Questionnaire design texts commonly recommend emphasizing important words, including capitalization or underlining, to promote their processing by the respondent. In self-administered surveys, respondents can see the emphasis, but in an interviewer-administered survey, emphasis has to be communicated to respondents through audible signals. We report the results of experiments in two US telephone surveys in which telephone survey questions were presented to interviewers either with or without emphasis. We examine whether emphasis changes substantive answers to survey questions, whether interviewers actually engage in verbal emphasis behaviors, and whether emphasis changes the interviewer-respondent interaction. We find surprisingly little effect of the question emphasis on any outcome, with the primary effects on vocal intonation and the interviewer-respondent interaction. Thus, there is no evidence here to suggest that questionnaire designers should use emphasis in interviewer-administered questionnaires to improve data quality. As the first study on this topic, we suggest many opportunities for future research.