Is it possible to derive statistics about bodies of English text  in general from measurements of a dictionary ? For example, if I have a chart of the letter most likely to come before any given letter in the alphabet taken from a dictionary, is that applicable to, say, a journal article?

A dictionary isn't distributed like the usual English text, so I'd say no. Richer metrics would be required to make that sort of translation.
~~~~