Adataink különféle Creative Commons licencek alatt jelennek meg.További információ
Ha szereted ezt a tartalmat, kérünk, fontold meg az adományozást.
Üzenőfal (7 319 bejegyzés)
Tippek
Mielőtt bármi kérdésed lenne, kérlek, olvasd el a GYIK oldalt.
Célunk az barátságos légkör fenntartása a civilizált beszélgetések érdekében. Kérünk, olvasd el a magatartási szabályzatot.
cafoc64474
12 órával ezelőtt
CK
tegnap
CK
tegnap
araneo
2 nappal ezelőtt
Alex_M
3 nappal ezelőtt
Alex_M
4 nappal ezelőtt
kumakyoo
5 nappal ezelőtt
Tom9358
5 nappal ezelőtt
Alex_M
5 nappal ezelőtt
kumakyoo
5 nappal ezelőtt
🍎 I generated word frequency lists for each language.
https://aitstudy.com/temp/wordcounts/
These are all TSV files
word + tab + count
bz2 compressed files.
I see you created site named AI study.
You could create dictionaries between languages based on how many times words appear in translation pairs.
Also you could use English as pivot language for sentences (although not 100% accurate) to build even bigger sentence pair list between languages.
Also you could train translation models (example MarianMT).
I'm sorry if I bothered you.
Currently, I don't have time to do these type of things myself.
Question about index list.
I want to see an alphabetical list of say German words. Not sentences, but just words.
When I notice a word which I do not know and which I wish to learn, I click on the word and see a German sentence with this word.
I have a book with pairs of sentences, and this book includes two indexes for both languages with corresponding page numbers. It is very convenient for finding and learning unknown words.
Is there such functionality on the website or via API?
No, there is no such functionality on the website or via the API.
Thank you for the information.
Meanwhile I found the alphabetical index lists online. For example, for German language it is: https://de.wiktionary.org/wiki/...l:Präfixindex/
But this index is overwhelming as it includes all the words including rare terms.
It would be interesting to have an index list of words which native speakers use in the sentences.
I believe your idea would be very useful, but it might be a bit out of the scope of Tatoeba. What I mean is that Tatoeba focuses mainly on building the corpora and making it available to the world, while others can build upon this resource to make something more specific, such as an alphabetical list of German words.
That being said, the search engine that powers tatoeba.org precisely has word indexes for every language. There is a simple tool to dump these indexes, so we can consider exporting them as text files. However, because the purpose of these indexes is only to allow very fast retrieval of sentences, this data might be a bit too "raw" to be directly usable by language learners. But please let me know if you or anyone else is interested by such dumps.
Thank you for your input. For me personally a text file would not be useful. What I had in mind is the list of distinct words on an HTML page, where a word could be clicked to see the sentences in which it is used.
And it would be nice to have a possibility to sort it in alphabetical order and in frequency order.
Certainly, it would be a complex task. For example, there is a German word: Zugeständnis, but plural form of this word is: Zugeständnisse. I am not sure if both variants should be in such a list or only one. And what if a word exists in the database with a spelling error? That's to say there would be for sure an issue of rawness which you mentioned.
I asked about this index list only because I try to learn new words in a sentence. And it would be easier for me to spot in such a list words which I do not know yet. Especially if it is a word with high frequency.
Ich hab' mir kürzlich ein kleines PHP-Script geschrieben, dass aus dem Tatoeba-Datendump eine Liste der griechischen Wörter extrahiert und zu jedem Wort notiert, wie oft dieses Wort im Korpus vorkommt. (Mir ging es dabei darum, die Sätze im Korpus nach "Einfachheit" zu sortieren, mit der Idee, dass häufige Wörter einfacher sind, als seltene Wörter. Das hat ganz gut geklappt, auch wenn es nicht perfekt ist.)
Für dein Anliegen könnte man das Programm benutzen, um eine entsprechende Wortliste für den deutschen Korpus zu erstellen (evtl. beschränkt auf die 10.000 häufigsten Wörter). Die Ausgabe könnte man so gestalten, dass dabei eine HTML-Seite entsteht, die Links auf die Tatoeba-Suche enthält, sodass du mit einem Klick Sätze mit diesem Wort erhältst. Mein Programm kann dabei unterschiedliche Varianten eines Wortes (Zugeständnis/Zugeständnisse) nicht berücksichtigen, macht also zwei Einträge daraus. Aber, soweit ich weiß, kann die Tatoeba-Suche das bei einigen Sprachen und Deutsch war da dabei.
Ich denke, es ist nicht viel Arbeit für mich, das Programm entsprechend anzupassen. Ich könnte dir also anbieten, so eine HTML-Seite zu erstellen und dir per E-Mail zu schicken (schreib mir einfach eine private Nachricht mit deiner E-Mail-Adresse, dann mache ich das). Evtl. könnte ich das Programm auch so überarbeiten, dass ich es auf GitHub allen zur Verfügung stellen kann, mal sehen...
Vielen Dank! Selbstverständlich können Sie mir den Link per privater Nachricht schicken. Oder, vielleicht, Sie könnten ihn hier posten, damit auch andere Teilnehmer die Liste ansehen können?
Ich hab' das Programm mal bei GitHub hochgeladen: https://github.com/kumakyoo42/tatoeba_stuff
Das Programm selber heißt "count_words.php". Man benötigt dafür die Sätze-Datei von https://tatoeba.org/de/downloads (sentences.tar.bz2). Diese muss entpackt sein. Als Beispiel habe ich die Top 10.000 der deutschen Wörter ebenfalls hochgeladen (top10000_deu.html).
Ich hoffe, das ist in etwa das, was du gesucht hast.
It is exactly what I need! Plus, there is a number near each word which shows how many times it was used in the sentences.
It works for every language (the language code to be uses is ISO 639-3, i.e. 3-letter code, not 2).
I changed the script a little for myself, - I removed converting to lowercase. It's easier for me to distinguish nouns this way.
I've already found a word which I did not know, and I could click on it in the list and see some sentences with it. Incredible programming! Thank you!
I created three HML pages based on the PHP script from https://github.com/kumakyoo42/tatoeba_stuff
These are frequency lists for German, English, and French languages, twenty thousand words in each:
https://labellechose.ch/frequency-lists/deu.html
https://labellechose.ch/frequency-lists/eng.html
https://labellechose.ch/frequency-lists/fra.html
Click on a word and the sentences with this word are displayed, click on the arrow and the definition from TFD dictionary is displayed.
These pages were created for personal use. They are located on the self-hosted mini-server which is not always online. The links are published for demonstration only as part of this discussion.
Thank you both!
Here are English and Japanese TSV files that you can download.
https://aitstudy.com/temp/
eng_wordcounts 2026-02-19.tsv
eng_wordcounts-lemmatized 2026-02-19.tsv
Vocabulary size: 76704
Lemmatized vocabulary size: 67092
jpn_wordcounts 2026-02-19.tsv
jpn_wordcounts-lemmatized 2026-02-19.tsv
Vocabulary size: 47632
Lemmatized vocabulary size: 42192
Ennek az üzenetnek a tartalma szabályellenes, ezért rejtve maradt. Csak adminok és az üzenet szerzője láthatja.
Ennek az üzenetnek a tartalma szabályellenes, ezért rejtve maradt. Csak adminok és az üzenet szerzője láthatja.
Ennek az üzenetnek a tartalma szabályellenes, ezért rejtve maradt. Csak adminok és az üzenet szerzője láthatja.
Tatoeba was updated today. What’s new?
- The icon of the Marathi language was updated, thanks to @sabretou and me. Sabretou added the original icon a very long time ago, but now believes it is outdated, and that there is a better, more representative option for Marathi.
More background info: https://github.com/Tatoeba/tatoeba2/issues/3238
New vs. old image: https://github.com/Tatoeba/tato...d1ca024cf6dd95
- The new API at https://api.tatoeba.org/, dubbed "v1", is now considered stable, so people are welcome to build tools upon it. (An API allows external programs or apps to directly connect to Tatoeba and browse the corpus.) This "v1" release is the culmination of months of work to provide a well-documented, modern, stable and easy-to-use API. If you are using the previous "v0" API, you are encouraged to migrate: https://en.wiki.tatoeba.org/art...i-migration-v1 The "v0" API will keep working for some time to ensure backward compatibility, but people should not build new tools based on it.
- @kumakyoo made his first contribution to Tatoeba’s code: https://github.com/Tatoeba/tato...l/3244/changes Thank you very much! If you would like to get involved with the development of Tatoeba, feel free to contact us. We need help from coders, but also for interface translation, documentation as well as UI/UX designers.
I think the API v1 being considered stable is a big step! Congrats, gefeliciteerd!! 👏
Ennek az üzenetnek a tartalma szabályellenes, ezért rejtve maradt. Csak adminok és az üzenet szerzője láthatja.
🍎 These have been updated.
https://www.manythings.org/bilingual/
Bilingual Sentence Pairs
https://www.manythings.org/anki/
Tab-delimited Bilingual Sentence Pairs
These are selected sentence pairs from the Tatoeba Project.
Thanks to all of you who make this possible.
✹✹ Stats & Graphs ✹✹
Tatoeba Stats, Graphs & Charts have been updated:
https://tatoeba.j-langtools.com/allstats/
Cool. Couldn't these be linked somewhere on the website?
Well, I don't mind. :-)
I added it to the "Projects using Tatoeba" wiki page:
https://en.wiki.tatoeba.org/art...using-tatoeba#
Yeah I always find it difficult to find the link.
Tatoeba me dit que la phrase
"Il ne faut pas appeler richesses les choses que l'on peut perdre.",
citation de Léonard de Vinci,
n'existe pas encore, donc je cherche à l'ajouter, ce que je ne peux pas faire car dans un second temps, Tatoeba me dit que cette phrase ne peut pas être ajoutée car le robot Voltaire l'a déjà fait...
Ich habe den Satz freigeschaltet: #7709449. Er kann jetzt wieder gefunden werden.
Merci pour la réactivité. Dankon pro la rapida respondo. Vielen Dank für die schnelle Antwort.