If I understand correctly, he can do the math problem quickly and accurately -on paper- if in the form of an application problem, but not if in the form of a naked computation. Is this the situation? That is, is the difference in computational speed and accuracy (aka fluency) evident regardless of response medium (paper or oral)? Because if it is, then I would not be too quick to dismiss a possible second exceptionality, even if he has no fine motor issues per se. If, on paper or orally, he can quickly and accurately solve math problems when they are in context, but not when they are in isolation, that may be reflective of a learning style which is highly contextualized, so that meaning is essential to the most efficient operation of his cognitive and academic skills. The anecdote about Ramona/Wimpy Kid would tend to support this hypothesis.
This kind of profile not uncommonly features relative weakness in basic, automatic skills, like math facts, handwriting, and spelling. Which, of course, include the dysgraphic class of exceptionality. It may be that he will eventually develop strategies for injecting meaning into fundamentally rote tasks, which will help with the Ramona/Wimpy situation, and possibly begin to see the intrinsic meaning in mathematics, which will help with raw computations.
Obviously, if the response medium (oral vs paper/pencil) is the primary correlate with his fluency, then fine motor skills would be a direction to look.
