This AI has to be fine-tuned to be deliberately verbose and jargon-heavy, either to make a good impression "Wowee, AI/author very very smart" (most likely), or (benefit of the doubt) to improve results on other LLM models drawing on it in the future.
Anyways, this tidbit is interesting (if correct at all)
(?) Not sure what homogeneous refers to in this context, whether it changes to something akin to "50-50 mix" of dialects instead of pure scribal dialect, or whether the statistical test points to there not being a slightest change between the scribes at all. I guess the reading depends on how one parses "at the hand boundary".
Funnily enough, this result is only brought once towards the end of the paper. No more details or attention was given to it anywhere on the previous ~30 pages. Lmao.
Anyways, this tidbit is interesting (if correct at all)
Quote:Folio 115r, where the hand changes partway down the page, is counted apart from the five hands’ 226 pages. Classified as a whole page it prefers the mixture to either pure dialect, and a change-point test favors a homogeneous (?) mixture over a switch of dialect at the hand boundary by nearly sixteen orders of magnitude.
(?) Not sure what homogeneous refers to in this context, whether it changes to something akin to "50-50 mix" of dialects instead of pure scribal dialect, or whether the statistical test points to there not being a slightest change between the scribes at all. I guess the reading depends on how one parses "at the hand boundary".
Funnily enough, this result is only brought once towards the end of the paper. No more details or attention was given to it anywhere on the previous ~30 pages. Lmao.