On github projects use issue labels 'good first issue' or 'help wanted' or similar, but I've also read a counter viewpoint https://dzhavat.github.io/2020/07/08/the-good-first-issue-my...
(1) which are in languages you known or are willing to learn enough to contribute in?
(2) which require contributions; sometimes open source projects are marked 'feature complete' and will only accept bug-fixes etc.
(3) of the remaining which are about a problem space that you can understand, want to try to understand, and are willing to contribute to.
If there is no candidate at the end of that list you can always search GitHub (or GitLab, SourceHut etc) by language and popularity and try look for feature, bug, or documentation contribution that way. I found for myself a good starting off point to be the things I am already using that were accepting new features that I wanted to write features for.
---
Exceptions:
Of course documentation can be (programming) language agnostic if you can generally understand the flow of the logic and it isn't too technical so if documentation is your thing have at it!
Bug-fixes, depending on the project, may require debugging which can be trickier than writing features and tests (in a happy-path world).
---
Example:
You use a few libraries, some in Go, some in JavaScript, some in C. You only know Go and don't want to learn JavaScript and C at the moment; so naturally you filter out all but the Go libraries you are using. Of the remaining you are excited about 1 or 2 of them and so you peruse the codebases, the open issues, and any mailing lists for feature requests, pending bug fixes etc, and eventually after understanding how the library is structured and generally how the codebase works you can work on a feature or bugfix.