This paper describes a set of simulations which explore the way different features of lexical organisation affect the probability of finding a pair of associated words in a set of five randomly selected words. The simulation is equivalent to giving Ss a set of five words and asking if they can identify a pair of associated words among them. The paper speculates that it might be possible to extrapolate from a simple test of this sort and derive some interesting claims about the number of links connecting words in L2 speakers' lexicons.