This article offers a close study in using TEI to represent the distinctive forms of multilingual glossing found in premodern texts, including where existing norms for encoding practice may not always leave room for the specificities of medieval practice. Drawing on the data set from a new edition of Walter de Bibbesworth’s Tretiz (a thirteenth-century rhymed vocabulary of French, extensively glossed into Middle English across its many surviving manuscripts), it outlines how <term> and <gloss> elements might usefully be deployed to illustrate the relations between lexical items both on the manuscript page and across a broader manuscript tradition.