As it's so tied to Chinese culture, it makes sense to have the tag be in Chinese, especially as there's a dedicated way to refer to them in comparison to English, as when I googled it, the translations "lucky sayings," or "words of good fortune" were also offered. The request to have tags in English is so that they're not repeated across multiple languages, but it seems unlikely that this tag would end up being repeated, and having it in English may actually make it more likely to be repeated (unless other languages have similar concepts?).

Could you possibly put it in both traditional and simplified with a / between? That way if someone searches in either it will come up (unless it already works like this, I haven't tried searching in Chinese before).

Merry Christmas, CK. The holiday special quest is sweet!

お元気ですか？
いつもありがとうございます！

What kind of features would you have in mind for such a platform?

I've daydreamed for a while about a collaborative and open-source framework for writing open-source language/grammar textbooks containing structured data for things like grammar topics, vocab lists, and exercises. There are several language textbooks that I really admire, but they frustrate me for 2 reasons. (1) Their data (e.g. list of grammar topics, vocabulary per chapter, etc) is not delivered in a structured format and can only be extracted by OCR/scraping (which often has errors) or manual data entry, so it's hard to use it in tandem with other resources such as Tatoeba. (2) Textbook licenses make it hard to use textbook content for legal reasons as well.

Another part of this daydream is the idea of a language textbook with non-linear progression between chapters. Each chapter or exercise would have a list of "dependencies", i.e. other chapters whose content it depends on, but a reader would be free to traverse the chapters in any order so long as they complete each chapter's dependencies before reading it. All chapters together would form a DAG (directed acyclic graph) structure, not just a tree.

There are a few permissively licensed language textbooks out there, but they are usually one-off projects by some institution or professor that don't allow ongoing contributions, and they still lock up their data in an inconsistent/un-parseable format. I also find that a lot of more recent language textbooks de-emphasize grammar and instead focus on trying to simulate immersion through activities and media, which I personally find unhelpful.

So for now this is just a bit of a pipe dream. However, if I had some help and knew that a considerable number of people would be interested in contributing to open source language web-textbooks, I might try programming a framework for such textbooks myself.

I forgot about it until reading this but there's a site called LangCorrect, which is not quite like what you described, but still may be interesting.

If I remember correctly, it's a writing prompt type thing in a bunch of languages. You can get corrections from native/fluent speakers as well as help people learning your own native language.

You can just strip the verb ending. 読 would match 読む, 読みます, 読んだ, 読んで, 読まない, 読める, 読もう, etc.

Hello,

Searching in Japanese on Tatoeba is currently limited in that it only works a character level, not at a "word" level. Put differently, every character is considered as a word of its own. We would like to improve that situation, but we don’t have the resources to do so at the moment. Note that Tatoeba is an open project and contributions are welcome.

If you are looking for a way to automate the process of retrieving example sentences containing any form of a given verb, I recommend that you download the Japanese Tatoeba corpus as a file and to process it yourself with a natural language processing toolkit. Weekly exports are available at https://tatoeba.org/downloads.

If you are looking for a way to manually search for sentences containing any form of a few verbs, let me explain some ways this can be achieved currently. I'll take the verb 振る（ふる） as example.

1. First, you need to use quotes around Japanese keywords:

振ります
→ will match sentences containing the four characters 振, り, ま and す anywhere in a sentence, in any order (but the provided order is prioritized).

"振ります"
→ will match those four characters contiguously.

2. You can use @text and @transcription to match furigana, too:

@transcription "ふります" @text "振ります"
→ will match sentences having "ふります" in furigana and "振ります" in sentence text.

3. There is a caveat: the furigana will not be specifically matched against the corresponding kanjis, for example:

@transcription "ふ" @text "振"
→ could potentially match the sentence 降りるふりをしました。 because of the presence of unrelated ふ elsewhere in the sentence.
→ however the probability of such mismatch gets very rare if you use two or more characters in @transcription

4. Regarding your question about how to match all forms of a verb. Let’s first consider verbs which reading is unambiguous, such as 思う. If you search for the root "思", you will get unwanted matches such as 不思議. The easiest way to exclude these is to provide a suffix to 思. Because 思う is a godan verb, there are only 6 possible suffixes:

思う（おもう）
→ 思わ, 思い, 思う, 思え, 思お, 思っ

We can give all the forms separated by the OR operator "|":

"思わ"|"思い"|"思う"|"思え"|"思お"|"思っ"
→ this search shows all sentences containing the verb 思う conjugated

5. If the verb is ichidan, and the root has more than one character, you can just search for the root:

"食べ"
→ this search shows all sentences containing the verb 食べる conjugated

6. If the verb is ichidan and the root has only one character, because of the caveat explained at 3, you need to provide a second character. It gets tricky because there are many possible second character:

見る（みる）
"見る"|"見ま"|"見さ"|"見ら"|"見た"|"見な"|"見ろ"|"見よ"|"見て"|"見え"
→ brings a few false positive such as この指輪ね、祖母の形見なの。

7. If the verb reading is ambiguous, you can combine example 2 and 4:

振る（ふる）
→ 降ら, 降り, 降る, 降れ, 降ろ, 降っ

So the final search is:
(@text "降ら" @transcription "ふら") | (@text "降り" @transcription "ふり") | (@text "降る" @transcription "ふる") | (@text "降れ" @transcription "ふれ") | (@text "降ろ" @transcription "ふろ") | (@text "降っ" @transcription "ふっ")

(We need to use parenthesis because the implicit AND operator has a
higher priority than OR "|")

Something else that might save you time is to use Jim Breen's wwwjdic.

Look up a verb, click the "links" link, find the "verb conjugation" link, then copy all the forms, and find what you need to search for as gillux explains.

This should save you a lot of typing, Here's an example of one verb conjugation page.
https://www.edrdg.org/cgi-bin/w...A4%A8%A4%EB_v1

Here also has an "example search".

https://www.edrdg.org/cgi-bin/w...A4%A8%A4%EB_1_

These responses by @gillux and @CK contain useful information that should be made available on the Tatoeba wiki as well.

Here are a few more links that you may find useful.

The Python NLP library called spaCy has pipelines available for Japanese:

https://spacy.io/models/ja

Using this library, you can download spaCy models and use them to analyze a sentence. Part of this pipeline includes a component that attempts to split the sentence into individual words, and reduce each word to its "base form". I personally use this library to generate fill-in-the-blank vocabulary exercises for German and Russian (a la Clozemaster) by searching for Tatoeba sentences containing words whose base forms are a target word that I want to study.

You may also find this useful:

https://github.com/cl-tohoku/J-UniMorph

This dataset purports to list all (most?) inflected forms for many Japanese words. Not only that, but the data file also lists how each form is inflected.

Thanks @gillux and @CK, I had not known about the character level only search. I was initially just sniping off the last character of the verb but as you said, I'd be running into issues with words containing just the root. I think obtaining all the conjugations separately and searching for that is probably the way to go for now.