdsourcing chosen, it is important to understand how to improve the quality and quantity in translation crowdsourcing. The quality includes metrics like accuracy and clarity, while the quantity denotes overall how many people and how much work can be gathered in the given period of time. Both quality and quantity are crucial to the outcome, and as the localization team, you need to pay attention to them for a good crowdsourced localization.

2.1. Quantity

First, improving the quantity. Quantity in translation crowdsourcing is almost equivalent to motivation. The more motivation there is, the more work people put in, and the higher quantity goes. There are six sources of motivation according to "Change Anything" by Patterson, Grenny, Maxfield, McMillan & Switzler, which are six reasons that affect a change or decision. Among those the first, second, and fifth points seem to suit the situation of translation and localization projects, under the assumption that it can recruit a portion of volunteers but still require many more, and these points are: Make the undesirable desirable; Surpass your limits; and to Design rewards and demand accountability. Some best practices to improve quantity following the prior motivations are as follows.

• First, add some comments from current volunteers with regards to how the project can help improve a resume to the recruiting advertisements. Examples can be like “I gained practical experience in the Company XYZ localization project, and I believe that it can be a great addition to my resume”, or “My interviewer was very interested in my experience volunteering for XYZ localization project, and I actually got an offer for this great job”. This strategy is more targeted at students and new graduates, since through research and investigation, their most common motivation is to gain practical experience and strengthen their CV. The localization team can pinpoint the idea that this project is very helpful for their career, allowing them to have the experience without spending too much time or effort.

• Second, try the best to design/customize the crowdsourcing platform for ease of use. This can allow volunteers to start working without much trouble, as well as to get a hold of “crowdsourcing software skills” with minimal effort. This is because that nobody will actually start if a lengthy manual and a 10-hour training on the software is required prior to the start, and the passion for volunteers comes and go extremely fast without timely feedback. A slow onboarding process will definitely lead to many candidates moving on from this project, while a simple one can offer volunteers much more satisfaction and motivation, improving the overall morale.

• Finally, provide plenty of feedback, including badges, medals, volunteer certificates, and public appreciations. This is probably the most common form of motivation, as people desire to be recognized and have something they can consider special. An award, even though not monetary, can make the project feel worthwhile to a volunteer. Many people also consider these kinds of “proof” as a great display of skill and another strong reinforcement to their portfolio, further increasing the motivation for them to help the localization team in its crowdsourcing project.

2.2. Quality

Second, improving the quality of the translations. This can be done in all three stages of the project, before, during, and after.

• Before the project commences, the localization team will need to communicate with other departments and stakeholders to make sure that all the content to be translated is finalized. The content picked should be parts that are most commonly used by clients, and that can have immediate use and impact after translation. Trying to pull off everything together can be overwhelming and make it impossible to perform good quality control. Usually, the application itself and especially its user interface is a must-do and the part that users are looking forward to the most. The user manual, update notes and templates can also be added to the crowdsourcing project early on as they are very helpful for new users.

• Once the content is set, the volunteers will need to be trained and evaluated. The training should be as simple and fun as possible, with an animated or game playing style. The training should also involve the key terms to keep consistent and the most basic workflow of the software. The applicants should be evaluated for the sake of quality, but the initial assessment should only cover experience and time willing to spend, in order to maintain their motivation. The rest, including specifics of past projects, can be optional and adjusted later on.

• During the translation project, the localization team should establish a points system for translators, which adds and subtracts points based on their actions and results, and in turn affects the creditability of their work. Other components of the translator community can involve a help center for guides and especially technical difficulties, community managers for better cooperation, and a gamified wall of fame/shame to demonstrate those who contribute the most as well as those who are “making trouble” with too much use of machine translation or prohibited words. There can also be a voting system where all translations need a majority vote amongst at least three people to be approved for use. The community will make a larger poll if there is a conflict, and the standard for “voting for” or “approval” should be set along the lines of being error-free, natural, and properly conveying the meaning of the original text even if there exist alternative translations.

• The best practices after completing the translations will include an internal review with automated quality check, functionality testing, additional button or spaces for user feedback, and UI size changes to match text that has expanded or shrunk. The performance of volunteers should also be rewarded with ideas such as a “level up” in contribution leaderboards or small souvenirs, but there must be enough caution to not use any type of monetary reward.

3. Crowdsourcing Translation Proposal for Notion.so

3.1. Notion.so

The digital world under the pandemic is asking for more and more from work, study, and entertainment software alike. Increasing online learning requires stronger notetaking software, while the sheer volume of remote work being done has raised the standard for project management ones. One of the best solutions to such problems combining the best of both worlds, is Notion. It is a great software to write and structure notes, as well as to manage tasks within a team: “Write, plan and get organized in one place”. Unfortunately, not everyone who needs it has the most convenient access to Notion. The CEO of the developer is Korean, and the supported languages of Notion only include English, Korean, and Japanese, excluding Chinese, Spanish, and the other remaining languages around the world. However, there is a massive user population of Notion amongst Chinese students, with communities across 3 platforms including Wechat, Telegram, and Notion Workspace, and more than 10000 contributors on each platform. According to relevant posts on third-party social media, many Notion users are waiting for a localization of the software into Chinese. Therefore, it is extremely valuable for Notion to consider performing such a localization, and potentially expanding it to an even wider variety of languages and user bases.

The content sections that would need to be localized in the Notion application include three main parts. First, the application itself, which includes the user interface and user-facing text, as well as all kinds of buttons, hints, and helpers. Second, the official guide for the application, which includes the user manual, help center articles, and update notes. Third, user-generated content, which includes lots of templates, experience sharing, and high-quality DIY community skill shares supported by the high user-freedom nature of Notion. The method and direction that may suit Notion the best under the current circumstances may be crowdsourcing, since Notion has the need to expand language support, and there exists a massive amount of users who know various languages and are actively making contributions to the community.

3.2. What is Crowdsourcing?

There are several important features of crowdsourcing in translation and localization. There are many people involved which results in an accumulation of many people’s intelligence, talents, and efforts. This can work fantastically with the current user population of Notion, and it also helps meet the actual needs and habits of users in the target language, almost a “made by clients, made for clients” kind of idea. Crowdsourcing also shapes the process to be all of the people working at the same time period, and working for a common goal. Alongside the fact that crowdsourcing is usually unpaid volunteer work, this minimizes the time and financial cost to the developers, and makes the launch of a localized version of Notion much more feasible. Some of the successful examples of crowdsourcing localization include the subtitle translation for TedTalk and the Facebook website localization. It took only 24 hours for the “crowd” to complete the French localization for the entire Facebook website, and it was also applied to Spanish, German and more, with at least 24000 phrases to translate for each language. The successes on similar styles and similar scale projects should prove that crowdsourcing localization could be the best option for the future Chinese and other localizations of Notion.

3.3. Proposal

Download the Proposal (MIIS Only)

3.4. Pitch Video

Download the Video (MIIS Only)

Download the Slides (MIIS Only)

4. Conclusion

With high user expectations for Notion to be localized into Chinese and many other languages, as well as the existing community who are likely to be willing to help with the localization project, it is a great idea for Notion to adopt the crowdsourcing localization strategy for their next step in integrating users from more countries and origins. By employing some of the best practices at maintaining and guaranteeing the quantity and quality of crowdsourced translation, there is a high chance that Notion can provide us with its freshly localized versions in the near future.