“But with more than 400,000 existing sentences in English, you have a rich supply of sentences to translate into Finnish. Any translation you contribute is guaranteed not to be an orphan. If you want to contribute a sentence with a particular term in Finnish, why not use the search function to find sentences with corresponding words in English? And if a few of them are missing, find English speakers and ask them to contribute English sentences that contain them.”

Hear, hear!

Direct those you'd ask to translate your spreadsheet to:
http://tatoeba.org/eng/sentence...n/fin/none/eng

>> If you have someone who's willing to check your work they could just as well simply translate the sentence …
>
> Not if they don't speak the language of the original sentence ...

No, but then they can't tell if it's a good translation, either. Tatoeba isn't a collection of sentences, it's a collection of translations.

>> But creating some sort of a culture where people can add poor data to the corpus is inviting shoddy work and harms the project.

> Completely agree, that's why I insist on first finding a native speaker willing to correct you and only then add the sentences.

Absolutely. That should be the standard.

No one is arguing that it's in any way harmful. It is, however, unnecessary since a proficient translator doesn't (by definition) need hints.

This all goes to the question of what it takes to translate a sentence into a different language. I don't even translate all sentences between by best two languages. There are some where I'm simply not familiar enough with the idiom or usage.

I'm of the opinion that when a language corpus comprises a few thousand sentences (an easy feat for even a single contributor) we should start to raise the bar on the quality of sentences.

For Tatoeba to be a great learning resource we need to demand a high standard of sentences. There are great projects out there for people wanting practice in writing sentences in languages they're learning (http://lang-8.com/ is a good example) but Tatoeba contributors should hold themselves to a higher standard.

… effectively duplicating your work. If you have someone who's willing to check your work they could just as well simply translate the sentence …

But then again, if you want to use this as language practice and have someone willing to teach you by pointing out your mistakes, then great. All the power to you.

But creating some sort of a culture where people can add poor data to the corpus is inviting shoddy work and harms the project.

C) Translate English sentences the meaning and tone you understand into Finnish or add Finnish sentences without any English translations. This will guide proficient Finnish to English translators a list of sentences to translate: http://tatoeba.org/eng/sentence...n/fin/none/eng

Either one of your other two options will leave the corpus with poor and incorrect translations. You needn't “give hints” to the idiom or meaning as the translator should understand these themselves.

In response to a comment by Demetrius but split to a separate thread due to the change in subject.

“I don't mind my status being taken away. I don't have much time to dedicate to Tatoeba anyway. :)”[1]

There is no harm in having extra corpus maintainers. If anything it might get people back. For the time you do spend on Tatoeba, it makes resolving issues easier and reduces the load on the more active maintainers who then would have to respond to your requests.

I think we should have a very low bar to grant corpus maintainer rights. Basically, you get them if you understand the project well enough that you won't make mistakes and won't run amok.

[1] http://tatoeba.org/eng/wall/sho...#message_10940

In my limited experience and from looking at his contributions, he seems to have a clear understanding of the structure of corpus and there is every reason to believe he will benefit the project.

Can you explain a bit better what you call a “translation pattern” and how you evaluate these?

PS. If someone is looking for a coding project related to Tatoeba, a useful one would be a linking tool that lists indirectly linked sentences in two specific languages. Possibly with the option of restricting it to one's own sentences or some other useful filter.

I've just been forwarding spam messages to her account (TRANG).

> Soweit ich weiß, ist es technisch nicht möglich, seine eigenen Sätze mit "OK" zu taggen.

Interessant! Ich dachte das das „OK“ Tag genau so gleich wie andere Tags sei. Habe gerade versucht ein meiner eigenen Sätze mit „OK“ zu taggen aber es ging nicht.

> die angegebene Muttersprache theoretisch falsch sein könnte. Es stellt sich auch die Frage, wer eigentlich ein Muttersprachler ist.

+1

halló
ég heiti Tina og rakst á prófílinn þinn í dag á tatoeba.org og þykir hann frábær og langar líka að vita meira um þig mér þætti vænt um ef þú sendir mér tölvupóst aftur í tölvupóstfangið mitt (...) svo ég geti gefið þér myndir af mér svo þú vitir hvern ég er, bíð bráðum eftir yndislegu svari þínu.
fröken Tina

Merry Christmas Shishir! Translated with grammatical errors and all.

(email address redacted as we wouldn't want the poor thing to be inundated with emails from people she hadn't so carefully selected to receive it)

The sentences in my post now both show up.

Even if it would cause a bit of an overhead, I think this would be an extremely useful feature.

"by the way: Who put the tag? When? How about deleting?"

You can see who put the tag up by placing your pointer over the tag. Most browsers will then display the contents of the element's "title" attribute, which contains the user's number, and time and date it was added. You can look up the user by going to the "Show latest activity" page which is still indexed by user number.

Mine, for example, is:
http://tatoeba.org/eng/users/show/1005

In the case of that sentence with the "delete" tag, the user number is 1396. You can replace the number in the link above and you'll find that the person who added the tag is CK. He added the tag in 2010, on the 28th of September at 14:08:25, France local time.

As blay_paul noted in the comments, the sentence is probably from lyrics and thus possibly a copyright violation. It should only be deleted once we're sure it doesn't belong here. Sometimes I've seen "delete" tags put up where one user is suggesting deletion, but wanting to get someone else's opinion (the "delete?" tag doesn't work because Tatoeba doesn't like question marks in tags).

As for copyright violations, I've begun culling these.

"When putting a tag - why not adopting the sentence, if it is orphan?"

Generally, adding a tag demands that one know a limited amount about the sentence -- usually less than whether it is natural or not. In the case of situational tags, one might know in which situation it would be used (say on the bus), if natural, but not whether it is natural.

With tags such as the various Japanese sentence forms (such as "ですが。") one needn't know whether it is a natural sentence to spot the pattern. The maintenance tags (the ones preceded by an @-mark) I recommend adding a comment to describe what needs to be done (except in the rare cases when that is obvious).

"And to have a better place to discuss subjects and to find them back: I think we need a forum, where we can easily have a structure of our discussions and subjects."

I'm not so sure about this forum idea, to be honest. We don't have that much traffic here on the wall splitting it up is, I think, premature. I think having everything in a single place, furthermore reduces chatter and keeps the focus on the sentences. :-) The chatter has a fine place on IRC (come join us, if you like!)

I do share your view that it would be nice to be able to better find past wall posts. A search feature should solve that.

I suppose we'll end up splitting the Wall up eventually, but I see it as a necessary evil with plenty of down-sides. From other projects I've taken care of, newcomers are generally unsure where to post and added complexity reduces the effort people are willing to put into learning about them.

Checking a similarly old sentence of mine, I noticed something odd. Searching for "settu það á eldinn"[1] turns up a single sentence:
"Sæktu það og settu það á eldinn."[2]
which was added 13 days ago and is indirectly linked to:
"Náðu í það og settu það á eldinn."[3]
which was added 14 days ago, and should show up in the search results as well.

Specifying the source language or restricting the search result to "eldinn" doesn't turn up setnence [3].

[1] http://tatoeba.org/eng/sentence...rom=und&to=und
[2] http://tatoeba.org/eng/sentences/show/885972
[3] http://tatoeba.org/eng/sentences/show/885081

@alexmarcelo: In case you weren't sure, you don't have the language filter[4] set.

[4] http://blog.tatoeba.org/2011/04...-7th-2011.html

+1

Well, in my experience these guest books tend to be stale, spam-ridden and uninteresting.

Of course, should others like it, there is nothing wrong with adding it. Especially since it could be given an unobtrusive link below the latest Wall messages links. You do make a valid point about the friendliness[1] and as Tatoeba takes on more users, it may well serve the Wall to be split up along topical lines.

Should it prove popular with newcomers, it might actually make a useful tool to alert them of the project documentation and their profile. Perhaps signing the guest book would include their user profile to encourage them to say something about themselves.

You haven't quite sold me, but I may be warming up to the idea. :-)

[1] Personally I'd rather improve the culture on the wall than retreat with the friendly interactions that used to dominate this venue to another venue.