[chryuop]

Yup, very confusing...above all when I know other 3 languages and they all mix up. You see in the Latin languages I know there is a huge difference between the o and de: one would indicate the capability of speaking a language and the other the actual action of speaking it. Using the Japanese with the Latin languages the difference would be like this (of course I speak only about the verb hanasu, not valid for any verb):
日本語を話す。　　I have the capability of speaking Japanese.
日本語を話している。 I am speaking Japanese now.
日本語で話す。 I am speaking Japanese now.
Of course then you can use the potential form.
In fact my Japanese-Italian dictionary gives me the example　ｲﾀﾘｱ語を上手に話す
as having the capability of speaking well the Italian language and not speaking at the moment Italian in a good way.