Traditionally, spoken language has been characterized as relatively implicit (or context dependent) when compared with written language, which is seen as explicit. This paper examines the relationship between modality and context sensitivity. An informal but controlled experiment was designed to elicit comparable texts in eight different conditions, each condition serving as an instance of the permutation of the three variables under investigation: content (abstract vs. narrative), modality (written vs. spoken), and sender/receiver relationship (individual, real audience vs. multiple, imagined audience). Separate texts were elicited from 32 undergraduates and all referential noun phrases in each text were coded according to a modified version of Prince's (1981) taxonomy of given/new information. The major finding was that the main determinant of implicitness, when defined in terms of Inferrable vs. Evoked information, was the variable of content, not that of modality: abstract texts contained more Inferrable information (in Prince's sense) than did narrative texts, regardless of modality.