K. and F. show that by analyzing self-corrections embedded in specific contexts one can obtain information about two central questions in language acquisition: What constitutes the child's knowledge about language at a given point in time? How could the child have acquired this knowledge?