Toy Problem of the Week

I was raking around on the net and found this problem and coded up a solution. Once we get at least 10 solutions I'll post mine and a link to where I found the problem.
Any language, any libraries (but declare them) allowed. Aim for clarity and conciseness. Assume the input is well formed. Mine is 5 lines of scheme code (6 if we include requiring a library). No calling an "answer" function that "you wrote already and put in an external lib" ;)

Consider the problem of turning a list consisting
of a mix between symbols and non-symbols into a
list of lists of the symbol and its following
non-symbols. That is:

Input:    ({<symbol> <non-symbol>*} ... )
Output:   ((<symbol> (<non-symbol>*)) ...)
Example:     (a 1 2 3 b 4 5 c d 8 9 e)
           -> ((a (1 2 3)) (b (4 5)) (c ()) (d (8 9)) (e ()))

>>45
OP already mentioned that he meant Lisp-like symbols:
1, "1" or '|1| are very different kinds of objects, one is the integer 1, the other is a string containg the character #\1, and the other is a symbol named by the string "1". Things may be slightly different in Scheme, but 1 and '1' are different things. OP's problems essentially boils down to grouping down objects given a predicate which separates the objects into 2 disjoint groups, the actual meaning of symbols or non-symbols is not as relevant, sadly, at least some half of the solutions are string-based, so they miss the point of the problem, and work with some limited amount of input.