Previously, I mentioned Malcolm Parkes’s transfer units and again in relation to how scribes worked. I’d like to not a relatively recent article that gives some modern, empirical support to the transfer unit, albeit somewhat indirectly.
In “Syllables as functional units in a copying task” (Language and Cognitive Processes 21.4 (2006): 432-452), Sonia Kandel and Sylviane Valdois examined children from different classes in elementary school while executing a copying task. A word appeared on the screen and the children copied it using a graphic tablet; the children were filmed to record the points at which they lifted their gaze. The idea behind the task was that “the child produces a gase lift because he/she does not have enough information on the spelling of the following letter-string” (436). It concludes that ‘the younger children used syllables to articulate visual parsing and motor programming. The older children used whole word visual units but still organised their graphic production according to the syllable structure” (432).
In short, if we can imagine young children copying syllable by syllable and then increasingly word by word (with age and practice), the sense unit, a clause or phrase that can be seen as a unit of meaning, appears to be a good heuristic for thinking of the way a well trained and experienced scribe would work.
Also of passing note, the article notes the paucity of copying tasks in the study of handwriting production, stating “A copying task seems to be a precious research and pedagogical instrument to investigate the links between perceptual and action mechanisms underlying spelling processes” (448). So perhaps we can look forward to more contemporary research that may have bearing on past practice.