Sure, but some of the failing cases are already there. I gave a quick look at your testing code and couldn't understand it, so I need to do a deep dive. Does your test code test the opposite, and also the same case? I mean, for example:

bass -> basses

Does your testing code tests

singular(bass)
plural(bass)
singular(basses)
plural(basses)

?

Sure, but some of the failing cases are already there. I gave a quick look at your testing code and couldn't understand it, so I need to do a deep dive. Does your test code test the opposite, and also the same case? I mean, for example:

bass -> basses

Does your testing code tests

singular(bass)
plural(bass)
singular(basses)
plural(basses)

?

I'm not sure, but I'm guessing not. Historically, the project assumes valid inputs, so a call of singular(bass) or plural(basses) is considered invalid. That is, the project doesn't perform general-purpose semantic language comprehension, but instead provides simple inflection given known inputs.

I'm happy to accept more sophistication, but the status quo is somewhat limited.

Actually some of the failing cases are like you said, singular(bass) or plural(basses). I will give it a thought about how to do some kind of verification before doing singularization/pluralization.

Added test cases like jeans -> jeans

Oops. I merged this not double-checking the tests and I see now they're failing, so I've force-pushed without those changes. Please re-submit.