Yes, provided that it's general purpose and doesn't win.
Fixes for unintended behavior in an existing language past a sandbox post date are obviously OK, to address the point about bugs in the OP.
For languages entirely created after a sandbox post date, it depends on two things:
- The language must be general purpose. That means, you should be able to use it to solve a wide range of other code challenges competitively (what I mean is that it should not be optimized for a single challenge). In most cases, it should also be Turing-complete. Ideally, the language should be specifically for a single tag (eg, a language with lots of string builtins for string).
- Your submission in that language must not win. Otherwise, it's just unfair for the other challenge participants. If the implementation in your language happens to get the highest score for the challenge, don't post it and just wait for the next challenge within the language's scope. (Note that you should not intentionally leave your implementation ungolfed to submit it without winning. That defeats the purpose of this point.)
The original reason for the standard loophole is to prevent the case of "I invented a language XYZ that does this task in 0 bytes," and to be fair to others. As long as your language doesn't violate those things, I think it's fine.