An analysis of oral and written discourse focuses on the hypothesis that marked structures serve to code discourse boundaries and signal the advent of new discourse units. The hypothesis is examined against three independent sets of data: (1) discourse from a larger study, in which subjects provided oral or written narrative descriptions, in English, of the same event (from a movie); (2) an identical study except that subjects provided descriptions in Mandarin Chinese; and (3) analyses of two texts, both novels in English. In each analysis, markedness distribution in initial and non-initial positions was examined. It is concluded that all three studies provide independent supporting evidence for the hypothesis from both written and oral narrative discourse. Contains 18 references. (MSE)