In linguistics, the term T-unit was coined by Kellogg Hunt in 1965.
[1] It is defined as the "shortest grammatically allowable sentences into which (writing can be split) or minimally terminable unit."
T-units are often used in the analysis of written and spoken discourse, such as in studies on errors in second language writing.
The number of error-free T-units may be counted, as in Robb et al. (1986),[2] or changes in accuracy per T-unit overdrafts of compositions may be measured (Sachs and Polio, 2007).
Elements not counted as t-units include backchannel cues such as mhm and yeah, and discourse boundary markers such as okay, thanks or good.