Sounds like a dumb conjecture, how do you explain reading and talking on the phone then?
We learn to read aloud, and there's usually a long transitional period of subvocalization. And even people who can read silently, if they want to very carefully understand something, often subvocalize it - even moving the lips silently. My conjecture is this process never goes away entirely.
> talking on the phone
Same mechanism. We hear the sounds and imagine ourselves making the muscle motions to produce such sounds.