> I am aware but without a spec we don’t know which is which.

I don't think I fully agree with this. A more accurate statement might be that "without a spec we don't always know which is which". You obviously don't need a spec to pin down at least some of a language's behavior, and even then the presence of a spec doesn't necessarily mean that you can answer arbitrary questions about what behavior is intended (c.f., various defect reports in the C/C++ spec asking clarifying questions, pointing out holes/deficiencies, or even admitting that certain behavior is not intended but being unsure about how to forbid it)