Other than a more broad "how is the language as a whole faring?" test, which might be done through surveys or other product-style research, I think this is just plainly a hard problem to approach, just by the nature that it's largely about user experience.