What the corpus reveals.

Every letter in the corpus is classified by the author's relationship to the patient. Comparing register features across these four voices reveals two independent shifts. The first is a person shift: patients narrate in the first person; everyone else — family, attending physician, Cullen — uses the third person at roughly similar rates. This shift happens at the boundary between self-narration and narrating someone else's illness, regardless of medical training.

The second is an epistemic shift: Cullen hedges roughly three times as much as attending physicians, who have examined the patient and describe findings with relative confidence. Cullen, working from text alone, qualifies with "may," "might," "perhaps." This suggests the clinical register was already shared between practitioners — but the epistemic stance within it varied by proximity to the bedside.

The four-voice comparison above shows a register gap between patient and physician voices, but treats it as static. Here the same features are re-computed decade by decade — so we can ask whether the gap is widening, narrowing, or stable across Cullen's career.

If the register gap between patient and physician voices remains stable across decades, the clinical register was already established before Cullen's career began — pushing its origins further back than the hospital. If the gap widens, we may be watching a genre consolidate in real time. The table above lets you check each register feature independently: pronoun use, hedging, passivization, sentence complexity.

When a patient and their attending physician both write to Cullen about the same illness, the same clinical reality passes through two different registers. Below, we compare patient-authored letters against attending-physician letters only within cases where both voices are present — controlling for the illness itself. What distinguishes the two is not what they describe, but how they describe it.