Gutza wrote: Wow. I'm really not the only one who thinks that? :-) OK, here are my major problems with Wiktionary: * English-centric (as has already been said) * not automated enough. Linking from [[biscuit]] (an English word) to [[Keks]] (a German word) should automatically add a link from [[Keks]] back to [[biscuit]]. A lot of identical formatting/layout should be automated (perhaps have a template automatically show up in the edit window when you try to edit a not-yet-existing page?). * too crowded in places. [[e]] should not contain all its meanings in hundreds of different languages on a single page. My suggestions for major improvement would be: * wait for MediaWiki to support multi-language projects within a single Wiki. This is already in the plans for Wikipedia. Once that is done, you can have one page [[e]] in Spanish, another [[e]] in Italian, etc. * add more options to the MediaWiki software to customise the appearance and behaviour, including: * * allow for the inter-wiki links (which should really be called inter-language links once it's a single wiki) to be displayed in a completely different fashion, i.e. instead of: _dansk_ | _dansk_ | _Deutsch_ | _English_ | _English_ | _Français_ at the top and bottom, display them as a dominantly visible list only at the bottom: == Translations == * Dansk: _kiks_, _biskuit_ * Deutsch: _Keks_ * English: _biscuit_, _cookie_ * Français: _biscuit_ * * allow for automatic reciprocal adding of inter-language links * make one major change of policy/aim/goal. I think it is completely redundant and futile to try to define (explain the meaning of) all words of all languages *in* all languages. I think a word should only be defined in its own language, and if you want its meaning explained in another language, then you should really look up its translation in that other language. There are just so many things on Wiktionary that are language-independent (e.g. the translations for each word, or pronunciation written in IPA) that duplicating it hundreds of times seems really dumb. So far are my ideas. Timwi

Peter Jaros wrote: I don't think that's a problem at all. If you know Hoobiflitz well enough to appreciate the subtleties of the language, then you can just read the definition of guba in Hoobiflitz. If you don't, then you wouldn't be able to appreciate the subtle implications of a word anyway, so a few suggested translations can give you the general idea. Besides, several inaccurate translations often give you a more accurate picture of a word's meaning than a single inaccurate translation. Suppose, for example, the three possible translations for guba were "offer", "sale", "promotion". The combination of the three gives you more information than each single one. What I'm getting at here is that having a list of well-chosen translations is as good as a definition. Timwi

Ray Saintonge wrote: Really? Here's what I tried: clicked on "Random Page" 10 times. Here are the results: 1. http://wiktionary.org/wiki/Pronunciaci%C3%B3n 2. http://wiktionary.org/wiki/%E3%92%AD 3. http://wiktionary.org/wiki/Mayoress 4. http://wiktionary.org/wiki/Bijutel 5. http://wiktionary.org/wiki/Decibel 6. http://wiktionary.org/wiki/Balul 7. http://wiktionary.org/wiki/%E6%97%86 8. http://wiktionary.org/wiki/Airtight 9. http://wiktionary.org/wiki/Brush 10. http://wiktionary.org/wiki/%E9%95%9A There is not a single one hit in the ten above which isn't appropriate for automated parsing: * No. 1 would be reverse translated from Spanish to English (note that http://wiktionary.org/wiki/Pronounciation doesn't exist). * No. 2, 3, 7 and 10 would be completely skipped in parsing because they don't contain any proper translations. * No. 4 and 6 would be reverse translated from Volapük to English (note there is no mention of Balul at http://wiktionary.org/wiki/January, and http://wiktionary.org/wiki/Jeweller doesn't exist). * No. 5, 8 and 9 would be properly translated to the respective languages for which translations are available. What would be wrong with this? What counter-examples do you have in mind? --Gutza

Arwel Parry wrote: Ok, maybe I didn't get the concept across really well. You are aware that we didn't get to that point in history where we humans all lay back and expect automatons to do all the work for us, right? Writing a parser to create the proper cross-links between definitions in the English Wiktionary and other languages' Wiktionaries would not be expected to result in complete dictionaries for all other languages. But the same happens with human contributors: you happen to know that "glas" has those two meanings, but some other contributor may not. Some other contributor may write the "glas" article in the Welsh Wiktionary and only include the first definition. If you happened to come across that article, you might add the other definition as well. I don't see why we need that human contributor at all if "glas" were to be found on the English Wiktionary article on "sky blue", for instance. Not to mention that if "glas" also showed up on "grass green" then both definitions would be available for the Welsh "glas", using a smart automated parser. In any case, the point is not to expect the automated parser to result in a prefect any language to any language dictionary, but rather to do what automated things tend to do today: make our lives easier, as in creating a reasonable basis for other Wiktionaries, as well as keeping things in sync. Just my 2c, Gutza

On Jan 8, 2004, at 9:43 AM, Sascha Noyes wrote: Take, specifically, the Spanish 'libre', the English 'free', and the French 'gratuit'. Without care, one might end up saying that since sp:'libre' means en:'free', which means fr:'gratuit', sp:'libre' also means fr:'gratuit', which is of course wrong. The transitive property does not always apply to language. Here out little automaton friend might run into difficulty, even if he *is* only aiming to "make our lives easier". Peter --- Funding for this program comes from Borders without Doctors: The Bookstore Chain That Sounds Like a Charity. --Harry Shearer, Le Show

Gutza wrote: Most of the above have their own problems. 1. Spanish. It at first seems like a simple matter of using the cognate "pronunciation", but in Law it can also mean "publication". Coming back from English there is also the usage at a wedding where the parties are "pronounced husband and wife" 2. An elder brother. Not all languages have special words to differentiate between an older and younger brother. Just because there is no translation there now, does not mean that it will never be added. 3. Are we talking about a female mayor, or the mayor's wife. 4. Two meanings are given: the jeweller (or jeweler in the U.S.) and his shop. How do we determine which is meant. English does not have a proper single word for the latter. At least French has "bijoutier" and "bijouterie" 5. This is the most stable of the lot because it is defined by international standards. 6. Probably stable. Names of months usually are unless we are talking about diferent calenders, which would be stretching the issue even for me. 7. This one relates to pennants or streamers, or even flags, but not in the sense of national flags. 8. The literal meaning seems obvious, but one still has to allow for the more metaphorical, as in "The lawyer had an airtight case." 9. The ordinary meaning might seem easy enough, but a brush in an electric motor would not be translated the same way. 10.It can mean a cloth for carrying a baby on one's back, or as a verb to "tie" or "bind" So this leaves me with only two out of ten where one does not run into ambiguities. This is far too much ambiguity for automated translations to be a viable option. As US President Kennedy once said, "I am a jelly doughnut". Ec