I think many people fully understand this. When people rave about, say, haskell being nearly as fast as C, what they mean is that you can write much higher level, and thus shorter and more concise, code in a way that can be automatically optimized by the compiler quickly.
In other words, he's correct but also assuming incorrectly that because other don't spell it out, they don't know.
Oh it's more than that. Ever tried to write a trie in C? Yea I didn't think so. C/C++ are great languages for missing the forest for the trees. I'd call Java is faster than C/C++ for this very reason.
This is one of my irrational prejudices regarding programming. It annoys me a little when people talk about their favorite language being "as fast as C" or "X times slower/faster than C," because such things simply aren't properties of languages, but of particular implementations on particular architectures. It bothers me almost as much as when people try to claim that "C/C++" is a programming language. :P
I admit, this really just makes me sound like some old kook with a huge beard who lives in the old programmers' home, or something, but I manage to restrain myself from commenting about it most of the time. So, rather than listen to me, I propose checking out the c2 wiki's page on the subject: http://c2.com/cgi/wiki?AsFastAsCee .
His argumentation misses an important variable: The programmers time involved. Fix this, and fix the machine, and programming languages do have speed.
And certainly, the statement 'This problem solution in language X developed in Time T running on machine M is faster than the problem solution developed in language Y developed in time T, running on machine M' (often condensed into 'X is faster than Y' is certainly possible and valid, given that the programmers developing the program are about even in skill. (I don't want to assume the same programmer here. Language benchmarks have been shot because someone wrote a good C-Program and a horrible Python program and complained how slow the python program is).
It's a nitpicky argument; competent people understand the meaning.
Languages have relative speeds on the same hardware if similar algorithms are developed by experienced programmers. E.g. ML is hundreds of times faster than script languages on 4MHz CPUs.
Of course when you get into 'nearly as fast' territory then all bets are off. And skilled optimizers can usually tip the scales.
In other words, he's correct but also assuming incorrectly that because other don't spell it out, they don't know.