I would guess that rate of benefit relative to divergence from idiomatic purity might also be significant. If a little clutter gains a 50% improvement with language A but only a 10% improvement with language B, language B having a 15% advantage under idiomatic expression may not be that significant.

But, yes, terms like "the speed of C" do not seem especially useful.

Even "equal effort" ignores the complexity of the effort-to-benefit function as well as maintainability issues and other considerations like finding good programmers who want to use the language.

Reality seems to be very complex.