BOLDOG ÚJ ÉVET TATOEBA!
SOK SZERENCSÉT KÍVÁNOK
2020-BAN!
MRAZ

It looks like it's not the same issue. I personally don't care much whether it's collapsible or not, my problem is that the language selector appears on the screen before the rest of the menu, and because of that I click a link I don't want to click instead of selecting na language.

Kaikki kielet ovat täällä samanarvoisia, joten kai minäkin sitten käytän Google-kääntäjää:

:::
Tunnetko lokin, joka näkyy jokaisen lauseen sivun oikealla puolella? Se näyttää tietoja siitä, kuka kirjoitti, käänsi ja editoi lauseen ja milloin. Vai pyydätkö jotain erilaista?
:::

Lauseen muokkaus- ja linkitystiedot ovat hyödyllisiä vain, jos tietää jonkin lauseen, joka on lähiaikoina käännetty toiselle kielelle. Siinäkin tapauksessa pitäisi vielä katsoa, että onko lause linkitetty jo aiemmin olemassaolevaan lauseeseen, vai onko se käännetty kirjoittamalla kokonaan uusi lause. Tämän voinee päätellä lauseen kirjoittamisen ja linkityksen ajankohdista.

https://en.wikipedia.org/wiki/Translation_memory

A translation memory is basically a database containing pairs of language units (called segments). It is used to make translations using software applications like Trados, MemoQ, Across, Transit, etc.

We should make a distinction between machine translation and using a translation memory (although the dividing line between the two starts getting blurred somewhat. With a TM nothing gets translated by itself (except 100% hits), but you still get a lot of help and you can also do concordance searches for expressions, parts of a sentence, etc. Anyway, there are already applications, or at least I know of 1 application where you can build your own translation engine using a high quality TM.

Thanks, CK, I am planning to buy software that can create a translation engine, so maybe, I'll try Tatoeba, just for the heck of it.

> One reason we encourage people to contribute in their native languages is so that we can get closer to providing data of excellent quality.

I agree that people should contribute in their native languages. However, I do not think that EVERYONE should ONLY contribute in their native language. Nor do I think that if they did, we'd achieve the highest level of quality.

People should contribute in whichever languages they know very well, which in most cases will include their native language. However, that does not guarantee data of excellent quality. Native speakers make the fewest outright errors (though, as you know, they're not perfect). But a lack of errors is not the only measurement of quality. If our English corpus consisted of a million error-free sentences of the pattern "Tom gave Mary X Y for Z", it would be greatly inferior to a more diverse set of sentences, even if 1% of them contained errors. It's far easier to correct errors than to introduce diversity. The more I look at English sentences contributed by proficient non-native speakers, the more I come to the conclusion that Tatoeba would be much poorer without them. Not only do they constitute something between a fifth and a third of our English sentences, but they bring different perspectives that cannot come from a small handful of people (especially when some of those contributors use automatic scripts to generate their sentences).

It's true that some non-native speakers can write sentences that are not fully natural but are difficult to correct. But that can be avoided by efforts on the part of both the authors (who need to develop an accurate assessment of their own abilities in a language) and the correctors/commenters (who should provide specific critiques that allow the authors to improve both their abilities and their self-assessment).

> While we may never get to an error-free corpus, I think it's a worthy goal to have.

No, we will never get to an error-free corpus. But if people are serious about trying to eliminate errors, they should correct them, and address the authors as collaborators, rather than as obstacles who are interfering with a vision in which people only contribute in the languages they started learning by the age of three.

Thanks soliloquist, I didn't inow that.