I had this problem a while back as well. This is what I did:
1. Create an output buffer and store the contents of the PHP output in a variable. (ob_start, ob_get_contents)
3. Tokenise the variable using the delimiter and you will have all the 'text' bits of your php file.
4. You should then be able to replace each token in that particular PHP file with either its translated equivalent or a function call to retrieve the text.
You can include code at the start and end of your PHP files to perform this sub-processing. I used the above process successfully on a very large site which has now been successfully converted into multiple languages. Hope this helps.