It has been demonstrated (in the research described here) that people hearing sentences on a tape recording process underdetermined sentences differently than people hearing them in a conversation. It has been shown that this difference is caused by the presence of underdetermined elements but not ordinary transformations. The effect of ordinary transformations in both modal and non-modal sentences is the same in both conditions. The possibility that the effective factor here is condition alone or underdeterminacy alone has been ruled out. Instead an interaction between condition and sentence type has been demonstrated. The demonstrated interactions was predicted from the nature of underdeterminacy. Underdetermined elements are any elements whose main function is to convey information about the state of a speaker by injecting that information into an otherwise purely factual proposition. Hence, these elements should not have their real meaning in the absence of a speaker. Results from counting equivalent and non-equivalent substitutions along with other findings suggest that subjects in conventional tape experiments do not attend to sentence meaning to the same extent as do subjects in ordinary conversations. It may well be that earlier findings can be attributed to the unnatural importance of syntax in a tape experiment. (Author/AMM)