User:Contraexemplo/Outreachy/Final report

• MediaWiki.org's style guide needs reworking, especially as most of its problems boils down to the lack of a clear difference between good and inapropriate contributions. The improvement of the source text will consequently improve future translations.
• The path to become a technical translator needs to be more clear. New users looking for ways of contribution should be presented with introductory pages rather than directly to the tools they will use. Visual media, such as short videos and screenshots, should be used more.
• Approaching universities to promote the role of technical translator is a unscalable strategy due to a series of social and geographical restrictions. It failed to recruit new translators in short term. However, it is unknown whether this strategy is effective in long term.
• The creation of a Brazilian Portuguese localization team with only new technical translators was successful. This is strategy is universal, as it can be applied to any language. Scalable, as it can accommodate as many translators as wanted and has no geographical or social restrictions. Lastly, it promotes the growth of a more integrated translation community as encourages teams to document their work and work together. In long term, this creates a legacy for newcomers to enjoy.

Introduction

Many Wikimedia projects have many years of existence, and MediaWiki, as the software that serves them all, is already sixteen years old. Although it is a solid product with multiple features and extensions, forms of use, and a large team of developers, its documentation and localization practices need to be revised and redesigned as much as the software and its wiki need technical translators.

This report explores and discusses existing practices from the perspective of a newcomer in the Wikimedia movement, describing the possible learning curve of an inexperienced person and pointing out aspects that need to be improved in the process to become a technical translator. It also reports the strategies applied during the two designated time periods, explaining its positives and negatives, scalability and effectiveness.

Current state

Documentation

One of the most important factors for an effective translation is a clear and well written source text. Ambiguous or inconsistent paragraphs may cause the translator to be confused about the intended message, causing the loss of priomordial information during the translation process. This ultimately disturbs the target audience from understanding the instructions provided, which can lead to misuse of the software. Therefore, creating style guides is imperative — in addition to helping the technical writer produce content consistent with the project identity, they establish conventions that make reading the documentation easier.

MediaWiki.org's style guide

The existence of style guide for documentation in MediaWiki.org is extremely positive. However, there are several aspects in it that need improvement, which are essentially caused by the same problem: the style guide fails to clarify the difference between a quality contribution and an inadequate contribution. The use of practical examples, introduced by a brief and clear text, is a more interesting solution than directing to the English-language Wikipedia style manual. The Atlassian writing style guide, for example, follows this concept. This will make content simple to read, easy to understand and apply without overwhelming the technical writer.

Localization

The How to contribute page points to two contribution paths related to localization: the translation of the software and its wiki. These forms of contribution have some similarities, such as the use of the same software (MediaWiki) and extension (Extension:Translate) in the translation process. They are also intrinsically related, as the documentation constantly mentions terms from the user interface to guide the reader. However, they have some fundamental differences that should be noted.

Translating the software

The translation of the software is done through the Translatewiki.net platform, which is independent of the Wikimedia Foundation and doesn't have any other association. Therefore, one of its fundamental characteristics is a process of entry isolated from other contributions in Wikimedia projects.

In this platform, to contribute you must make a registration and do a series of test translations to receive translator privileges. If the submitted translations are considered adequate when compared to existing translations, the user receives an e-mail message authorizing them to start translating any project registered in the platform.

Translating the documentation

The documentation is translated in MediaWiki.org, which is considered part of the Wikimedia projects. Unlike Translatewiki.net, creating an account to make contributions is not required, neither is submitting test translations. Access to the Translate extension can be done through page listing groups of messages according to its translation statistics, which is mentioned on the How to contribute page; through its direct access page; or by clicking on the "Translate this page" message when accessing any page marked for translation.

To assert that the differences and similarities between these two paths is a source of confusion to an inexperienced person is an understatement. As stated before, the fact that the translation of the documentation relies on translations used in the software makes it necessary for translators of the documentation to be present on both platforms to guarantee there is no divergence between terms of choice. However, the page that introduces forms of contribution in MediaWiki.org, How to contribute, does not make explicit such aspects as the Meta:Babylon/Translations does, or directs the interested party to the relevant documentation (Help:Extension:Translate). Exposing newcomers to tools they will use, without introductory explanations, creates significant obstacles for new contributors.

Translation community

Connecting with the existing translator community can be a challenge for newcomers. The How to contribute page does not mention ways to contact experienced translators, as Meta:Babylon/Translations does; in it, there are only forms of communication relevant to the development of MediaWiki software. The same thing happens in the user documentation of the Translate extension. Therefore, the problem that the novice technical translator should be familiar with the organization of the Wikimedia movement, looking for pertinent information in Meta-Wiki, persists.

Meta:Babylon/Translations directs the reader interested in software translation to the support forum of Translatewiki.net, to the internationalization mailing list, and to #mediawiki-i18n ^connect on IRC. For those interested in translation of announcements and documentation, it suggests using the translators mailing list and the #wikimedia-translation ^connect channel on IRC to get in touch with other translators. On the subpage dedicated to documentation, Special:SupportedLanguages ​​page is also mentioned as a way to find translators of the same language in MediaWiki.org.

It is important to note that Meta:Babylon/Translations presents such means of contacting translators as forms of support. This express a work culture within the Wikimedia movement that values ​​the independence of its members, giving them the power to choose what to work in and how. Thus, communication between translators is an exceptional situation. In the dedicated mailing list, notifications about content that needs to be translated by a given deadline are the most common messages to be sent.

Documentation for translators

In addition to the user documentation for the Translate extension, there is no help documentation directed to the translators. The absence of specific documentation to guide translators in their choice of terms and writing style, when they are not extremely familiar with the software and the Wikimedia movement, generates inconsistent translations. These conflicts ultimately disrupt the readers' understanding of user guides, making it difficult to comprehend processes and use software resources.

Volunteers that wish to translate software or user guides in Mozilla projects, for instance, are introduced to the work culture through their Quick start guide. Their localization efforts are extremely organized, with translators being encouraged to create teams for their locales and provide a "L10n kit" with glossaries, conventions, style guides or anything else they find useful to support their activities.

Those practices make it possible to continue the translation efforts if previous volunteers are not available in the future. The creation and maintenance of such documentation must be seen as a way to support future volunteers, providing them with proper training and tools thus making the process of becoming a translator less difficult to follow.

Overview of documentation translation

The Help:Contents page was used as a reference in the analysis of the documentation in MediaWiki.org, as it is a page constantly mentioned as a gateway for users seeking help on MediaWiki.org and is frequently searched by those who access the wiki. Keeping in mind the list of priority languages, I collected data on the translation rate of each page mentioned on Help:Contents, average and absolute number of views. Finally, a ranking taking into account each translation rate and number of monthly accesses was made.

As described in my fifth report, Chinese, Catalan, Portuguese (Brazilian and European), French and Polish are the six languages ​​with the highest translation rate. The positions of the Chinese and French languages ​​in the classification by monthly views are not discordant. Finally, of the six languages ​​mentioned, only Chinese, French, Brazilian Portuguese and Polish are present in the list of the ten most accessed languages.

Swedish, Hungarian, Persian, Finnish, Turkish and Arabic are the languages ​​with the lowest translation rates. The Swedish and Turkish positions do not stand out from their rankings in the list of most accessed languages. But, surprisingly, the classification of the other languages ​​mentioned differ from that previous ranking. The Help: Contents page in Arabic, for example, is the seventh one with the most monthly hits.

In order to understand these results, it is necessary to consider the level of proficiency in English of the priority language-speaking countries, the access of the speakers of these languages ​​to the Wikimedia projects (since they are censored in countries like Turkey) and the recognition of the projects in these countries (as evidenced by initiatives like the Inspire campaigns). The message that this data conveys is obvious: there is a growing demand from users who speak these languages ​​for translated documentation, and this is reflected in the number of accesses to their corresponding page.

Outreach strategies

As highlighted in Translation strategy, there is interest in reaching out to non-Wikimedians to present the translator role as a way to get in the Wikimedia movement. As stated in Improving MediaWiki’s documentation and localization practices, "(...) to find new translators, we need to look for places where diversity is welcomed and open knowledge is valued. We also need people that speak their native language well and also understand English at, at least, an intermediate level. Because of that, reaching out to university students and professors is our best bet, given this kind of collaboration has been growing in the last few years."

Taking into account I am an undergraduate and a new Wikimedian, verifying the accuracy of this statement seemed to be a good balance between coming up with new outreach approaches and taking advantage of my abilities and undergraduate status. Both strategies were tested in my home country, Brazil.

Approaching universities

Target universities

• University of Brasília (UnB)
• Federal University of Goiás (UFG)
• Federal University of Uberlândia (UFU)
• University of São Paulo (USP)
• Federal University of Santa Catarina (UFSC)

Applied methods

• Use of visual advertising material promoting the role of technical translator in MediaWiki.org.
• Companion text makes a small introduction to the Wikimedia movement and directs the interested to the Translation quick guide.

• 
  First version: Text in English
• 
  Second version: Text in Brazilian Portuguese
• 
  Third version: Explicit instructions

Administrative body

• Contact, through e-mail and face-to-face meetings, with:
  • coordinations of courses of Languages, Translation, Engineering.
  • departments responsible for advertising and journalism iniatives within universities.
  • centers responsible for teaching foreign language courses within universities.

Student body

• Contact, through e-mail and messages in social networks, with:
  • students from educational institutions.
  • Student entities (such as academic athletic associations).

Results

The administrative body of the Brazilian universities mentioned is not easily accessible to the general public. First, there is a technical difficulty: many of the institutional email addresses actively block popular email providers such as Gmail, accepting messages from other educational institutions only. Then there is a social aspect: even if you can successfully send a message, many employees are trained to ignore any messages from strangers. Finally, after much insistence, even if communication is established, the loss of contact happens very easily.

Federal University of Uberlândia

My most effective contact with an administrative body happened with the coordination of the Translation course of the Federal University of Uberlândia. However, this only happened because of my visit to the university. Were it not for my questioning to the course's secretary and the help of an employee who contacted the coordinator immediately after introducing myself, my messages would have been ignored.

I have been talking to the coordinator of the Translation course since December 2017. The slowest part of the process was to explain the operation of the Wikimedia movement and the translation process in MediaWiki.org. Many concepts sound confusing to the outside public, requiring a large number of clarifications and examples. After the introductory phase, we began to discuss forms of collaboration between the university and the Wikimedia movement. He proposed to join efforts to develop a set of best translation practices for MediaWiki.org, and continuous dissemination of the role of technical translator to Translation students.

Federal University of Goiás

My experience with their staff was not very positive, and much of this is due to the delay in making some progress and the need for face-to-face meetings to effectively present my project. My face-to-face conversation with the director of the communications department, for example, was quite receptive, but the lack of response in subsequent communications stalled this task. The meeting with the coordinator of the Center of Languages ​​(NucLi) was also welcoming, but bureaucratic limitations (the end of her term) and temporal limitations (the beginning of students' classes will happen after the end of my internship) did not allow our conversations to advance for the time being.

Other universities

Reaching out to other universities was quite dependent on the action of their students, who volunteered to disseminate advertising materials among groups that participate in social networks and mailing lists. Despite the effort, there was no significant response. The task, once again, stalled.

Conclusions

This strategy is not scalable. It requires a lot of time, effort and good connections with people from the target universities to get some results. In addition, much of the effectiveness achieved is due to visits to some educational institutions mentioned, which brings geographical restrictions. However, since it has been used for a short time, it is not yet possible to conclude this is an ineffective strategy.

Creating a translation team

Target audience

University students of courses specialized in foreign language study and/or translation studies with intermediate level of fluency in English.

Applied methods

• A tweet asking for help from volunteers within the target audience, rewarding the time spent on the experiment with a 20-hour certificate, was published under my personal profile.
• The post led those who got interested in this to the Project:PD help/Outreachy (Round 15) page, which explains the purpose of the experiment.

Results

It took 13 retweets for three people to come to me. Of these three, two have created accounts on MediaWiki.org and one has become a technical translator (User: Viviannesh). Informally, an Information Systems student also volunteered and became part of the team (User: QtK6z).

A translated version of the Translation quick guide has been made available to help the two beginners to get familiar with the tools they would use. Both were encouraged to explore them to the fullest and write down their questions and impressions. In the Project:PD help/Outreachy (Round 15) page, a list of target pages was made available, allowing each user to choose a page to translate. In two weeks, despite their busy routines, the page translation percentage rose from 24% to 65%.

From the account of the experiences of both, I was able to collect the following information:

Motivations

Among the motivations of the two users are the desire to acquire experience in the language and/or the translation profession itself, and the willingness to contribute to a great project. However, both described the granting of a certificate of participation as an attractive aspect.

Appraisal

Both described the Translate extension as intuitive and easy to use. Difficulties are focused on content-related issues, not the translation process itself.

The quick translation guide was a great help, mainly for exposing concepts through visual content.

Difficulties

Understanding the contribution process is a task that requires time and practice. MediaWiki.org directs interested parties to translate the content of the wiki to pages that are not intuitive and without introductory content. In addition, content seems to require experience with movement and software to be translated effectively.

Without external guidance, there is no way to know which pages should be prioritized. The available tools do not appear to weigh urgency or importance, making public only statistics related to the status of the translation of the message group.

Splitting the content of the page into small messages makes it easier to work, but it makes it difficult to understand the context. The lack of auxiliary tools (glossaries or conventions) made them feel insecure to make translation decisions.

Conclusions

This strategy is scalable. It can be applied to any language, with as many technical translators as are available. Setting a goal (a certain group of pages to be translated, for example) causes the group to focus on the activities without much disturbance.

The strategy was successful in recruiting new translators — both expressed a desire to continue contributing. Having an experienced translator on hand to guide them also has shortened the time to go through the learning curve — supervision makes the correction of early mistakes more effective.

Conclusion

MediaWiki.org documentation and localization practices need reviewing and reworking. It is necessary to make the documentation clear and consistent to provide a good source text to translators, thus restructuring the existing style guide with more direct and effective recommendations. There is also a need to reorganize the localization culture so that both novice and experienced newcomers can contribute by creating a system of collaboration between them through teams and by establishing conventions that standardize terms present in the documentation and MediaWiki software.

Approaching universities is a task that requires too much time and effort, and usually comes with social geographic restrictions, making it unscalable. However, as this strategy was applied in such a short period of time, it is not possible to assert if this strategy is completely ineffective.

"Passive" advertising (as done in the first strategy just promoting the role of technical translators) is not as effective in short term as "active" advertising (as done in the second strategy when making a more direct call for volunteers to make part of a localization team). The positive outcome of the second strategy — successful recruitment of two new technical translators to make part of the Brazilian Portuguese localization team in MediaWiki.org — suggests that a more communicative, supportive translation community is an essential trait to attract more new technical translators. Whereas issuing a certificate of participation may not be the main reason both mentioned as their motivations to become technical translators, the possibility of providing one for translators (and possibly other contributors) should be considered. The easiness to apply this strategy regardless of geographic or social restrictions makes it scalable.

Recommendations

• MediaWiki.org's Style guide should be more clear when establishing the difference between contributions of quality and those which need rework. It should not rely too much on third party documentation; instead, it should highlight which guidelines must be followed.
• The page How to contribute should direct those interested in contributing as technical translators to a introductory page explaning the differences and similarities between both paths of contribution, which has to provide essential information such as translation best practices, means of contact with other translators and documentation to support their activities. Meta:Babylon/Translations serves this purporse well. Efforts to make the translation process more clear must include the improvement of the content of this page.
• Introductory documentation regarding translation tools should be short and make more use of visual media such as short videos and screenshots. Actual user documentation should not be dismissed, and must be recommended as further reading.
• Translators should be encouraged to create teams and organize their translation efforts, rather than working completely alone. In addition to that, the creation of their own glossaries, conventions and style guides must be stimulated, thus making it easier to newcomers to join the movement as technical translators.

Although these tasks require a lot of time and attention, the development of auxiliary documentation for writers and technical translators is essential for an organized maintenance of the software and the ecosystem created around it. The objective, finally, is to eliminate points of failure in the contribution processes, avoiding redundancies and increasing the efficiency of contributors.

Acknowledgements

I would like to thank both my mentors, Johan Jönsson and Benoît Evellin, for the support throughout my Outreachy internship, either by finding valuable information, encouraging me to speak my mind and make decisions, or offering interesting opportunities to develop my skills and understand better the Wikimedia movement as a whole.

I also want to thank

• Srishti Sethi, for being such a wonderful coordinator and always providing great support.
• Sandra Fauconnier, for providing Commons:Structured data/About, for my training in December concerning marking pages for translation. It was a great way to put to practice what I have learned.
• Ed Erhart for helping me review, improve and publish my piece about documentation and localization practices on MediaWiki. Reporting my findings was an important part of this internship and I surely learned a lot with your observations.
• Niklas Laxström, for patiently answering my questions about the Translate extension and Translatewiki.net.
• My teammates, User:QtK6z and User:Viviannesh, for their help in the second strategy and their willingness to stay in the movement and improve MediaWiki.org's translation practices.
• Renata D'Avila, Outreachy intern with Debian, and V Körbes, Outreachy intern with Kubernetes, both for their support and welcoming conversations about writing, remote work, the internship itself and ways of communicate effectively with our mentors.
• Maru Arvigo, Lawgorithm researcher, for proofreading my essay for the Wikimedia blog, my last couple of bi-weekly reports, and constantly encouraging me to write more.
• My wonderful fiancé, for proofreading most of my bi-weekly reports, helping me with my visits to universities, constantly listening to me talking about my work and believing in me even when I didn't.
• All the students that volunteered to pass the word about technical translations in their universities and coordinators that welcomed me, in particular, Igor A. Lourenço da Silva, coordinator of the Translation course at Federal University of Uberlândia (UFU).

Appendices

List of priority languages

Language	ISO 639-1
Arabic	ar
Catalan	ca
Czech	cs
German	de
Greek (modern)	el
Spanish	es
Persian	fa
Finnish	fi
French	fr
Hebrew (modern)	he
Hungarian	hu
Italian	it
Japanese	ja
Korean	ko
Dutch	nl
Polish	pl
Portuguese	pt
Portuguese (Brazil)	pt-br
Russian	ru
Swedish	sv
Turkish	tr
Ukranian	uk
Chinese	zh

Data from Help:Contents

Translation of all pages in Help:Contents			Accesses to the correspondent Help:Contents page
Language	Translation completion	Ranking	Views	Monthly average	Ranking
Arabic	15.85%	18	28,279	943	7
Catalan	83.18%	2	18,078	603	16
Czech	32.40%	14	20,238	675	13
German	67.30%	9	63,877	2,129	2
Greek (modern)	16.25%	17	3,143	105	23
Spanish	59.78%	10	34,641	1,155	5
Persian	15.13%	21	20,199	673	14
Finnish	15.15%	20	18,278	610	15
French	73.78%	5	35,035	1,168	4
Hebrew (modern)	27.63%	15	17,259	575	19
Hungarian	12.15%	22	18,885	596	17
Italian	36.00%	13	21,373	712	11
Japanese	68.53%	8	31,117	1,037	6
Korean	46.25%	12	24,997	833	8
Dutch	20.88%	16	20,976	699	12
Polish	70.83%	6	22,829	761	9
Portuguese	74.73%	4	17,782	593	18
Portuguese (Brazil)	82.25%	3	21,893	730	10
Russian	70.30%	7	43,302	1,443	3
Swedish	8.33%	23	5,915	197	22
Turkish	15.63%	19	17,187	573	21
Ukranian	56.28%	11	17,249	575	19
Chinese	86.53%	1	69,192	2,306	1

See also

• Notes on suspected problems and possible solutions to them
• Notes on outreach strategies
• Internship daily notes, organized by month: December, January, February and March
• Improving MediaWiki’s documentation and localization practices, post on Wikimedia blog