Talk:WikiTranslate

From Meta, a Wikimedia project coordination wiki

Multilingual idiomatic dictionary[edit]

I understand this proposal is about creating a multilingual idiomatic dictionary, right? This would certainly be a good thing to have. At the same time, the implementation seen at the test project is not yet convincing. There, you have a separate pages for every translation, e.g "je suis", "I am" and "Ich bin" (I'd rather say "ich heiße", btw) are separate pages. That means that if I want to add another language, I would have to

  1. create a new page for the new language; and
  2. edit all the existing pages

This may be tolerable as long as you have no more than, say, five languages, but this is not scalable. Imagine, you have "Good morning" in 20 languages - That would mean twenty different pages, like de:Guten Morgen, ru:Доброе утро etc. pp. If you want to add another language, you would not only have to add another page but also edit 20 existing pages. This is clearly not feasible. There should be a better technical solution in place. It may well be, that mediawiki is not the best available option here. --Johannes Rohr 11:55, 18 April 2007 (UTC)[reply]

You are right. Maybe:
  1. Add different pages with one translation template, than you only have to edit that template and create a new page.
  2. Or from the translation a redirect to the English translation, than we only have to edit that page and also add a new redirect.
What do you think? SPQRobin 12:14, 18 April 2007 (UTC)[reply]
I think that this will take much more research and probably also development effort. Such a project, if taken serious, should be planned for scalability and maintainability from the ground up. Redirects are a quick-and-dirty fix, but they are clearly not maintainable. If such a project grows, things are bound to get messy. I can only reaffirm, if this project is going to get serious, a dictionary-specific solution should be envisaged. Instead of setting manual redirects, a database-like solution should be used, where all translations are stored in metadata, so that the software can generate pages like "Je suis", "Ich bin" or "I am" on-the-fly from the same set of data.
Furthermore, consideration has to be given to ambiguities. If you take the simple phrases in the test project, it is striking that even the simple English phrase "I am" has more than one possible translation to other languages, depending on the context. When "I am" is used in the meaning of "My name is", the preferred German translation would be "Ich heiße", for the generic "I am" (e.g. I am a doctor) I would say "Ich bin". So there would have to be a web-like solution allowing for assymetric, multi-directional relations between phrases. I don't think a vanilla mediawiki installation can satisfy these requirements. --Johannes Rohr 13:41, 18 April 2007 (UTC)[reply]
I know, putting redirects isn't a good solution. And what do you think about a template for the translations? About the example that you said: If there are different contexts, we can add (like in the wiktionary) a little bit information. I have made an example, http://editthis.info/translatewiki/Ik_ben. SPQRobin 17:12, 18 April 2007 (UTC)[reply]
It's not just different contexts, but also other very minor differences with big consequences: on the test site I saw the sentence "Happy birthday". The correct Dutch translation would be "Gefeliciteerd met je verjaardag" or, more formal, "Gefeliciteerd met uw verjaardag". In English both 'je' and 'uw' are 'your'. This would mean that every single sentence that has 'you' or 'your' in it, has two Dutch translations. How manageable is that? Fruggo 18:59, 20 April 2007 (UTC)[reply]
I have an idea: We can add a link to each word/sentence with that problem, and that link goes to a page where it is explained. SPQRobin 20:00, 20 April 2007 (UTC) (Eigenlijk wel grappig, we praten allebei Nederlands maar zijn in het Engels tegen elkaar aan het praten :-) )[reply]

Wikitravel Phrasebooks[edit]

Wikitravel.org have developed (/are developing) a set of phrasebooks:

http://wikitravel.org/en/Wikitravel:List_of_phrasebooks

These are aimed at giving travellers some translations of basic phrases. I really like this idea, so I want to point you guys at it. I appreciate it's not really the same idea as this proposed project, but if any of you are bursting to do some translating, feel free to contribute!

Unfortunately wikitravel is under a different license, so you strictly speaking you're not allowed to take any of the text from there, to use within this project.

-- Harry Wood 10:23, 20 April 2007 (UTC)[reply]

I don't know if we can do something with that phrasebook, also because there is a different license. Although it is a good idea. SPQRobin 20:03, 20 April 2007 (UTC)[reply]

Wasn't wikitranslate his idea? See his user page (or the one linked to in the (in Firefox: blinking) text in the turquoise box).--213.168.118.215 18:04, 5 June 2007 (UTC)[reply]

It could be. I think there are much people with an idea like this. I will ask him for joining this idea. SPQRobin 18:43, 5 June 2007 (UTC)[reply]
I will! See your disc page on de.wikip. THX!--213.168.111.227 10:07, 7 June 2007 (UTC)[reply]

OmegaWiki?[edit]

Have you considered participating in OmegaWiki?

(English) A collaborative project to produce a free, multilingual resource in every language, with lexicological, terminological and thesaurus information.

(asturianu) Un proyeutu collaborativu pa crear un recursu llibre y plurillingüe con información lexicolóxica, terminolóxica y sinónimos en cada llingua

(Bokmål) Et samarbeidsprosjekt for å lage en fri, flerspråklig ordbok på alle språk.

(Brezhoneg) Ur raktres stroll zo e bal sevel ur vammenn ditouriñ digoust ha liesyezhek en holl yezhoù, ennañ titouroù geriaouel ha termenadurel hag un tezorus.

(български) един съвместен проект, целящ създаването на свободна, многоезична база данни на всички езици, която да съдържа лексикологична, терминологична и речникова информация.

(Català) Un projecte de col·laboració per tal de crear un recurs lliure i multilingüe amb informació lexicològica, terminològica i tesaurus en cada idioma.

(česky) Kolaborativním projektu, který má za cíl vytvořit volný, mnohojazyčný zdroj ve všech jazycích s lexikologickými, terminologickými a slovníkovými informacemi.

(Deutsch) Einem gemeinschaftlichen Projekt mit dem Ziel, eine freie, mehrsprachige Datenbank für den Wortschatz aller Sprachen mit lexikografischen und terminologischen Informationen zu schaffen.

(ελληνικά) ένα έργο συνεργασίας για την παραγωγή ενός ελεύθερου και πολύγλωσσου βοηθήματος σε όλες τις γλώσσες, με πληροφορίες πάνω στις λέξεις, τους ορισμούς τους, την ιστορία τους.

(español) Un proyecto de colaboración para crear un recurso libre y plurilingüe con información lexicológica, terminológica y tesauros en cada idioma.

(Esperanto) Kunlabora projekto celante produkti liberan multlingvan vortaron en ĉiuj lingvoj, kun leksilogiaj terminologiaj informoj kaj tezaŭro.

(français) Ceci est un projet collaboratif visant à produire un dictionnaire libre et multilingue dans toutes les langues, avec des informations lexicologiques, terminologiques, et de thésaurus.

() פרויקט שיתופי ליצירת מקור חופשי רב-לשוני בכל שפה, הכולל מידע לקסיקוגרפי, טרמינולוגיה ותזאורוס.

(Nederlands) Een samenwerkingsproject om een vrije, meertalige bron te maken in iedere taal, met lexicologische, terminologische en thesaurusinformatie.

And many more languages...

The project can be found at http://www.omegawiki.org. Siebrand 01:34, 14 September 2007 (UTC)[reply]

Many related projects; see also the archives. +sj | help with translation |+ 02:07, 23 August 2008 (UTC)[reply]

  • Some points:
    1. Note that for finding simple words translations, you absolutely don't need to create a new wiki, Wiktionary is enough for that (and there already exists external tools that can look for words in all Wiktionaries).
    2. Ideally, this wiki should have specific extensions, to really take into account the recognition of grammar, and allowing reordering sentences or finding inflexions where appropriate. Such extensions should also be able to get data from the existing Wiktionaries, with an internal cache for parsed and indexed Wiktionary articles.
    3. In addition, I would not name your project "TranslateWiki" but "TranslatorWiki" (once it will be able to translate sentences, rather than just simple words), which would more appropriate and would avoid the confusion with BetaWiki (which is also known as TranslateWiki, but related to the localization of MediaWiki, its extensions, and some other open-source projects, and that should really have a better name than "Beta", probably something like "LocalizeWiki").
    verdy_p 16:14, 26 August 2009 (UTC)[reply]