Promoting Fortran in non-English languages

In this post, I talk about some simple ideas to promote Fortran and the work of our community in non-English languages. :united_nations:

We have already talked about non-English languages here:
https://fortran-lang.discourse.group/t/should-fortran-discourse-be-open-to-languages-other-than-english/385
And we discussed that some students (especially undergraduate) can be intimidated coming in our Discourse by two barriers : English level and technical level.

But another problem is the referencing of sites in foreign languages. For example, in France Google is dominating (>90 %), followed by Yahoo, Bing, Qwant and Ecosia. If I type ā€œFortranā€ in the french versions of those search engines, the results are:

  • Bing: Fortran-lang.org is at the 3rd position.
  • Ecosia: 2nd (in fact Ecosia uses Bing results).
  • Yahoo: 4th.

It seems good, butā€¦

  • Qwant: mainly results in French on the first page. And impossible to find Fortran-lang.org and affiliated sites, even going deep in the results (limited to 5 pages).
  • Google: like in Qwant, and I canā€™t find Fortran-lang.org and affiliated sites in the 10 first pagesā€¦

Those engines privilege results in the national language, as most people expect, and itā€™s a good and rational thing. But itā€™s a kind of bubble effect and in the scientific fields you need of course to break out of that bubbleā€¦ OK, I know that I have just to add an english word in my query to influence the engineā€™s behavior: with ā€œfortran languageā€, Fortran-lang.org is 5th in Google and 1st in Qwant! But most undergraduate students wonā€™t do thatā€¦ (I am trying to teach that to my 20 yearā€™s old students.)

1) Identifying and participating in non-English sites

As a consequence, we can search for non-English forums or Wikis related to Fortran and participate. For example, the first (and only?) french forum about Fortran is:
https://www.developpez.net/forums/f157/general-developpement/algorithme-mathematiques/programmation-parallele-calcul-scientifique-haute-performance-hpc/fortran/
where I have posted some news about Fortran and our community (Discourse, fpm, Milanā€™s bookā€¦) and I try to give answers when I can. OK, there is only a few new questions each month in the Fortran forum, but that site has a good page ranking and is considered as a reference by french developers. Therefore, people will easily find there news in their own language showing that Fortran is evolving and has a community. A first step before possibly joining us.

2) Writing papers in computer magazines

We can promote Fortran and its tools by writing papers in our national magazines. For example, in France the Diamond Editions are publishing very good magazines around open source: Linux Pratique, GNU/Linux Magazine France, MISC, Hackable magazine. The pay is pretty good, and the author can even choose to be paid less if he chooses a Creative Commons BY-NC-ND license for its paper, that will then be freely put online on their site some months latter: https://connect.ed-diamond.com/
Writing a paper for a magazine is typically a 20 hours work.

We could also write books, but itā€™s a huge effort, as Milan knows :wink: , and for a low audience on a national levelā€¦ (OK, it depends on your countryā€™s sizeā€¦)

3) Teaching Fortran

If you teach Fortran to your students you will of course speak about Fortran-lang and its projects. Or if you have Ph.D. students. And maybe put your PDF online.

4) Any other idea ?

7 Likes

I opened an issue at the fortran-lang.org repository to explore the possibility for localization of the webpage.

3 Likes

Thanks for bringing this up. As I commented on @awvwgkā€™s link, I think itā€™s a good idea to translate the pages to other languages.

I think it depends on the size of your country. My old country is small and as such I encourage to use English, as the lingua franca. I am unapologetic about that and simply ā€œrequireā€ it for anybody to learn enough to at least understand technical documents.

That being said, it is helpful to translate the webpage, and it does provide an opportunity to connect with more people and opens up more avenues to collaboration (asking people to translate the pages) and as a way ā€œinā€ for people.

It is also important to eventually get to the top in (local) search engines, so we have to do whatever it takes.

1 Like

One thing is the translation of webpages but another, non-trivial, is the vocabulary/terminology. Probably (at least it is so in Polish) the last version of Fortran which had some printed books published and the vocabulary (mostly) commonly accepted was F77. With the advent of new concepts, constructs, attributes and little-to-no printed material, it was up to (few as they were/are) academic teachers to translate. What has emerged is mostly a sort of calque. It is acceptable in speech but when written, looks ugly. Some terms have also changed in English vocabulary. Example: F77 used to have subprogram - function or subroutine. In Polish there were podprogram, funkcja and procedura (!). Now we have procedures everywhere so if we adopt procedura for the general term, we have nothing left for a subroutine.

2 Likes

If translating pages is decided, I of course agree to work on the french translation! And of course, it could be a collaborative work if there is other people interested here.

We can either do a 100% human translation or try to use an automatic translation tool such as https://www.deepl.com/ and revise by hand the result to accelerate (?) the work. Those tools can be very efficient, although the translation can be a little ā€œheavyā€ when there is a lot of technical terms in the sentence. Another problem could be on the copyright side: I donā€™t know if there is any restriction concerning those tools. We should read first their Terms and Conditions.

1 Like

Yes, some parts could be more difficult than others to translate, depending on the amount of technical terms.

In France, it is well seen to avoid mixed language (ā€œFranglaisā€ as we call it), when possible. I have just read the Fr SymPy webpage and at the bottom, the verb ā€œForkā€ was not translated, which yields the ugly ā€œForkezā€. We could use the verb ā€œBifurquerā€ but it is not used in that sense. Although it can be used as a transitive verb, according to my dictionary, it is very rare. It is generally used only as an intransitive verb. So itā€™s difficult to avoid ā€œForkezā€ā€¦ Of course, one must be familiar with a language to feel the ugliness of a term.

Each translation team could define somewhere a technical glossary of its preferred translations for technical terms, when several translations are possible, to get something homogeneous.

In French, a sous-programme can be a fonction or a sous-routine. But a sous-routine can also be called a procƩdure, which is preferred because a fonction could also be called as a sous-routine (which returns something).

2 Likes

In the case of technical terms, a criterion that can be used is whenever there is a word, in the local language, that fully or partially represents the same idea, use it in that language. Otherwise, the term can be used in English, highlighted or in another type font.

Thus, a table of equivalences between technical terms in English and languages ā€‹ā€‹other than English can be constructed. This table would always be available in all versions of the page.

1 Like

The problem is that such a word is rarely (or never) unique, the only one possible translation. When there are no books published in local language (say, a Metcalfā€™s MFE translation), we have no ā€œlocal standardā€ to follow.

1 Like

The selection of a specific word or term would require a consensus among the translators for a given language.

This consensus can occur according to the frequency of use, for example.

1 Like

@alozada, welcome to the forum!

@msz59 excellent point about vocabulary / terminology. Even before translating pages, I would love to have a section called vocabulary or terminology and have all the main terms translated to other languages.

As @alozada said, sometimes there is no established term. In that case we can at least list some candidates.

I should amend what I said above, that while I require English for documents that are to be read by international participants, I also like to speak Czech with just Czech people and I do not like mixing English terms in it. Having an agreed upon vocabulary of technical terms would greatly help.

I feel a vocabulary might be our first step as it would help translating down the road and help make progress in promoting Fortran in non-English languages, the original topic by @vmagnin.

3 Likes

I am glad you read the SymPyā€™s French translation. It helps to realize how hard it is to actually produce a high quality up to date translated version of pages. Itā€™s quite a big task. But we can at least start with a vocabulary as I mentioned in my previous comment.

2 Likes

I have read the Terms and Conditions of the free version of DeepL:
https://www.deepl.com/pro-license#free

Both private and business use of the DeepL Translator (free) service are free of charge.

There is no restrictions: we can use it to ease our translation work.

Sorry @certik , no Czech language yet in that European tool :disappointed_relieved: :wink: But it appears in the long list of languages available in https://translate.google.com/ (I have not read their conditions).

1 Like

Some linguistic tools from my bookmarks that could be interesting:

Translation tools

Terminology tools

  • https://iate.europa.eu/home ā€œIATE ( Inter active Terminology for Europe) is the EUā€™s terminology database.ā€
  • https://www.btb.termiumplus.gc.ca/ ā€œTERMIUM Plus Ā®, one of the largest terminology and linguistic data banks in the world, gives you access to millions of terms in English, French, Spanish and Portuguese.ā€ Especially interesting for the English / French translation. It even knows what is a forkā€¦

Bilingual dictionnaries

2 Likes

I can help with the Spanish version, now I am the owner of fortran.es

3 Likes

@awvwgk has now set the plugin to internationalize the site. See: Localization of the webpage Ā· Issue #197 Ā· fortran-lang/fortran-lang.org Ā· GitHub

1 Like

I think that a potentially big audience that we should consider (and that would benefit significantly from this work) are Mandarin (and Cantonese?) speakers. In that case the ā€œbubbleā€ that @vmagnin has correctly described is much ticker and more resistant.

The problem is to find someone able to revise the automated translation that might be very poor. I know that itā€™s a tougher problem compared to what discussed so far, but I think there is a lot of untapped, rising potential among Chinese developers and the effort might pay dividends. (And no, unfortunately, I donā€™t know any Mandarin and Cantonese, unfortunately :cry: .)

2 Likes

Hi Emanuele @epagone ,
yes, we should not put online automated translations. We can use it as a tool to accelerate (or notā€¦) our work, but my opinion is that it should always be revised and amended by a human being.

Perhaps there is already Chinese speakers here? And there is probably a lot of speakers in the American labs.

It does also depend on the Jekyll server support concerning Unicode characters.

Agreed.

It is with this hope that I typed my message: to throw the idea on the table and gauge if any Mandarin speaker is part of this discussion board.

Good point, I didnā€™t think about that. I donā€™t know.

1 Like

Well, there are. DeepL free plan limits the number of whole files (3/month) but also ā€œLimited characters & volumeā€ for ā€˜text translationā€™

2 Likes

@msz59 , you are right: if you want to translate files itā€™s three .docx or .pptx. And if you just translate paragraphs, you are limited to 5000 characters (at a time).

1 Like