Localisation 5 August 2026
User-Generated Content Translation: How Web 2.0 Changed Localisation
Reviews, questions and ratings may be the most read part of your site. So by what criterion, by whom and how far is that content to be translated?
On an e-commerce site the most read text is often not the product description written by the company. It is the reviews left by users. The same holds for forums, question-and-answer sections and community pages. This content does not belong to the owner of the site, yet it is largely what determines the site’s value.
Contents
User-generated content translation is one of the fastest growing and least regulated areas of web localisation. The source states that the collaborative and interactive capabilities of Web 2.0 have led to the production and translation of a vast amount of content (Jiménez-Crespo 2024, pp. 174-176).
Below we set out where this content comes from, which platforms translate it, what questions it raises in terms of quality and liability, and what sort of framework the corporate side can put in place. The subject may look like a technical preference, but the real decision has to do with whose responsibility a given piece of content is.
Web 2.0: the read-write web
The inventor of the WWW drew attention to the social nature of the network from the very beginning: “The Web is more a social creation than a technical one” (Berners-Lee 2000: 113). That observation may be counted as having foreseen the age of social networks, and it sets the frame for today’s content economy.
Web 2.0 describes the stage at which the production of content became more dynamic and user participation moved to the centre. Berners-Lee calls this the “read-write web”. That stage brought with it not only social networks but electronic commerce and the sharing economy as well.
Web 3.0, for its part, took personalisation further. Content is shaped to the user’s preferences and context; the same account travels between different devices. For localisation this means that the same page may look different to two users.
This personalisation also makes quality control harder. There is no fixed page to be checked; there is a view assembled for the user. During the testing phase, therefore, as many scenarios as possible have to be tried and the way the page is assembled under different language preferences has to be followed.
What does user-generated content translation cover?
User-generated content is content produced not by the owner of the site but by its visitors. Product reviews, forum posts, question-and-answer threads, uploaded images and subtitles fall into this group. Its volume is large, its quality variable, and it flows without stopping.
Translating this content differs from classic localisation on three points. First, the source text has not been checked; spelling mistakes, slang and incomplete sentences are ordinary. Second, the volume cannot be foreseen. Third, how much of the content will be translated cannot be settled at the outset.
Its value, on the other hand, is high. A local user finds an account of experience written in their own language more convincing than corporate text. Many platforms therefore choose to make user content visible, even if only through automatic translation.
The decision about scope is taken here with two questions: which types of content will be translated, and what level of quality will be aimed at. Readability may be enough for reviews, while for a user question containing a safety warning it will not be counted as enough.
From consumer to producer
Web users have become the co-creators of digital media. The literature calls this figure the “co-creating user” (Banks 2009) or, in an older term, the “prosumer” (Toffler 1980). In time those same users took on the role of volunteer localiser as well.
This transformation produced a twofold result for the translation profession. On the one hand the volume of content being translated rose far above professional capacity; on the other, the question of who does the translating, and with what authority, became blurred.
The source describes this development as an additional layer added to the evolution of localisation. Volunteer translation, that is, is not an alternative taking the place of professional processes; it is a separate plane growing alongside them.
For the corporate side the practical consequence is that the definition of scope widens. Localisation used to be limited to the texts the company itself produced. Today an important part of the site’s visible surface comes from users. When that content is left out of scope, the localised site covers only a portion of the source version.
Volunteer translation platforms
The online collaborative translation platforms used for localising web content have become commonplace today. Among them the source counts platforms for subtitling, social networking, microfinance, conference talks, software localisation and encyclopaedias (Jiménez-Crespo 2024, p. 175).
These platforms have an interesting feature: while they are themselves localised by users, they at the same time offer easy-to-use workflows for the localisation of other web content. They have become structures that are both translated and translating.
For the corporate side the real lesson here has to do with tools. What makes volunteer flows possible is the disappearance of the technical barrier. That same ease can be used in in-house translation processes too: gathering contributions from within the organisation no longer calls for heavy infrastructure.
Even so, a volunteer flow calls for a model of governance. Who will approve, which term list will apply, how will disagreement be settled? A channel for contributions opened without answering those questions can turn in short order into an inconsistent heap of text.
Participatory cultures and fan translation
The WWW allowed the rise of the structures the literature calls “participatory cultures” (Jenkins 2006) and “fan cultures” (Hills 2002). The “user-generated translations” (Perrino 2009) that emerged in connection with them added a separate layer to the evolution of web localisation.
In these communities translation is not paid work; it is an expression of belonging. Because the motivation is different, the criterion of quality is different too. The community more often than not puts speed and accessibility ahead of formal perfection.
What the corporate side can draw from this is the following: your users may already be translating your content. Bringing that within a framework rather than ignoring it is the more realistic way of protecting both quality and brand consistency.
The simplest form a framework can take is to give the community a term list and a short style guide. Users who contribute mostly want to know the right equivalent; the problem is not unwillingness but the lack of access to a source. Sharing those two documents can raise the quality of volunteer contributions visibly.
The professional process compared with the volunteer flow
The two approaches are not rivals; they are different instruments for different types of content. In most organisations both are used at once: product information goes through the professional process, user reviews are translated in an automatic flow. The table below summarises which is reasonable in which case.
| Criterion | Professional process | Volunteer or automatic flow |
|---|---|---|
| Suitable content | Contracts, product information, safety warnings | Reviews, forum posts, archive content |
| Speed | Can be planned but limited | Very high |
| Quality control | Recorded and measurable | Dependent on community moderation |
| Liability | Defined by contract | Mostly unclear |
| Cost | A fee per word | Infrastructure and moderation cost |
Quality, liability and transparency
The most critical matter in translating user content is transparency. If it is not stated that the output is machine translation or a volunteer contribution, the reader takes it for text written by the organisation. A faulty translation in a review can be perceived as the brand’s own mistake.
The second matter is the boundary of liability. Statements that may have consequences for health, safety or the law should not be left in unmoderated flows. Content of that kind is put into a separate queue and passed through human review.
The third is rights. The content belongs to the user; the platform’s authority to process, translate and publish it must be defined explicitly in the terms of use. For reviews holding personal data, the requirements of data protection legislation are assessed separately.
The fourth is measurement. Whether translated user content is actually being read has to be followed. The resource spent on translating an archive nobody reads yields far more when it is shifted to current content.
The fifth is continuity. User content is not a one-off delivery but an uninterrupted flow. The decision to translate it is therefore not a project decision but a business decision. Unless it is settled who is to look at what and how often, the order established in the first months breaks up within a few more.
The Türkiye context
In Türkiye user reviews play a decisive part in the decision to buy. On local platforms the question-and-answer sections are read almost as much as the product description. When that section is left empty in the Türkiye version of a foreign brand, it creates a visible gap in terms of trust.
On the language side there are two practical problems. First, a significant proportion of users write without using Turkish characters, which lowers the quality of automatic translation. Second, abbreviations and regional usages are frequently resolved wrongly in machine translation.
On the legal side it is the KVKK that governs. A name, a telephone number or order details may appear in a review. Sending that content to a third-party service for translation amounts to a transfer of data and calls for a privacy notice and a legal basis.
What does this mean for you?
The decision to translate user content is a policy decision more than a technical one. The sequence below lets you put that policy in writing.
- Determine which types of user content will be translated and which will remain out of scope.
- Write down the target quality level for each type: reading for information, or text ready for publication?
- If automatic translation is to be used, decide how this will be made known to the user.
- Make human review compulsory for content that may give rise to liability.
- Have your terms of use and your privacy notice updated so that they cover the translation operation.
These five items protect both quality and the legal footing. You can reach our other articles on the subject from our knowledge centre, and for your own content flow you can use the quote form.
This article is for general information and is not legal advice.
Frequently asked questions
What is user-generated content?
It is content produced not by the owner of the site but by its visitors. Product reviews, forum posts, question-and-answer threads and uploaded images fall into this group. Its volume grew rapidly with Web 2.0 and it came onto the agenda of the translation industry.
Does volunteer translation take the place of professional translation?
Generally it does not; it stands alongside it. Volunteer flows are useful for content that is large in volume and changes quickly. In areas that carry liability, such as contracts, safety warnings and legal texts, a professional process and a recorded quality control are needed.
Is it right to translate reviews by machine translation?
For reading for information it may often be enough. It is expected, however, that the user be told plainly that the output is machine translation. Translating statements that may give rise to liability in this way also produces risk; content of that kind should be put into a separate flow.
Is there a legal side to translating user content?
There is. The content belongs to the user, and the platform’s right to process, translate and publish it must be defined explicitly in the terms of use. For reviews holding personal data, the requirements of data protection legislation are assessed separately.
Sources
- Jiménez-Crespo, M. A. (2024). Localization in Translation. Routledge, pp. 174-176.
- Berners-Lee, T. (2000). Weaving the Web. HarperBusiness, p. 113.
- Jenkins, H. (2006). Convergence Culture. New York University Press.