My guess would be that his eyes read faster than his mouth does. His eyes are reading ahead, and what is coming out of his mouth is lagging behind, and eventually he forgets what the next word is and has to go back with his eyes to find his place. That's why it's annoying to track with a finger, because it slows down the reading in the brain and makes it seem like a babyish pace.
This makes me think about something I realized about good readers a while back. In order to be expressive and fluent at reading aloud, we have to scan forward. It's how you know what to emphasize and what kind of tone to use. It's also why poor readers sound so robotic and boring. DS is a very expressive reader, suggesting that he absolutely does this. So maybe you're right. He also tells me that he doesn't lose his place when reading in his head, but of course, he could just be saying that.
My concern is that if there is an issue, it won't be caught at school for a long time, if ever.
