I have several responses to this criticism:
The forty-year history of high-level computer languages tells us that the "hot" programming language changes every 5-10 years anyway. While an employer may want someone who can "hit the ground running" for a short-term project, a more important qualification for long-term employees is the ability to adapt to new techniques and languages.
Whatever language I teach to a college freshman (not to mention a high school student) will be at least partially obsolete by the time that student is graduated. Accordingly, the student's job prospects are better served by learning to learn a new language quickly and by a solid understanding of fundamental concepts of programming that hold across languages, than by spending years mastering the details of one language that was "hot" at the beginning of those years.
A student who's taken one year of programming classes shouldn't be writing substantial programs professionally, any more than a second-year medical student should be diagnosing and treating patients. So if learning Scheme prevents first-year programming students from getting a programming job, that's perhaps a good thing. (Do you want to ride on the airplane whose navigational system was programmed by a college student with a year's experience?)
These arguments are even stronger when applied to high school programming courses, because high school students are (presumably) another year or two farther from the job market. If a high school or beginning college student has the choice between learning C++ syntax and studying algebra, the latter may more effectively advance the student's career in computer science: while most computer programming is no longer about numbers, being comfortable with algebraic concepts (e.g. "variable", "operation", "function") correlates strongly with success in computer science.