Re: [libreoffice-l10n] Bavarian and Nipmuck - report
Sgrìobh toki na leanas 12/12/2015 aig 22:51: That doesn't mean that work on spell checking and grammar checking should not be done. I didn't say that either. It does mean that whoever creates the spell checker and grammar checker have a hardcopy (paper) copy of the printed dictionary and grammar checker of the language in question, and understands both what in the book is contentious, and why it is contentious. Yes but these are skills different to those of a localizer. Not every localizer is a grammarian and vice verse and the same applies to the writing or even analysis of dictionaries. How many people should be on a team is as dependent on cultural factors, as it is on practical factors. Agreed The vice of a one-person team, is that there is nobody to hand off responsibility for the project off to, if the sole team member is no longer able to work on the project. Agreed. Or at least to the extent that if a single localizer fails to plan for the future once the project(s) have reached a degree of maturity. The vice of multi-member teams, is death by paralysis. The inability to come to mutually acceptable solutions, when questions/problems arise. In the corporate world in Europe and North America, researchers have found that eight people on a team, is the most effective size, for a project to be successful And my rough and ready analysis has told me you can expect about 1 active localizer for each half million speakers or so. Like you seems to get an almost invariable proportion of lurkers to participants on forums, some ratio like that also seems to apply to l10n. Which means 8 is an almost unachievable number for many languages. All the more reason to plan but also an argument for not just turning down folk because they haven't got a team. a) The primary issue with translating the help file after the rest of the UI, is that it does not get done. (Take a look at the number of localizations of LibO, where the Help file is not translated into the target language.) That is one way of looking at it. The other is that of cost vs benefit. It would be nice if someone actually did any research but using the in-built help in my experience has almost become a joke, like the tools Windows includes for automatically fixing issues. In most cases, it does not have the answer you're looking for and/or is out of date. Even commercial projects with large development teams often product shoddy Help, even for key features/issues. Like the Help on using Hunspell in Trados. To be honest I'm surprised LO has not tried to determine if the in-built help is actually worth the effort from the end-user POV in contrast to online help and/or active user forums etc. For a small team, it is certainly the smallest bang for you buck in my view. b) By starting with the Help File, one can incorporate it into the Documentation Work Flow, ensuring the documentation is consistent across the various mediums. Otherwise you end up with situations like the US English, where the written documentation and the help file contradict each other. Even worse, is when both the built-in help file, and written documentation are wrong Which makes the Help stuff an even worse place to start. The advanced/expert users of the software. No. They hit Google and find the answer on some forum, wiki or some such platform. I would bet the farm on it. I realize that Apple pioneered "Prohibit documentation wherever and whenever possible", but all that really results in, is to ensure that the user is unable to use the product as designed I'm not an Apple disciple, if that's what you're hinting at. If the software is well designed, you shouldn't have to resort to Help much. It is no more heartbreaking to translate the Help file, as it is to translate the rest of the UI, or the documentation that is in other languages, into the target language. Fundamentally disagree. The UI on the whole is not that bad and in some places, actually a lot of fun (with the Calc functions are probably the worst bit of the UI) - on the whole, they're manageable chunks and you can use TM to great effect. Help is full of big chunks with eye watering tags that confuse not only translators but also TMs. And like translating EULAs, all the while you're thinking "who's actually going to read this". It's a bit of a cliché but "more research would be welcome" :) Michael -- To unsubscribe e-mail to: l10n+unsubscr...@global.libreoffice.org Problems? http://www.libreoffice.org/get-help/mailing-lists/how-to-unsubscribe/ Posting guidelines + more: http://wiki.documentfoundation.org/Netiquette List archive: http://listarchives.libreoffice.org/global/l10n/ All messages sent to this list will be publicly archived and cannot be deleted
Re: [libreoffice-l10n] Bavarian and Nipmuck - report
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 09/12/2015 14:31, Michael Bauer wrote: > 1) Orthography > Terrible reason to turn down a project. Most l10n projects LO has invo lve languages where spelling is a potentially contentious issue. Nonetheless, every one of the 3,000 languages that have been reduced to writing has a published dictionary, and grammar. I'll grant that regardless of which language one is talking about, there will be contentious issues regarding the accuracy of the book in question. That doesn't mean that work on spell checking and grammar checking should not be done. It does mean that whoever creates the spell checker and grammar checker have a hardcopy (paper) copy of the printed dictionary and grammar checker of the language in question, and understands both what in the book is contentious, and why it is contentious. It also means that when automated tools are used to pare out foreign words, ensure that only foreign words are omitted. (How many iterations of the Afrikaans spell checker were needed, before "die" was included in it? The same thing happens in other languages, but usually aren't as blatantly obvious omissions to even the most casual user.) >Team Size > Errr no. 1 dedicated localizer is more than enough. How many people should be on a team is as dependent on cultural factors, as it is on practical factors. The vice of a one-person team, is that there is nobody to hand off responsibility for the project off to, if the sole team member is no longer able to work on the project. The vice of multi-member teams, is death by paralysis. The inability to come to mutually acceptable solutions, when questions/problems arise. In the corporate world in Europe and North America, researchers have found that eight people on a team, is the most effective size, for a project to be successful. > 5) Start with documentation/help > No.It would raise the wrong expectations, if you give the average user a screen that says Filte, unless highly cynical, they would expect the rest in the same lingo too. a) The primary issue with translating the help file after the rest of the UI, is that it does not get done. (Take a look at the number of localizations of LibO, where the Help file is not translated into the target language.) b) By starting with the Help File, one can incorporate it into the Documentation Work Flow, ensuring the documentation is consistent across the various mediums. Otherwise you end up with situations like the US English, where the written documentation and the help file contradict each other. Even worse, is when both the built-in help file, and written documentation are wrong. > As to the Help, who reads the Help? The advanced/expert users of the software. I realize that Apple pioneered "Prohibit documentation wherever and whenever possible", but all that really results in, is to ensure that the user is unable to use the product as designed. >its the worst starting point and a soul-destroying task. It is no more heartbreaking to translate the Help file, as it is to translate the rest of the UI, or the documentation that is in other languages, into the target language. jonathon -BEGIN PGP SIGNATURE- Version: GnuPG v2.0.22 (GNU/Linux) iQIcBAEBAgAGBQJWbKT8AAoJEKG7hs8nSMR70CwP/iw5yyVbNkrdwXgDuI9Z7Iw8 IzFJXk1N10uWhorLOKP2oCe/OvHeagLiOsj9vYjnIuFxPV9puHbci/u9GCgt4bYy ldcOPngm9WzJIvFwoJZkxIFogNjXrYglPmN47ak14dGEjnBsMQUMAIAMcPafx/JL dnHnDaOKmNFymvztALRMBHhHxHqBkIrZle76sdpkzwpo1ggPAiOe50utBj1sT1qn V+DsobD3LEX/fDXdUIFoOhsSB4Ko+kVvSrkpnQQvbsPkaj4o/NSw57f5J/JSaOwo W/l/a+OcqioT04rmBzJTa3PcWXuwPZ5wshpkvSCbFceXrblqd65FqYcehaE8PC3I KMDX0Ykrwuam24K6OhdCX0Q73Hqn4r5oOJovRcbWi+0lOfIV3/jiQBbdtzFF8ag3 6cB8T8nq4clcxmRv7x7lZEv3t7pkSRs8JbBezCg1sxCQbIplSSI0iWBNi8QxUzFd Fhqsf7DwPswCtGz+SdGCxPgWpBvJU6+AImjzIoQaccGd8s+RWVzbRIh0lICFtSkp kfGoqFJI/JZTUY1PtdluqmR6e6xK0SrDBm33Q8doz82nl9IwLmEsX4+PyyDCqNa4 b6LBSmkMHUUGmMpIx8xlwlehGAsgfzTq41gf4Mr0h/kvy+7DQi/I7+trIw1935Y3 61hnROdhTwlO/YYwy5+o =lj6Q -END PGP SIGNATURE- -- To unsubscribe e-mail to: l10n+unsubscr...@global.libreoffice.org Problems? http://www.libreoffice.org/get-help/mailing-lists/how-to-unsubscribe/ Posting guidelines + more: http://wiki.documentfoundation.org/Netiquette List archive: http://listarchives.libreoffice.org/global/l10n/ All messages sent to this list will be publicly archived and cannot be deleted
[libreoffice-l10n] Bavarian and Nipmuck - report
Somehow the mail client ate most of my email, reposting, sorry... --- Sorry for the delay in responding, Im travelling. I think I disagree with most things that have been said in this discussion so far. Let me try and go through them one by one... 1) Orthography Terrible reason to turn down a project. Most l10n projects LO has involve languages where spellling is a potentially contentious issue. Perhaps the really big locales have very settled spelling systems but even they are not immune. For example, I doubt that anyone is enforcing either pre or post spelling reform spellings in the German project. Some locales actually deliberately use l10n to help standardize spelling. 2) Team size Errr no. 1 dedicated locaizer is more than enough. I have a day job and I also do virtually all the l10n work on Mozilla, LO, WorPress (both), VLC, and several other projects. In fact, a single localizer can be more effective in some instances provided they put in sufficient time and effort. In fact having a team for Scottish Gaelic initially would have been a hindrance, not a help because there would have been ENDLESS debates around terminology and spelling. In a non-standardized language, a single translator can produce translations which are superior than those of a team, provided they are fluent and generally good with technology. 3) Its extinct or critically endangered Well, so is Scottish Gaelic, less than 60k speakers is hardly a stadium full of people... l10n is a key part of any revitalization effort in a society which is not cut off from technology. It is perhaps the one way in which a marginalized language can gain a foothold on the screens of the next generation, small as it may be. A program with a UI in a marginalized language has a big wow factor if done well. If you localize Diablo III into German, people just expect that, its not news. Translate it into Nipmuck and itll be all over the airwaves. Wikipedia or even Ethnologue are not the pinnacle of information when it comes to smaller languages. On several occasions have I come across languages marked as extinct in one, but not the other or vice versa or even where both were simply wrong. For example, they had a Basque Creole lumped in with a Romani language code in once instance. 4) Better to translate literature Yes and no. Im a very good localizer but Im totally useless at translating literature or poetry or songs. Its called a specialism, no translator worth their money translate EVERYTHING. Id be equally useless at writing non-technical content. 5) Start with documentation/help No.It would raise the wrong expectations, if you give the average user a screen that says Filte, unless highly cynical, they would expect the rest in the same lingo too. As to the Help, who reads the Help? Ever? Unless they dont have web access. Even if some folk use it, its the worst starting point and a soul-destroying task. 6) Professors say to prioritise proofing Maybe but that depends on the locale. To create a spellchecker you first need either really good dictionary or ody of well spelled texts, plus someone who can do code to some extent because doing a Hunspell package is not entirely straight forward. Grammar checkers are equally nice but not a priority to begin with I would say. Small languages often have not codified their grammar fully and thus if you just write some rules, youll just annoy everybody. In the end, these are just opinions. They are neither uniform (I disagree for one) not are they based on research. 7) Firefox That is actually the best alternative suggestion Ive heard in this debate. It might make sense to look into that. But either way, LO and Firefox are both must-haves really so it doesnt make that much of a difference which one you start with. Firefox, since it has Android and iOS versions now, would get you more bang for your buck faster though to begin with 8) Machine Translation Worst idea ever. MT relies on massive bilingual corpora - and thats just the start of the headaches. The last thing a language like Nipmuck needs is a MT system that cost them huge resources to produce and which outputs semi-gibberish at best. Irish is in a much better position regarding English/Irish data and yet Google Translate produces Irish which either makes you laugh yourself silly or makes you cry. Long story short, my view is, welcome to both, just have a moment to consider the implications regarding time/effort/other challenges and if you still think its a good idea, good on you. Michael -- To unsubscribe e-mail to: l10n+unsubscr...@global.libreoffice.org Problems? http://www.libreoffice.org/get-help/mailing-lists/how-to-unsubscribe/ Posting guidelines + more: http://wiki.documentfoundation.org/Netiquette List archive: http://listarchives.libreoffice.org/global/l10n/ All messages sent to this list will be publicly archived and cannot be deleted