 Hello everybody. Welcome to this talk on expediting the Brazilian-Portuguese documentation process. I'm very glad to be part of this LibreOffice Conference 2021. My name is Timothy Brennan, Jr. I live in Brazil. I've been living in Brazil. My family has been in Brazil for three generations. My children are the fourth. My wife and children are Brazilian and I speak both English, American English, and Brazilian Portuguese fluently. And I am very happy to be part of this process, helping out with the LibreOffice and the translation process and everything else involved. So let's talk about one of the issues that we have with documentation. So software releases in LibreOffice, especially after LibreOffice started its release cycle every six months. Very fast, but the documentation has a hard time keeping up with the release cycle. So it's a slow process because just the beginning, the first part of the cycle translation takes a long time. Then you have to go through two revisions, you know just one isn't enough, and finally there's the publication process. So we had to speed up the process for us to be able to catch up and keep up with the software. So there are some changes and some strategies that we took on and so one of the things is translate the main document once and then do a correction of the translation and subsequent documentation be reviewed rather than just simply always translating the full document from 100% English back into Portuguese again. It's very expensive in terms of time and effort. So we also when we do this we're going to keep and maintain different formats like ODT, PDF, HTML, EPUB for ebooks and stuff like that. So when we update a new version we can update also our centralized repository, I guess you could call it, where we keep our documents. So with that all we need to do is review the Portuguese version rather than translate the whole thing from the beginning. So what do we do to speed up the translation part? We actually use an online open source service, we're going to talk about that very soon. We upload the file, ODT file, have the machine translated and downloaded, that's simple. So this eliminates also the language barrier that many times could exist when some Brazilian translator is fluent in spoken English or sometimes isn't able to keep up with the nuances of some of the things in English. So machine translation really lowers the language barrier a lot and it is much better to have an imperfect text than no text at all. So this is one of the things that we've been very emphatic on. Now the big heavy work is revision. So we correct the sentences because machine translation, as everyone knows, is not perfect. It helps a lot but there are a lot of funny mistakes we'll see about those really soon. We also have to really go through a lot of consistency check to make sure that what was translated and what has been translated fits with the software. And after that part is done, which is the bulk of the work we do today. Publication. And finally we reach the level of publication. By the time we get there the hard work has been done. Publication does involve a lot of work on the final revisors. So what do we use? We use makecat.com. Obviously there are others. We're not espousing only makecat but we got open source. So that's one of the big things that we focus on having an open source machine translation. Also makecat.com does not require to install any software so we can do that online. And the process is very simple. Get the original file in English, upload it, have the machine translate it and download it. Simple as that. We can also add a standard translation member if we want to. We haven't had a need of it yet. And then that translated document can also remain online. We can go back and download that document again if we have to. So it's a nice way to keep a little backup of the translated file. So we maintain the ODT format. It keeps all the styles, all the fonts, and everything, the tags that we have in it from the ODT format. So also, technical English is easy to translate in machines. So there's a very low percentage of errors or a high percentage of correct answers or translations. So that's another advantage of machine translation. Now the big work is correcting the machine errors and translator errors. And that requires, you know, at least for you to be good in Portuguese. So everyone on the team is fluent in Portuguese. And have a good working knowledge of English and be able to compare between one language and the other. So there are a lot of contextual errors that have to be checked. Software elements have names that can confuse a translator, for example. So here are some examples of some of the things we have to correct. In Portuguese, you'd say, click on the organize tab and it translates Piquinha Organizar Abba, which is wrong in Portuguese. You have to switch the order. Adjectives and nouns are reversed in Portuguese. So Piquinha Abba Organizar, so we have to keep our eye out for that. Some humorous and funny things that happen, and we also have to sometimes stop and really stop and really think about what we're reading. Like for example, bold can mean bold as in a strong font, but bold as we know in English can mean audacious. Well, when it translates bold to audacioso in Portuguese, it's just so funny because it has nothing to do with bold font. And we have deck. Deck can mean a deck of cards, but it also can mean a deck like an extension of your house in front of your house, something like that. So the word Baralho, specifically deck of cards, has no margin for other meaning. So a lot of times we have to change that. In fact, we've changed deck to pineal, which is something more like panel. Works much better in Portuguese, no ambiguity there. Outline. Outline we use for, you know, when we're outlining a picture, when we're outlining a document, stuff like that. In Portuguese, the terms can be very, very different. So when we say, when the machine translates contorno, for example, that has no meaning in the context of outlining a text, contorno would be something more like a draw when you're drawing a rectangle or a square and you do a line around that, that line, that perimeter is contorno. So we've had some laughs with that. Also left as in left and right and not left as in the past tense of leave. So sometimes the machine will misinterpret it and put down de show, meaning go away. Table, which we use for where we eat on are also the table on our document. Meza has nothing to do with tabela. So we have to keep our eyes open for that. And shift, meaning the shift key, which does, in the old times of the typewriter, shift the letters from small caps to upper caps. But in Portuguese, it's totally meaningless. Mudança means change of transformation. Character in a movie, in a book, personagem in Portuguese has nothing to do with caractere, which is what we're looking for in this document. Break, as in line break, column break, things like that. More the translation sometimes means rupture, like what in the world? And then all of a sudden, oh yes, break, of course. And shift plus end, which is finguturno, meaning the end of the shift for workers, such as doctors, nurses, and people like that. So that's part of the bulk. And we have some good laughs with that because obviously a machine is not a human. So it can't keep up with all the nuances our brains are capable of as humans. So we also revise and correct for better reading in the target languages in Portuguese. Sometimes it may be grammatically correct, but it doesn't sound right. So for example, the translator might mimic the English language. And so we have to take a look at that, say, hey, this is unnecessary. This is extremely redundant. This sounds weird. So for example, we have this here already. If the element already exists, we just get already out in the context this happened in Portuguese. So because it's just really not necessary. For example, in Portuguese, the pronoun you is translated voce, where many times voce is unnecessary because the conjugation of the verb that follows already tells what person it is. So we can eliminate that. And sometimes I use a control H for voce. I go one by one, check them out and very rarely do I leave one in. So we also focus on good writing for technical texts. And so that requires that we change sometimes a little bit a few of the terminologies and expressions. Grammatical mistakes are very rare, you know, but they're there, but they're obvious. They stick out. Now the golden rule for this consistency check process is the software is always right. What does that mean? Sometimes we'll find inconsistencies between the documentation, even in its original in English and in Portuguese. So we will go to the software. The software says such and such a thing. The documentation will say another thing and we will follow what the software says. And so we will follow what the software says. And so differences with the software will be reported. The interface and help translate on web light, we can do like fixed translation errors, context errors, terminology errors, grammatical and linguistic errors. All these can be fixed and they will also sometimes even serve as bug reports for who knows, even some software mistakes or something in the menu system or contextual errors that are confusing in Portuguese. And sometimes even we'll find English errors or an error in terminology or concept. So one of the nice things about being in a documentation team has been that we are able to help out with some development, you know, not on the level of a programmer. So how does a team being managed right now? So we have a channel on telegram, instant messaging for quick communication during the week, and we call people to join during the week who are on the team on PapuLibri, which is LibreChat in English. And we use documentation, GC documentation platform for us to chat. That's our one hour a week meeting. We discuss anything LibreOffice from technical issues to minutiae of how the proper hyphen or, you know, half hyphen or dash or whatever these things we've talked a lot about. We try to keep it an hour. Sometimes we go about 10 or 15 minutes over time. But then as our friend Tulio always says, let's stop now because we'll run out of subject to talk about next week, which is a joke. Of course, we never run out of subjects, but we'd go too late. But anyways, it's really nice these meetings on a weekly basis. And some of the earnings we've had, I've learned a lot of LibreOffice on a software level just by reviewing the documentation in Portuguese. It forces me to think in two languages and forces me to use the software in a manner that I wouldn't if I weren't doing this. So I learned a lot about styles, formats, advanced techniques, databases, mail merge, security, all kinds of stuff, safety and everything. So each of us grabs a chapter. We put our names on a spreadsheet. And when we're done, we let everyone know. And we check to see if anyone else has got another chapter. If that chapter hasn't been taken, we will take it and move on to the next one. So our knowledge of professional documentation creation has increased greatly. How to organize ourselves to get tasks done, the assembly of the final document where Olivier sets up the master document and the chapters and everything. Our language expands. My particularly in Portuguese language has developed quite a bit just by listening to different ones talking in Telegram and our chats on Jitsi and everything else, IT, terminology, knowledge and everything. So we get to participate in development of LibreOffice software with the international team. This is a really nice thing, too. We start getting more involved. And of course teamwork, I mean, what can I say? Teamwork is always the best. So after all of this time, we've done this. Here we have Estanche de LibreOffice, which is something equivalent to LibreOffice bookshelf. And this is, of course, an older picture. As LibreOffice is always updating, the documentation is also. And we have our team here of people. And so basically we have, from this point forward, I'm going to show you here. I'm going to go around this area here. And so we have Jackson Cavalcanti. We have Tulio. We have Olivier, which most everybody I'm sure already knows, too. And we have Cristina. We have Diego. And down here we have, what's his name? Rafael Lima. This is our roadrunner. His name is Raúl Pacheco. He's our roadrunner because he's so fast and he does everything faster than anyone. He leaves us in the dust with his work. So he's well known in our group for that. And we have here Luciana. And then we have Marcia. We have Flavio Chefer. And we have Vera Cavalcanti. And this is myself here. Everybody, thank you so very much. This has been a great opportunity to be a part of this team, worldwide team, international team of LibreOffice. I've learned a lot and I have so much to learn yet. Thank you very much for this opportunity and I am open for questions. And hopefully I'll be able to answer them. Okay. My video is finished. Yes, I have a question for Tim. What is the next steps that the team is taking in the future? Hello. So our next step that we're going to be taking is to develop more the documentation and the newer versions. We are going to be not doing a full complete translation, but working on the documentation we have. And I guess I don't know if that answers your question, but I guess that's the answer I would have right now. Yeah. That is interesting because since we have eliminated the translation and one of the things that we have asked upstream is that the new developments in the English documentation that we get a specific section to indicate the changes in the books so that we can go directly into the new parts and new pieces of documentation and just translate what is there. It doesn't prevent us to review the full book, but nevertheless it's a little bit easier for us to update because we don't want to go through the full cycle of translation. There is no need for that anymore. So the idea is to keep close look at the developments in the English documentation. And our experience in Brazil, I think it's very advisable to reproduce into other languages. One of the things that it really matters is that machine translation is lowering the barrier, as Timothy said. And it's much easier to get people that are knowledgeable in your language than people that are knowledgeable in two languages. So you can really increase the amount of people participating in the group of documentation by eliminating the language. So this is good, yes. Yes. True. I have done some translation work and definitely the machine language translation has expanded much our capacity to save on time and focus on quality documentation. Yeah. Any more questions, anybody? So, room is nothing? Okay, my question. So this is a very interesting talk because the Japanese community also challenges the documentation translation. But we can not success now. So machine translation is so easy, but so translation quality is not good now. Yeah. So translation in Portuguese is so machine translation quality is good or not, but that's your question about the quality. Yes, yes, quality. Yes, yes. The quality is good. I think one possible thing that helps is we share English and Portuguese share a similar script system. If my knowledge of Japanese writing system is correct, I think there are two or three writing systems, right? Yeah. So I don't know how that works on a machine translation level because I know nothing about Japanese language, I mean very little except. So it has worked very well because like I said on the presentation there, there are some small differences, but structurally, you know, Indo-European languages, the script are very similar. There are some just switches around grammatical structures, but the quality is excellent. I mean, I can attest to that. I've been a part of the process. The quality is very high, and I think it's like Olivier mentioned just now. It is something very interesting to check into, and I'm sure the people in Japan who would work on a team like this, who are working on a team like this, would have their own sets of challenges, but I believe it's these developing, facing these challenges is what helps develop new technology. I know that one of the advantages of using machine translation, especially for European language like Portuguese, is that the amount of vocabulary of the technical English is very shallow. Although Timothy mentioned some mistakes, the number of mistakes is very low, and the technical English is not a very rich English. So it's very limited in terms of vocabulary, and we pick one or two mistakes, but it's not like translating Shakespeare. Probably the number of mistakes would be much higher, so this is an advantage. To me, machine translation is like a bilingual dictionary on steroids, I guess you could say, with lots of options and capacities. It just enhances our translation capacity. I remember years ago when I would translate from Portuguese to English or vice versa, I would use an actual paper book dictionary and go through and everything. So it's just an enhancer. To me, it's like a great enhancer. It really saves a lot of time for us. The big work is to actually process the document, compare with the original, compare with the software. For example, I have to have a labor office in Portuguese, a second installation of a labor office in English to check what the language is in English. So then when I'm doing the documentation review, I'm also doing screenshots and everything. It's what we all do in our team. So it's very interesting. It really pushes us to develop as humans, even though we're using machines, we are developing ourselves. Yeah. And now chatting, chat room is one question. So that's guide. Sorry, I pasted the chat. Oh, okay. Guide, exhalation require meeting, or is it possible split efforts per individuals? Translation, okay. Meetings are very important because you have to talk about details, you have to talk about specificities of your target language. So for example, in Portuguese, like I said, those were humorous examples I gave, or discussions about the hyphen, the dash, the double dash, and things like that. For example, in Portuguese, like on Room 1, Felima was mentioning how the desk point and the thousands placeholder, they're reversed in Portuguese. Those are just small examples of what happened. So it's very necessary to have these meetings. I would say you can split the individual, we do split them up per individuals, but without the meetings, I don't think the process would work that very well. That weekly meeting, that human bonding that we develop on a weekly basis, really makes a big difference for us. It's just necessary. We're not machines. We're human beings. So this part about us getting together on a weekly basis, chatting, laughing together, talking about common subjects, but always our meetings are focused around labor office, and we try to stay focused. Sometimes we'll go off on these small rabbit trails, but that's what adds to the spice of being part of the team. You know, it's not me in my own office working alone. It's human beings bonding together. I think it's possible, but it makes a big difference when you have these meetings on a weekly basis. That's my answer for that question. Thank you. I just wanted to add something. Yes, the meetings, like Timothy said, is important for the bonding, but also, of course, since we have chapters and the document can be split, of course, by several, and you can distribute the workloads between several members of your community. But, of course, there need to be a coordination, and the coordination is rather simple. We expect, for example, that when we pick a chapter for revision and we put in advance dates, deadline, we try to honor the deadline. If we cannot honor the deadline because we are all volunteers, then we send a message saying, I'm going to be a bit late, but we manage the expectations, and we try to, of course, to be as strict in terms of commitments, which is important for us. So, yes, you can split the work into several individuals, but it's good to have a weekly meeting or a bi-weekly meeting just to follow and to keep the work active. I think there's a second question there. If I'm not mistaken on telegram, mind if I read it? Okay, there we go. Thank you. Have you tried translating complex, structure-wise docs such as basic cheat sheet? As far as I know, we're focusing on the documentation right this moment. We're not focusing outside of that. It may be a subject we might bring up, I don't know, Olivier could probably even answer that better than me, but right now our focus has been the official documentation. I just want to add that these basic cheat sheets, if they are the ones that were created by Jean-François Niffnaker, then this is a very nice piece of work and it's maybe interesting, yes, that we get it translated into Portuguese. Of course, there is a small amount of text and quite a lot of lines of code, so the code is not going to be translated, but yes, this is a nice thing to add. Yes, of course. Okay, so it's time up now. Thank you. Thank you. You're welcome. Thank you so very much. Great presentation.