This sociolinguistic model predicts linguistic changes such as acoustic and lexical variation in the speaker in response to the addressee. Speakers tend to select linguistic forms on the basis of the knowledge, intentions, or goals of their addressee. Arnold, Kahn, and Pancani (2012) reported that speakers will adjust their pronunciation and use acoustically reduced tokens when the addressee has contextual understanding of the exchange. Findings from their study also demonstrate that speakers changed their utterance initiation and determiner duration. Essentially, listeners that speakers know are anticipating what will be said will include less information in what they say, and will speak more quickly with less use of determiners (the).
Research on audience design demonstrates multiple ways in which communicators might alter pronunciation to fit the anticipated needs of those listening to them. For example, when words are more common and well-known, speakers may not pronounce them as prominently as when the word is not well-known or is unexpected (Lindblom, 1990; Wright, 2004; Smiljanic & Bradlow, 2009). This is in opposition to the speaker-internal view of why acoustic productions may vary, which posits that more familiar or expected words are simply easier to pronounce due to their familiarity and the fact that they are predictable, therefore resulting in less effortful and prominent acoustic productions of these words (Balota et al., 1989;Bell, Brenier, Gregory, Girand, & Jurafsky, 2009).
Audience design takes place for example in Adult-infant interactions where in order to fit the needs of their addressees, adults will adjust their pronunciation.
For a related discussion of how signalers vary their communication based on who they believe will attend to it, see audience effect.