What you call a long sandwich, how you say “pecan,” whether it’s “pop,” “soda,” or “coke” — your everyday words quietly map to where you’re from. Answer a set of questions and watch the map find where in the United States people talk the most like you.
How detailed do you want your dialect map to be? Select one of the modes below to start.
Your result is just one slice. The same survey becomes a playground on the companion page, which turns the numbers into seven interactive views:
This map uses data from the Harvard Dialect Survey, a study of American English by Bert Vaux and Scott Golder. For each question, the survey recorded the percentage of respondents in every U.S. state who chose each answer.
1. Scoring each state. Your similarity to a state is the average probability that a resident of that state matches your answers:
similarity(state) = average over your answers of P( resident of state gives your answer )2. Drawing the smooth map. We use Gaussian kernel smoothing to turn the state scores into a continuous surface, stretched to your range so contrast remains visible.
3. Cities & giveaway. Cities are scored by evaluating the smoothed surface at their locations. The "what gave you away" panel highlights choices that are far more common than nationwide.
The complete dataset behind this map is free to download. The CSV contains one row for every combination of state, question, and answer, with the percentage of that state’s respondents who chose it — 51 jurisdictions × 148 questions × their answer options (including 122 survey questions and 26 modeled questions).
The dialect-survey data originates with the Harvard Dialect Survey (Bert Vaux & Scott Golder) and is made available under the Creative Commons BY-NC-SA 3.0 license. Please keep that attribution if you reuse it.