How do actors simulate disfluency?

A dramatic actor's job is to present scripted material in a manner that appears to the audience as a convincing spontaneous scene. That is, the audience should feel like they are watching the unfolding of unrehearsed speech in real time. (Note that this is leaving out certain types of formulaic acting such as might be seen in a Shakespeare play). In order to be successful, what cognitive process is the actor actually going through? The processes cannot be identical to the processes they themselves would be undertaking when engaged in actual spontaneous speech off the stage (no matter how similar in context to something they portray on stage).

First, what is similar? If we can nail down some of the similarities then it may be easier to identify the variant points. The basic Leveltian process may be the same: The actor (1) conceptualize what they wish to communicate; they (2) formulate it into linguistic structures following syntactic, semantic, and pragmatic rules and conventions; and (3) send instructions to their oral apparatus to articulate these structures. At this point, though, one difference may become clear. If the dramatic performance is heavily scripted and exact wording is necessary to the progression of the story, then it is not clear that the actor can deterministically convert their conceptualization into the scripted formulation. That is, depending on various influences, they could conceivably produce their speech differently from performance to performance. But they do not (again, in the situation where the script is strict).

One possibility is that through training as an actor, they learn how to manipulate their conceptualizer into the exact same state from night to night so that they will output the exact same formulations again and again. From a neurolinguistic point of view, does this mean they would show precisely the same brain images from night to night?

If the performance is not so strictly scripted and the actor is allowed a certain license to vary word choice, timing, perhaps even syntactic configuration, then perhaps the situation is easier to explain. Nevertheless, they still have to simulate the conceptualization process in a repetitive way. Maybe this is not much different from anyone telling a story they have told many times before.

Still, though, the next problem comes with disfluency. One way that an actor can make their rehearsed speech sound more authentically spontaneous is with the inclusion of disfluencies. But what cognitive process does the speaker use to simulate this disfluency? Typical accounts of disfluency suggest that they are indicative of ongoing speech production difficulties. But if we're assuming that actors are in such complete control of their conceptualization so as to simulate the same state night after night, what causes them to have difficulty? Are they somehow actually introducing difficulty into their speech production process so as to authentically produce disfluencies? Or are they in complete control and the disfluencies are faked at a level that is beyond the capability of the average audience member to detect? If this latter scenario, would another professional actor recognize the "faked" disfluency?

The theme here so far seems to be that a trained actor is controlling themself. Does this mean that non-actors also can control themselves in comparable (but perhaps less proficient ways)? Can the average person also simulate disfluency in conversation when needed? Is it just another tool of prosody that can be manipulated in certain communicative ways? Certainly that's what some sociolinguists have suggested with such ideas as using hesitation to show politeness, or saving face, or other mitigating purposes.