The human brain has an astonishing capacity to understand spoken language, which is often fast, ambiguous, noisy and incomplete. An important mechanism underpinning speech comprehension is the ability to anticipate what someone is going to say next. This project investigates how the brain processes the first speech sounds of an incoming word to access the meaning of that word as quickly as possible. Using electroencephalography (EEG) to measure real-time brain activity at the millisecond scale, the four-year project funded by the Swedish Research Council will uncover the role played by predictive processing in rapid word recognition in Swedish, English and an artificial language. Taking advantage of our recently discovered brain potential – the pre-activation negativity (PrAN) – to test an unprecedented combination of factors crucial to word recognition, four experiments will reveal how different clues to a word’s identity present in the first few moments of an unfolding word allow the brain to segment the speech stream into individual words, and then anticipate the likely word ending. These clues include fine-grained phonetic information about upcoming sounds present in the signal, as well as the number of words that begin with similar speech sounds, and how frequent or expected these words are. By identifying and defining the neurophysiological responses to these clues, the field will gain important tools for research into speech perception, word recognition and general predictive processing.