If a primary text is to old to have a recording, how do historians deal with the known fact that non verbal communication is the majority of thought and expression of a person? If you read an old text, you're imposing your own sub vocalization on their words and we don't know what they gestured when they spoke or wrote their words.
I'm not entirely sure I agree with you that non-verbal communication is a major or necessary component of expression in all circumstances. Context and intent matters - if you're writing something down, especially something which isn't intended to be private, then you need to communicate your meaning to the intended audience with as little ambiguity as possible, which will likely mean deliberately avoiding the kind of expressions you mention. While it is of course possible (indeed, likely) that subtext, nuance and allusions can still be lost on someone who isn't the intended audience, this is why historians spend time and effort familiarising themselves with the cultural context of their primary materials. While this is always going to be an impefect process, particularly the further back in time our subject is, it's not exactly hopeless either, and we do the best with what we have. No historian ever works or expects to work with a perfectly preserved or unambiguous historical record. Any primary source - whether a video or audio recording, a written record, a diary or letter - requires interpretation and contextualisation to be meaningful. Doing this well is at the crux of historical practice.
Beyond the inherent imperfection of primary materials, I'm not entirely convinced that non-verbal elements of communication are that big a hurdle most of the time. Even dealing with oral testimony - something I do frequently in my own research - can be done effectively without recourse to visual preservation. Interviewees are well aware, after all, about the nature of the exercise, and that their audience is relying on reading a transcript or listening to the recording rather than watching them speak. This naturally affects the way they express themselves. The performative aspects of such testimony are still relevant - pauses, tone, emotionality and so on can be analytically useful - but it depends very much on the kind of work you're doing as to how vital they are. For most of the questions I was interested in, at least, the performance aspects of the recordings were fairly secondary. Recordings were useful to have, but relying on transcripts for some of the interviews for which the audio wasn't available did not bring my research to a screeching halt either.