That would not be obvious in the Finnish legal system. The original source would have to be unique enough to be considered a "teos", i.e. demonstrate unique artistic vision and be something another person would not have come up with.

This would be unlikely to be true of a short list a common greetings and their translations. A more unique translation would certainly qualify.

But the French copyright law is probably different.

It is true that lists can be copyrighted, but it's not as straightforward when you take into account fair use. You'd have to consider how much of the list has been copied, and of course whether the list itself had some originality or not.

For instance let's say I have copied 20 basic sentences from a very large dataset of thousands and thousands of sentences. It would be impossible for the copyright owner of the dataset to claim a copyright infringement.

If anything, we should be able to keep very basic sentences within Tatoeba, even if they have been copied, as long as they are not presented within Tatoeba in a list that looks nearly identical to the source.

For instance if I'm a new user and I've copied 20 basic sentences from a blog post, and those are my only sentences in Tatoeba, then there could be copyright infringement due to the sentences being listed on my sentences page. But if I unadopt all these sentences, and there become blended into the corpus, then there is almost no more risk of copyright infringement because I have effectively removed the list.

On the other hand let's assume someone wants to create a list of basic sentences in Tatoeba, and they came upon an interesting blog where the author compiled a list of good sentences to learn for beginners. They decide to re-create this list in Tatoeba. Coincidentally, all the sentences on that blog post already exist in Tatoeba. Well, the list that the user created in Tatoeba can be considered copyright infringement despite the fact that the sentences existed already. But we would just need to delete the list to solve the copyright issue, we wouldn't have to delete the sentences.

My point is that when dealing with copyright on lists, we have to look at the places where Tatoeba lists sentences ("My sentences" page, lists, tags, favorites). This is where the list copyright infringement can make sense and it is the list that we would need to delete, not the items in the list. For the items, we have to look at each of them individually and evaluate if on their own they have any originality or not. And if not, it is fine to keep them.

I'll remove the red mark on the sentences you mentioned.

I've also created an issue on this topic: https://github.com/Tatoeba/tatoeba2/issues/1847

Feel free to suggest other solutions if you have any better ideas.