For me BASIC is just another 3rd generation programming language. The question will not be if the code is faster but if it is fast enough and 99% of the time C is. If not assembly becomes the first option - or redesign the algorithm, use a faster board, not another language.
I think it would become more interesting if you came up with a "sensor actuator motor language" that could easily be used by younger children, e.g. for programming robots. Call it ROBO-BASIC. That would give it a focusgroup and a possible market in the form of schools.
Nevertheless building a compiler is a nice piece of work (even if it won't beat the GCC)
Finally a language like Python would also be interesting in some markets. Or a rule-language which consists (mainly) of condition=>action clauses