Language is full of variability: any idea can be put into words in multiple ways, and each utterance is produced with varying degrees of success. My research program examines how individuals draw upon their unique experiences in order to communicate in the sorts of realistically-messy and ever-changing language situations that occur in the world. 

In this talk, I will discuss two recent lines of work. The first line of work focuses on comprehending mistakes when reading and listening. I examine what an error like "The shirt by the socks *were clean" means, how a listener's understanding of it unfolds incrementally through time, and how this process is informed by the expectations about production errors formed about speakers of different backgrounds. 

The second line of work focuses on variations in the success of conversational turn-taking. Corpora tell us that conversational turn taking is about as good as any practiced motor behavior can possibly be. What are the cues that speakers use in order to plan coordinated speech so successfully? I showcase a set of novel experimental and statistical methods to test this question, asking if it is more useful to pay attention to the content or the timing of the prior utterance. 

These two different domains come to converging conclusions. As producers and comprehenders, we can juggle multiple representations for language. This is what allows us to be flexible and resilient, allowing communication to be successful despite language's infinite variability.