FUEL (Frequently Used Entries for Localization) aims to solve the problem of inconsistency and lack of standardization in computer software translation in a new and unique way. Initiated by Red Hat, the project is trying to give a better experience to end users of a localized desktop by resolving the issues of standardization and inconsistency.
It's hard for a naive user to digest five different replacements for a simple word on a computer menu like File or Save. Getting used to inconsistent terminology can be a major hindrance in path of the popularity of localized desktop. Therefore, FUEL's aim is to come up with the most appropriate and acceptable translations by collaborating with active communities and localizers who are having standardization problems in translation. The project tries to give native language desktops a consistent look and make the localized desktop more usable for native language speakers.
There are lot of problems in the localization process. When translating any text, often we can translate one word in many ways. Misinterpretation, ambiguity, and lack of context can create critical problems for users. For example, in Hindi, the word "copy" can be translated in 45 different ways. It's a similar story for words like save, print, and tools. And different localizers use different words according to their preferences.
In previous standardization efforts, people generally worked on their own, with no collaborative input. Those efforts lacked the participation of active localizers and desktop users. For example, some government organizations and institutions of India have labored to create computer glossaries with several thousand entries and made them available to the public -- but no one used them! Because of complexity, lack of clarity, and nonuniformity, few people supported those efforts.
By contrast, FUEL is not a simple glossary project. Its process can be divided in four broad phases called the FUEL journey. A project starts with a list of all major desktop-related entries that appear on menus and sub-menus on a desktop, its panel, browser, office suite, editor, email client, instant messenger, and terminal. Currently there are 578 list entries in the POT file. Initially, if any language localizers want to work on the issues of standardization by following FUEL, their community needs to concentrate on these 578 entries. Getting these entries ready for translation by attaining consensus among localizers and experts can help standardize desktop translation quickly. This approach improves usability more than others that take on different applications one at a time.
FUEL is an attempt to standardize terms for the whole desktop instead of concentrating on different applications separately. At present, FUEL incorporates representative entries from the GNOME desktop, OpenOffice.org, Firefox browser, Evolution email client, and Pidgin instant messenger, so that it can have at least all the entries that a normal user uses very frequently. Later, on demand from communities, FUEL can incorporate more applications in its list from different projects.
Final evaluation of translation of a FUEL entries list is not a task performed by one or two people; rather, all active localizers of a language community sit down with language and technical experts to discuss and agree on one final list. They also agree that after the release of the list, all localizers will use only the terms on the list for their English counterparts, which ensures the success of implementation phase.
The release of an evaluated FUEL list is not the end of the process. A language community won't get the actual benefit of FUEL without implementation of the translated menus, so FUEL incorporates an implementation step to make the process complete. It is the responsibility of translators to implement the released FUEL list.
FUEL provides a set of steps that any content-generating team involved in creating localized content can undertake to ensure consistently high quality of their translations. The process involves a version control system, a ticketing system, and a mailing list. Each FUEL list is provided in different formats (PO, ODS, and PDF) for the ease of the general public. FUEL's language coordinators send information to mailing lists of the language about any development related to FUEL for that language. By joining and talking on the fuel-discuss mailing list, anybody can comment back to the project members. On a yearly basis each project adds new entries to the main FUEL list. In this way, development of terminology is tracked like software development.
Every process in FUEL is open from beginning to end. If anybody wants to start working on any language, he must inform the whole community through the mailing list. Evaluation of FUEL entries is also done via the mailing list. After the evaluation process, again, the evaluated FUEL language list is released on the mailing list of the community and the fuel-discuss list. Anyone can file a bug against any entries by creating a new ticket from project page. If the community agrees on any modification in current list, they can change it. Thus the whole development process for terminology is similar to any open source software development process.
Hindi - a success story
Hindi, one of the most spoken languages of the world, is an encouraging example for the FUEL project. With hundreds of dialects, Hindi is one of the most difficult languages for standardization. Although CSTT of India has made one glossary for Hindi language for standardizing computer terminology, it has not been widely accepted among software translators and localizers. Sarai and IndLinux have been organizing translation review workshops since 2003, and the experience of these workshops helped a lot during the final evaluation of FUEL Hindi. In this year's translation workshop, after discussion, the Hindi community agreed on the need to work on standardization. In July, all active localizers, along with language and technical experts, sat together for two days to discuss and finalize Hindi terms for the 578 FUEL entries. Finally, on July 13, FUEL Hindi was released as a community-approved set of guidelines.
The Hindi FUEL release is now in its implementation stage, and the community of Hindi language has started working on implementation. The community has completed implementation of FUEL for Firefox and now started working for GNOME. The implementation process involves changes in the existing translations, replacing some entries with others from the FUEL list. The implementation phase is lengthy and labor-intensive. All major projects are already available with Hindi localizations, so it will take at least another release of involved projects to implement FUEL recommendations for Hindi language.
In the weeks since the successful release of FUEL Hindi, four more languages have actively started working within FUEL.
Words are the only wealth
The FUEL project page starts with the following quotation of the great Bhakt poet Tukaram:
"Words are the only jewels I possess,
Words are the only clothes I wear,
Words are the only food that sustains my life,
Words are the only wealth I distribute among people."
The importance of words is ultimate for the FUEL project.
If any language community wants to work on FUEL, it should start by sending information about using FUEL for their language to their community mailing lists and the FUEL mailing list. The FUEL process is open for all enthusiasts who want to join the evaluation process; collaboration and cooperation of whole community is a must for the entire process.
Achieving the end of standardization is not an easy task, as it requires a lot of change and rework in existing translation, so there is a possibility of slower growth of FUEL fully implemented languages, but the process ensures that, when a localized version is released after implementation, it will come with greater reliability and perfection due to the collaborative and open process. In short, FUEL is a collaborative, open, and transparent way to achieve standardization.