(2) The experimenter has one event with multiple key word sets.
Here the key word sets most likely have some common words and they are therefore not independent sets. The Null hypothesis says that each of these key word sets has, under the given experimental protocol, ELSs in a random arrangement as might be expected in a monkey text. The Alternative hypothesis is that one [at least one] of these key word sets, under the given experimental protocol has ELSs in a more compact arrangement than expected by chance.

(3) The experimenter has multiple events each with multiple key word sets.
Here the key word sets associated with the same event most likely have some common words and they are therefore not independent sets. The Null hypothesis says that each of the events has all of its key word sets having ELSs in a random arrangement as might be expected in a monkey text. The Alternative hypothesis is that, under the experimental protocol, all but one of these events have at least one of their key word sets having ELSs in a more compact arrangement than expected by chance.