|Syntactic Confectionery Delight|
I did something similar a few years back. I found that the biggest speed boost came from partitioning the dictionary in advance - making a sepearte file for each word length (ie: 1.letter.words, 2.letter.words, 3.letter.words, etc...) so you don't have to loop over words that obviously don't match.
It's obviously not worth it if you're just going to run it once, but if you this is somehting you plan on doing again and again...
Another approach I've had good luck with on a different type of word problem was to divide the dictionary up based on the first character in each word. That wouldn't help in your case, but a varient would...
Assuming your dictionary looks like this...
at bat cat car tab
...reformat it so it looks like this...
abt bat abt tab acr car act cat at at
...then when you get a scrabbled word, sort the letters, and look it up (straight substring match, no regex needed) to find all the matches. To decrease the search space, you can partition the dictionary by first letter of the "word" (first letter when the letters are sorted that is) or by the length of the words, or by both.