The marketplace affords consumers different modes by which to express their preference. These include, but are not limited to, pressing a button, noting something on paper, and vocal expressions. New technologies enable consumers to interact with vending machines and even kitchen devices. Normatively these different modes of expression should not yield different choices, as there is nothing about the option (or the consumer) that differs as a function of expression mode. Yet, we find in our research that mode of expression has a substantial effect on people’s choices. The study of response mode effects is not new; four decades of research on decision-making show that preference construction can be influenced by the form in which these preferences are elicited (Lichtenstein and Slovic 1971; Tversky, Slovic, and Kahneman 1990; Nowlis and Simonson 1997). However, to date no studies have tested the effect of response modality—that is, the form in which the preference is expressed—on people’s choices. Previous research hints at the possibility of differences due to modality. In particular, vocal and non-vocal preference elicitation modes differ in what is required to translate one’s preference into the desired outcome. Most non-vocal preference elicitation modes--such as pushing a button at a vending machine--require the translation of one’s choice (e.g., Snickers) into another medium (e.g., button B4) to obtain the desired outcome. Speech, on the other hand, is directly connected to emotionality: During a child’s emotional development, speech serves as a main vehicle for emotional expression (Chen, Kennedy, and Zhou 2012). Numerous studies have shown that speech and emotionality are indeed strongly interlinked (e.g., Pavlenko 2002). We conjecture that translating one’s choice into another medium might therefore reduce emotional resonance, which suggests that the more that the situation calls for explicitly vocalizing one’s preferences, the more that emotional, automatic preferences will guide choice outcomes. This would imply that speech, more than non-vocal expressions, would produce decisions reflective of automatic emotional reactions, a prediction that was confirmed in three experiments.