It is broken again.

Thanks!

Small question regarding tags. In spanish adjetives have gender.
For example the sentence "You're smart" can be "Sos listo" (male) or "Sos lista" (female).
What is the right tag for this?

¡Bienvenido!

#5050524

I used this site passively to mine flashcards for many years, but just
recently I started contributing to the project. So my knowledge about
the dynamics of the community is biased and erroneous.

1) I will love to have this. In fact, one of the reasons that I joined Tatoeba
is to introduce sentences that I want to have a translation. I don't know what
it the best way to collect "wished words or structures". But a nice feature
would be that from time to time the random sentence instead of showing a proper
sentence it displays a message like "Please create a sentence with the word X".

2) If I am not mistaken you will notice a disparity on the low frequency words
that are extremely dependent on the corpus that you use to generate the
frequency list (for example the work of Machado de Assis vs. the work of
Fernando Pessoa). Also you should watch out the inflexions. I think that a
complementary idea is to get the required words from the A1-C2 examinations.
The Goethe institute, for example, publishes them for their German tests.

3) I am indifferent, and I think that bill has a point here. From my short
experience I made a lot of mistakes, but they are corrected quickly thanks to
some people that are behind my sentences. If now you have lets say 10x more
contributions from the new reward system then the people that are making sure
the sentences are correct will be saturated, and the quality of the corpus will
be lowered.

4) The idea that you are proposing is not roughly the same as adding the tag
"@need native check" for all the sentences that are created by non-natives? I
think that non-native translations are OK as long as they are very simple, and
I understand the ones that don't want any non-native sentences at all. Suppose
you have a bunch of Portuguese sentences that you translate to English. If the
one that is proofreading the sentences does not know Portuguese, then she can
only judge your grammar. For example you could translate "Tom eats an apple" as
"Tom is using his digestive apparatus to process a fruit". She will accept it
because she maybe think your original sentence was a bit abstract. Now, if she
knows Portuguese, isn't it easier to just ask her to do the translation instead
of proofreading your translations?

Also, if you want to add a lang-8 like feature you will harm the people like me
that uses Tatobea (and not lang-8) to mine flashcards. Regarding untranslated
sentences, I don't see it as a big problem. For example, I am happy that there
are 100K more sentences in German than in Spanish. It means I have at least
100K sentences to practice!

5) I don't know the structure of Tatoeba database but you can have a database
of sentences and a database of audio files. In each audio file you associate
the text of what the speaker is saying (provided by the user when he uploads
it). Then, when there is a sentence with the same text as some audio file you
can just provide the audio. Editing and modifying a sentences should be
transparent to the audio database.

6) I would like to have shortcuts!, like Ctrl+T open the box to enter a
translation for the current sentence; Ctrl+N goes to the next sentence; Ctrl+P
to the previous one; Ctrl+R a random sentence, etc. I hate using the mouse and
the touchpad.

7)-8) I don't have an opinion.

Thanks!
Do you know the saying human stupidity beats artificial inteligencia?, well I am working on that :)

How?

Have the developers changed the auto detect algorithm?
Today, I am getting the wrong language around 30% of the time.

I think you can translate it to the same name in the target language.
I use Tomás instead of Tom in my Spanish sentences because the latter
makes them sound a bit artificial. Where I grew up Tom was actually a
name you would give to a pet, so I get really funny pictures if I have
to stay with Tom in my Spanish sentences.