In a text or in general a sequence, adjacency statistics measure the frequency with which words or tokens occur next to each other. For example, in the sequence BAAACBBAAC, the pair (or 2-gram) AA occurs three times, AC and BA each occur twice, and BB and CB each occur once. No three tokens (3-gram) in the example occur more than once, but in longer texts 3-grams (such as the words 'artificial neural network') and longer sequences may recur.
Used on page 410