For the Seward Family Digital Archive (sewardproject.org), I'm working encoding a piece of paper which was used to facilitate an in-person conversation between William Henry Seward, his wife Frances, and a deaf-mute artist named John Carlin. In the conversation, Frances wrote for herself and on her husband's behalf. In other words, we will frequently need to distinguish between the scribe in the text, and the "voice."
I'm thinking of using <handShift/> as we do for the correspondence in our collection, with the scribe attribute used to identify handwriting and the resp attribute to indicate that the text was dictated (by identifying the person speaking).
Has anyone in the TEI community worked with documents like these before, and how did you approach the encoding?