menu
Tatoeba
language
Register Log in
language English
menu
Tatoeba

chevron_right Register

chevron_right Log in

Browse

chevron_right Show random sentence

chevron_right Browse by language

chevron_right Browse by list

chevron_right Browse by tag

chevron_right Browse audio

Community

chevron_right Wall

chevron_right List of all members

chevron_right Languages of members

chevron_right Native speakers

search
clear
swap_horiz
search
sacredceltic sacredceltic June 3, 2011 June 3, 2011 at 11:15:23 AM UTC link Permalink

With 535 English sentences, we now know pretty much everything about John's life http://tatoeba.org/epo/sentence...rom=eng&to=und

{{vm.hiddenReplies[6454] ? 'expand_more' : 'expand_less'}} hide replies show replies
papabear papabear June 3, 2011 June 3, 2011 at 11:58:04 AM UTC link Permalink

It's not something I can take to its logical extreme, but I try to mix up the ethnicities of names I use. I wonder if lists of baby names around the world might help.

{{vm.hiddenReplies[6455] ? 'expand_more' : 'expand_less'}} hide replies show replies
sacredceltic sacredceltic June 3, 2011 June 3, 2011 at 12:10:02 PM UTC link Permalink

The problem, if you use various names, is that we end up with dozens of variants of "X is not as tall as Y" http://tatoeba.org/epo/sentence...rom=eng&to=und

I think that the multiplicity of these similar sentences with different persons names have no interest as example sentences and they create useless clutter.
We should find a way to deduplicate them. Of course, this implies either using standardised names (which would probably raise a big debate), or replace names with <filling holes>, which I would favour.
We could conceive a system, by which the sentence is stored and edited in a generic way without the names/with the holes, but when viewed, for the purpose of appearing natural, the holes could be filled with random names. Just my twopence...

{{vm.hiddenReplies[6458] ? 'expand_more' : 'expand_less'}} hide replies show replies
papabear papabear June 3, 2011 June 3, 2011 at 12:19:58 PM UTC link Permalink

Might that make it more difficult to add sentences, parse corpus sentences by computer, and have audio sentences match their text?

I do agree with your basic premise, however, and I have a couple more ideas we could explore the pros and cons of:

* A moratorium on adding sentences of a certain length in well-established languages. Ex: No more sentences of less than three words in English.
* A heuristic anti-duplication algorithm that forbids not just whole duplicate sentences, but near-duplicates.

Shiawase Shiawase June 4, 2011 June 4, 2011 at 2:54:53 PM UTC link Permalink

I've no problem with a set of generic names. I think it's actually quite useful in the context of example sentences.
Purely based on the number of mangled mail-merged messages I've seen, I'm not sure it's trivial to script properly for name placeholders. I think there's more useful things that the time could be spent on first.

On the further points by sacredceltic and papabear, it might be worth drafting some style guides for the sort of example sentences that are useful and tips for avoiding duplication (or near duplication with no good purpose).

sacredceltic sacredceltic June 3, 2011 June 3, 2011 at 2:40:36 PM UTC link Permalink

After investigations, it appears that we have even more insight on Tom's life than on John's: http://tatoeba.org/epo/sentence...rom=eng&to=und

{{vm.hiddenReplies[6462] ? 'expand_more' : 'expand_less'}} hide replies show replies
xtofu80 xtofu80 June 4, 2011 June 4, 2011 at 8:02:27 AM UTC link Permalink

From a computer science point of view, replacing names with generic tags certainly makes sense.
What I am worried about: are there languages which inflect names, for example changing a names ending if it is in the accusative?
This would complicate matters of course.

{{vm.hiddenReplies[6467] ? 'expand_more' : 'expand_less'}} hide replies show replies
Fingerhut Fingerhut June 4, 2011 June 4, 2011 at 8:50:48 AM UTC link Permalink

Well in German, you add a 's' in the genitive form ('Johns' instead of 'John'), you even encounter more severe changings ('Fritzens' instead of 'Fritz') - that might not be perfect standard, but have always been used quite often by many people, so you can't be sure you won't encounter that in sample sentences here (especially if people take sentences out of novels). And in both English and German you tend to add an 's' in the plural ('The Hansons', 'Die Mahlers'). I'm pretty sure you do similar things in other languages as well.

{{vm.hiddenReplies[6468] ? 'expand_more' : 'expand_less'}} hide replies show replies
sacredceltic sacredceltic June 4, 2011 June 4, 2011 at 4:32:56 PM UTC link Permalink

Ok, but if you replace Kurt by Otto, which is the point here, that wouldn't change, would it?

In French, names may affect the possessive "de" if they start with a vowel:

"le vélo de Pierre" but "le vélo d'Alain"
"La robe de Marie" but "la robe d'Anne"

But that rule is regular and requires just a test to determine whether the name's first letter is a vowel in this case.

{{vm.hiddenReplies[6472] ? 'expand_more' : 'expand_less'}} hide replies show replies
Fingerhut Fingerhut June 4, 2011 June 4, 2011 at 5:21:42 PM UTC link Permalink

Well, in German, it changes, when the name ends with an 's', 'x' or 'z':
"Ottos Kind" but "Max' (or, sometimes, Maxens) Kind"

Maybe one should just give out the rule, that every male is called 'Ben' here and every female 'Hanako'. Or something like that.

{{vm.hiddenReplies[6474] ? 'expand_more' : 'expand_less'}} hide replies show replies
sacredceltic sacredceltic June 4, 2011 June 4, 2011 at 9:04:33 PM UTC link Permalink

>Maybe one should just give out the rule, that every male is called 'Ben' here and every female 'Hanako'. Or something like that.

Yes, I thought about this, but it would trigger endless wars to decide which names should be retained, and at the same time, we need to illustrate all different linguistic cases, including this s/x/z german rule which is most interesting...

sysko sysko June 4, 2011 June 4, 2011 at 9:30:40 PM UTC link Permalink

a vowel or a h actually (le cheval d'henri IV)

{{vm.hiddenReplies[6478] ? 'expand_more' : 'expand_less'}} hide replies show replies
sacredceltic sacredceltic June 4, 2011 June 4, 2011 at 9:32:13 PM UTC link Permalink

de 'h' muet devant une voyelle...

Hans_Adler Hans_Adler June 7, 2011 June 7, 2011 at 10:58:23 PM UTC link Permalink

In languages with vowel harmony (e.g. Finnish, Estonian, Hungarian, Turkish) any inflections of a name will usually depend on the quality of the vowels in the name. I am not sufficiently familiar with any of these languages, but I would be surprised if it were impossible to build plurals or accusatives of names in all of them. The precise choice of name can also have a subtle influence on translations, as it may imply a setting (e.g. Tom vs. Julius Caesar vs. Ossama bin Laden).

{{vm.hiddenReplies[6550] ? 'expand_more' : 'expand_less'}} hide replies show replies
sacredceltic sacredceltic June 7, 2011 June 7, 2011 at 11:26:36 PM UTC link Permalink

Excluding certain names so they are not replaced by markers is not so difficult.

First "Julius" or "Napoleon" or "Barack" or "Ossama" are not so popular in example sentences for sentences where the names are indifferent (let's call them ahistorical sentences, as opposed to historical ones), so these firstnames would just not be in the list of common firstnames.
Again, the system I propose is not intended to replace all names by markers, but just to avoid clutter from duplicates with common names in ahistorical sentences. So it doesn't matter if does the job only 90% of the time, because it is still very positive from that point of view.

In the case when a historical character has a common name, there are several solutions to the problem:

"John loves Elizabeth" => no context, no family names, "John" and "Elizabeth" can easily be replaced by markers and the sentence can become generic.

"John Wayne loves Elizabeth II" => here, at least in English, the firstnames (from our list of common names) are followed by words starting with a capital letter, indicating a probable second name => stop of procedure = no replacement by markers.

In at least a number of other languages, there must be similar ways, more or less easy to determine that we're dealing with a specific character in a historical sentence.

Finally, in case of difficulty, such as "Winny the pooh" or "Billy the kid", we could use an exception list of personalities to be excluded from the procedure...

Demetrius Demetrius June 4, 2011 June 4, 2011 at 11:25:11 AM UTC link Permalink

[БЕЛ] Я асабіста не падтрымліваю ідэю замяняць імёны на паметкі. Звычайна, гэта можа палегчыць частку задач для праграмістаў, але, наколькі я разумею, «неапрацованыя сказы» заўсёды былі часткай ідэалогіі Татоэбы.

Калі ўвясці паметкі замест імёнаў, удзельнікі павінны будуць выконваць «папярэднюю апрацоўку» сваіх сказаў. Удзел у праэкце стане складнейшым, бо ён будзе патрабаваць болей папярэдніх ведаў.

Бягучая сістэма і так не вельмі інтуітыўная (і колькасць навічкоў, якія дадаюць анатацыі, таму доказ), а паметкі замест імёнаў зробяць яе яшчэ менш інтуітыўнай.

[УКР] Я власно не підтримую ідею заміняти імена на мітки. Звичайно, це полекшить частку завдань для програмістів, але, наскільки я розумію, «неопрацьовані речення» завжди були часткою ідеології Татоеби.

Якщо ввести мітки замість імен, учасники мусимуть виконувати «попередню обробку» своїх речень. Участь у проекті стане складніша, адже вона буде потребувати більше попередніх знань.

Поточна система і так не дуже інтуітивна (і кількість новачків, що добавляють анотації, тому доказ), а мітки замість імен тільки зроблять її ще менш інтуітивною.

[РУС] Я лично не поддерживаю идею заменять имена на метки. Конечно, это может облегчить часть задач для программистов, но, насколько я понимаю, «необработанные предложения» всегда были частью идеологии Татоэбы.

Если ввести пометки вместо имён, участникам надо будет производить «предобработку» своих предложений. Это усложнит участие в проекте, потому что оно будет требовать больше предварительных знаний.

Текущая система и так не очень интуитивна (и количество новичков, которые добавляю аннотации, тому подтверждение), а пометки вместо имён ещё больше усложнят её.

[АНГ] I am personally not in favour of name placeholders. Of course it can make some processing tasks easier, but keeping sentences “as raw as possible” seems to have always been an important part of Tatoeba philosophy.

Implementing name placeholders forces users to ‘preprocess’ their sentences. It will make contributing harder, because people have to learn more things before ding anything here.

The current system is not intuitive enough (and the number of new contributors adding annotations proves it), and name placeholders will make it even less intuitive.

{{vm.hiddenReplies[6469] ? 'expand_more' : 'expand_less'}} hide replies show replies
sacredceltic sacredceltic June 4, 2011 June 4, 2011 at 4:51:02 PM UTC link Permalink

>Implementing name placeholders forces users to ‘preprocess’ their sentences. It will make contributing harder, because people have to learn more things before ding anything here.

But it could be a transparent feature for the contributor:

The contributor could type "Jack loves Gill"
Before inserting the sentence in the database, a procedure could then lookup the typed words in a list of possible names for the given language and ascertain they are names (they might be a few ambiguities...)
When the names have been identified, the procedure would replace them with the generic markers and the sentence would be inserted with them in database (<name1> loves <Name2>
After the insertion, the sentence would appear to the contributor as it has been entered, so the contributor would not even be aware of the procedure.

Finally, everytime the sentence with the markers is displayed, the markers are replaced by random names from the names list for the relevant language.

So the only drawback would be that when the contibutor will display another time the sentence he created, he would see random names instead of those he entered, but I think it is a very light drawback that everyone would understand...

We don't need to have a mecanism that includes all existing names.
It could handle just the most frequent (Tom, John, Jack, Mary / Smith, Brown, Watson...for instance in English)
We could also devise a mecanism to enrich these lists automatically based on the frequency analysis of these names.

Cherry on top, there could be an option to prevent the procedure to execute in certain cases, when names must remain originals in the case of personalities for instance (it could also lookup a list of personalities to exclude...)

FeuDRenais FeuDRenais June 4, 2011 June 4, 2011 at 8:44:58 PM UTC link Permalink

Yes. Slavic languages have quite a few declensions for names. Turkish ones as well, and then you have things like vowel harmony to worry about.

Personally, I don't much like the idea of "efficient names". If it's a problem of database size, then sure. Otherwise, optimizing presentation methods would be a better approach to the problem being discussed, IMO.

{{vm.hiddenReplies[6475] ? 'expand_more' : 'expand_less'}} hide replies show replies
sacredceltic sacredceltic June 4, 2011 June 4, 2011 at 9:10:22 PM UTC link Permalink

>optimizing presentation methods would be a better approach to the problem being discussed

What do you mean?

{{vm.hiddenReplies[6477] ? 'expand_more' : 'expand_less'}} hide replies show replies
FeuDRenais FeuDRenais June 4, 2011 June 4, 2011 at 10:23:09 PM UTC link Permalink

I mean automate search results to not display similar sentences in the same language.

{{vm.hiddenReplies[6480] ? 'expand_more' : 'expand_less'}} hide replies show replies
FeuDRenais FeuDRenais June 4, 2011 June 4, 2011 at 10:28:54 PM UTC link Permalink

Or rather, when given 100 search results, optimize it so that the first 10 have the greatest variety and are the most mutually diverse (with diversity decreasing as you go up the pages).

That would require some math and some criteria to characterize what is "diverse" with respect to one another (i.e. ranking of importance for different vocabulary, with names having lowest importance), but it would be an elegant way to do it.

{{vm.hiddenReplies[6481] ? 'expand_more' : 'expand_less'}} hide replies show replies
sacredceltic sacredceltic June 5, 2011 June 5, 2011 at 9:16:44 AM UTC link Permalink

That's an interesting idea, but it doesn't reduce the clutter. For instance, when I'm reviewing untranslated sentences, I always bump into these boring repetitions...It also affects statistics because near duplicates are counted individually.

But I agree that techniques for rendering a set of sentences as varied as possible are certainly most interesting and desirable.

{{vm.hiddenReplies[6484] ? 'expand_more' : 'expand_less'}} hide replies show replies
FeuDRenais FeuDRenais June 5, 2011 June 5, 2011 at 12:51:18 PM UTC link Permalink

It would be nice to have a "similar sentences" tab somewhere on the side while you're viewing a particular sentence.

{{vm.hiddenReplies[6485] ? 'expand_more' : 'expand_less'}} hide replies show replies
jakov jakov June 5, 2011 June 5, 2011 at 5:57:20 PM UTC link Permalink

+1

xtofu80 xtofu80 June 4, 2011 June 4, 2011 at 10:43:27 PM UTC link Permalink

In the case of languages where the names are changed according to case, etc., a simple replace method will not work anymore, as the computer would in theory need to know the case the name is in. If a computer can then in practice change a concrete name into a special form is still another issue, and this would have to be implemented in all languages. I therefore doubt this can be realized.

xtofu80 xtofu80 June 4, 2011 June 4, 2011 at 10:52:27 PM UTC link Permalink

There are two issues at stake here:
1) People are creating many variations of the same sentence. This is especially true for simple sentences.
2) Variations naturally evolve by translation, if one word in one language can be translated differently into another language. If two translations are created, then these two are again translated into many other languages, etc.

The question remains whether all of these variants are helpful / relevant and if not, how we could avoid these.

Personally, when encountering sentence variants in Japanese which I translate into German, I either link them to the same sentence in German if they mean exactly the same, or I just translate one and then go on to other, more interesting sentences.

jakov jakov June 5, 2011 June 5, 2011 at 6:15:12 PM UTC link Permalink

Mi kontraŭas meti kodon en la tekstojn, sed mi pensas ke marki la nomojn eble faciligus al algoritmoj trovi similajn frazojn. Tamen estus aldona tasko, do eble prefere la nomoj automate rekoniĝus el listo: do estu "opt out".

slomox slomox June 6, 2011 June 6, 2011 at 3:59:29 PM UTC link Permalink

Well, there are of course many languages that inflect names. The Latvian name for "Bertolt Brecht" is "Bertolts Brehts" and the last name can change to for example "Brehtu" or "Brehta" depending on the grammatical case. See http://lv.wikipedia.org/wiki/Bertolts_Brehts The same inflection is also relevant for first names, place names etc.

So, if we want to translate "<male_first_name> was ..." we need to indicate the case for Latvian: "<male_first_name:nominative> bija ..."

And of course we need a collection of rules or even a case-by-case table which name matches which Latvian name form, the inflection forms etc.

In agglutinating languages we may need to put even more info as parameters. Perhaps the tense of the verb also changes the suffix of the name.

We need to discern male and female forms and if we start with first names, we could do the same with last names (we have probably many "Mr Smith/Jones/Walker is ill." sentences), city names ("Rome/Paris/Tokyo is a beautiful city."), or anything ("I own a horse/pencil/car.", "I love/hate/hit/cuddle you."). There's always redundancy.

We may also need frequency data for names, because it would be odd if the placeholder name "Petronella-Gunhilde" shows up just as often as "Tom", "John" or "Anne".

Hm, when I started writing this post I actually liked the idea. Now I think it's not feasible. Too many grammatical pitfalls.

{{vm.hiddenReplies[6495] ? 'expand_more' : 'expand_less'}} hide replies show replies
sacredceltic sacredceltic June 6, 2011 June 6, 2011 at 4:20:42 PM UTC link Permalink

I agree it is a difficult issue, but my feeling is that it is feasible and worth it, at least for firstnames.

We could start with languages where inflections are non-existent as in English or are simple, such as German or French.

I like the idea of "Petronella-Gunhilde" to replace boring Mary or Sally...
But seriously, my proposition was to create lists of names per language. And I don't think rare names need to be included because the probability to create duplicates with them is very small.
Actually, if somebody was to create a sentence with "Petronella-Gunhilde", the chance is that this choice of name would be purposeful. If it suddenly becomes a popular firstname in the service, then we can always add it to the lists.
Actually, a frequency analysis procedure could automatically enrich the lists.
Only common names like "Tom/Mary/Kumiko..." are used without purpose just to create example sentences, especially by newcommers who want to experience with the system, you will notice.

As for names of places, it is another issue, because you cannot always replace one by the other when the sentences refer to specific characteristics of these places, otherwise you would end up with "Mount Fuji is the highest mountain in Italy" or other such nonsense...

{{vm.hiddenReplies[6496] ? 'expand_more' : 'expand_less'}} hide replies show replies
slomox slomox June 6, 2011 June 6, 2011 at 4:33:37 PM UTC link Permalink

>We could start with languages where inflections are
>non-existent as in English or are simple, such as German
>or French.

But that would render these sentences untranslatable into languages that need to make more complex distinctions.

If we want to implement it it must work in all languages.

{{vm.hiddenReplies[6497] ? 'expand_more' : 'expand_less'}} hide replies show replies
sacredceltic sacredceltic June 6, 2011 June 6, 2011 at 4:38:59 PM UTC link Permalink

>But that would render these sentences untranslatable into languages that need to make more complex distinctions.

No, because, as I explained, the sentences would always display as "normal" with random names filled in. So nothing would prevent normal translation in other languages.
Only the encoding would be special for these languages, but that would be completely transparent to all the users, except to the sentences' authors, who would see random names in place of those they inserted...

Thanks to this transparency, we could have languages for which the mechanism would be ON, while it would be OFF for others. Actually, it could be ON for all, except the names lists would be filled for some and empty for others...

{{vm.hiddenReplies[6498] ? 'expand_more' : 'expand_less'}} hide replies show replies
slomox slomox June 6, 2011 June 6, 2011 at 5:14:46 PM UTC link Permalink

Okay, if the English sentence contains "<name1> hits <name2> in the face." what would the Latvian translation contain? Using "<name2>" in the Latvian sentence would render grammatically incorrect. Using an actual name in the translation wouldn't work too, because they are not the same. The Latvian user may got presented the placeholder name "William" and translated it as "Williams". Another user using another inflecting language got presented the placeholder name "Carl" and translated it as "Carlas". "Williams" and "Carlas" would appear as indirect translations of each other. How would you avoid that?

{{vm.hiddenReplies[6500] ? 'expand_more' : 'expand_less'}} hide replies show replies
sacredceltic sacredceltic June 6, 2011 June 6, 2011 at 5:31:13 PM UTC link Permalink

Suppose I am the author of the sentence in English (for which language the mechanism would be ON) and you're the translator in Latvian (for which language the mechanism is OFF).

I enter the sentence: "Jack loves Jill".
The system stores my sentence as "<name1> loves <name2>" because the mechanism is ON for English and "Jack" and "Jill" have been detected in the associated list.
You view my sentence as "Paul loves Robert" because <name1> and <name2> are automatically replaced by random names from the same list.
You translate it to "Plszxw liuwzq Rabyyuixed" in Latvian (I'm sorry I don't speak Latvian...)
The system stores your Latvian sentence NORMALLY (because the mechanism is off for Latvian) as you keyed it : "Plszxw liuwzq Rabyyuixed"
When I view your Latvian translation, I see the same : "Plszxw liuwzq Rabyyuixed".

{{vm.hiddenReplies[6502] ? 'expand_more' : 'expand_less'}} hide replies show replies
slomox slomox June 6, 2011 June 6, 2011 at 9:49:48 PM UTC link Permalink

But what would the English translation look like when I view the Latvian sentence?

{{vm.hiddenReplies[6507] ? 'expand_more' : 'expand_less'}} hide replies show replies
sacredceltic sacredceltic June 6, 2011 June 6, 2011 at 10:10:06 PM UTC link Permalink

>But what would the English translation look like when I view the Latvian sentence?

I'm not sure I understand your concern...

The display of the English sentence and of the Latvian sentence would remain independant as they are now.
If I want to translate "Carla is lovely" in English in "Charlotte est adorable" in French, I can already do that...
Firstnames don't translate into each other across cultures anyway...

{{vm.hiddenReplies[6508] ? 'expand_more' : 'expand_less'}} hide replies show replies
slomox slomox June 6, 2011 June 6, 2011 at 11:00:13 PM UTC link Permalink

>The display of the English sentence and of
>the Latvian sentence would remain
>independant as they are now.

When I view a Latvian sentence with an English translation, the Latvian sentence is printed in big, the English sentence follows with a green arrow and somewhat smaller and after that come indirect translations in small and grey. What would the second line say if the first line said "Plszxw liuwzq Rabyyuixed"?

{{vm.hiddenReplies[6511] ? 'expand_more' : 'expand_less'}} hide replies show replies
sacredceltic sacredceltic June 6, 2011 June 6, 2011 at 11:08:06 PM UTC link Permalink

So if you translated "Jack loves Jill" into Latvian "Plszxw liuwzq Rabyyuixed", when you look at you Latvian sentence, you would see an English translation with random names (keeping males and females apart for the purpose of grammar)

So one time you would see "Tom loves Mary" and the next time you would see "Jack loves Sally"...

sacredceltic sacredceltic June 6, 2011 June 6, 2011 at 5:44:43 PM UTC link Permalink

Now for inflections, we would have to implement rules.
Example in French:
we have "Paul porte le manteau de Pierre"
but
"Paul porte le manteau d'Alain" => "de " is replaced by "d'" in front of a vowel (and of a mute "h", this is another subproblem...let's stick to vowels now for the purpose of the demonstration)

There are 2 storage solutions :
1) the system stores "<name1> porte le manteau d'<name2>"
When markers are replaced, the pocedure analyses that there is a "d'" in front of <name2>, and thus, it should pick a random name that starts with a vowel.

2) the system stores "<name1> porte le manteau <possessive marker><name2>"
and depending on the random <name2> that is chosen, <possessive marker> would be replaced by either "d'" or "de "

Solution 2 is much better as it enables storing only one sentence for 2 forms.

boracasli boracasli June 6, 2011 June 6, 2011 at 8:33:50 PM UTC link Permalink

because latvian names and surnames;
if of a male, "s" to end.
if of a female, "a" or "ja" to end.

boracasli boracasli June 6, 2011 June 6, 2011 at 8:39:15 PM UTC link Permalink

and Carl is not Carlas in Latvian. Karls is better. If Lithuanian, use Karlas but you can use Carlas.

Karls (if latvian)
Karlas (if lithuanian)

{{vm.hiddenReplies[6505] ? 'expand_more' : 'expand_less'}} hide replies show replies
slomox slomox June 6, 2011 June 6, 2011 at 9:48:06 PM UTC link Permalink

The rules are more complex than "s" for male and "a/ja" for female. "Otto" becomes "Oto", "Constance" becomes "Konstance", "Henri" becomes "Anrī" etc.

"Carlas" was not meant to be Latvian, it was an example for "another inflecting language".

{{vm.hiddenReplies[6506] ? 'expand_more' : 'expand_less'}} hide replies show replies
sacredceltic sacredceltic June 6, 2011 June 6, 2011 at 10:15:21 PM UTC link Permalink

As I stated earlier, the system I propose doesn't need to work for all languages. Most duplicates are in main languages and symptomatically English and probably Japanese.
The mechanism I suggest would enable to code name markers for language A (where it is possible with no or few simple rules) and not for language B (because it is too complicated). And that would not interfere at all with the translation process, so that if we implement this system for English, French and German, that would change nothing for Latvian.

sacredceltic sacredceltic June 6, 2011 June 6, 2011 at 10:26:37 PM UTC link Permalink

and how do you write Ichiro or Tchang in Latvian? Translating firstnames doesn't make sense...

{{vm.hiddenReplies[6510] ? 'expand_more' : 'expand_less'}} hide replies show replies
boracasli boracasli June 7, 2011 June 7, 2011 at 11:53:26 AM UTC link Permalink

Ičīro for Ichiro
Tčāng for Tchang

{{vm.hiddenReplies[6516] ? 'expand_more' : 'expand_less'}} hide replies show replies
sacredceltic sacredceltic June 7, 2011 June 7, 2011 at 12:30:11 PM UTC link Permalink

Ok, but what is the point of translating "Ichiro is taller than Tchang" in "Ičīro latvian(is taller than) Tčāng"?

"Oto latvian(is taller than) Karol" is much better, because at least, Latvian readers know that "Oto" is a firstname, since if you're not in the manga culture or interested in japanese culture, you may not know that "Ičīro" is a firstname. You might wonder if the sentence doesn't tell about a mountain named "Ičīro"...

There are dozens of thousands of firstnames in the world. You can't expect them to have a "translation" (whatever this means) in every language.
What is the translation of Sioux's "Jumping dear" into Latvian?

OK, close languages have related firstnames, such as "William" in English is "Guillaume" in French. But what is the purpose of translating one into the other in an example sentence?
What is it that you translate when doing this? Meaning??? What does a firstname "mean"?

Actually, I have a firstname which is of Saxon origin and which has become very popular in France in the last 30 years. But the same firstname has almost disappeared from Saxon England and sounds so old-fashioned to English speakers that it makes them smile.
So although my firstname exists both in French and English, it doesn't conjure up the same images at all...So what's the point of keeping it when translating from French to English? Isn't it better to translate a rare/funny/ridiculous/popular firstname in language A into another rare/funny/ridiculous/popular firstname in language B so as to convey the same idea rather than the same letters, especially when these letters have no other meaning?

{{vm.hiddenReplies[6517] ? 'expand_more' : 'expand_less'}} hide replies show replies
slomox slomox June 7, 2011 June 7, 2011 at 1:39:06 PM UTC link Permalink

Tatoeba is about translation, isn't it? What you propose is more of an adaptation. Like "George likes fish and chips" <-> "Franz mag gern Sauerkraut" <-> "Jean aime baguette" <-> "Luigi piace pizza"

{{vm.hiddenReplies[6519] ? 'expand_more' : 'expand_less'}} hide replies show replies
sacredceltic sacredceltic June 7, 2011 June 7, 2011 at 1:55:05 PM UTC link Permalink

NO. You confuse people's NAMES and NOUNS. Because chips/baguette/Sauerkraut have a MEANING. But Luigi/Jean/George have NO meaning at all. They're just common firstnames in the countries where the language is used. Nobody cares about them for the purpose of making example sentences.

"Jean aime la baguette" and "Luigi piace pizza" MEAN something different.

but "Luigi piace pizza" and "Guido piace pizza" have exactly the SAME meaning.

Do you know this Luigi in this sentence? And even if you did think of him at the time you wrote the sentence, what does it mean to me?! So what the heck if he has another name?

boracasli boracasli June 7, 2011 June 7, 2011 at 3:35:19 PM UTC link Permalink

if Ichiro is a firstname, the latvian is Ičīrs. because Ichiro is a male given name.