Language Documentation (LD) and description have generated textual resources for minority tone languages of Africa. With the event of computers, text resources are being created mostly in the form of digital-born texts. Visual layout of texts in these languages follows a variety of representation models, even within the same writing systems. Some of the issues at stake are IPA vs Latin characters; omission vs surface representation of tones as accents; surface vs deep representation of tones, one tier vs multiple tier representation of linguistic analysis.
This paper brings out the limitations of tone representation and analysis as an upper layer of character strings, and suggests a model based on the TEI-XML markup language.