ABSTRACT
One way of presenting the results is to investigate how the performance constraints influence the probability of finding syllable x in the subsets of k syllables. By pooling all the 133 subsets of k syllables and by plotting the observations in the form of a distribution diagram, a way of answering that question is obtained. This has been done in Figure 23.2 where we usc a 7 by 19 matrix for the 133 syllables and where the largest vertical lines indicate that the syllable in question was observed as occurring in all 133 (=100%) of the simulations. When evaluating these results we should bear in mind tl1e distribution for a corresponding series of derivations in which the enumeration of syllables takes place in a completely unbiased fashion, that is without any performance constraints at all. That experiment wonld in principle give us a uniform distribution of vertical lines implying that all syllables would be equally probable. As can be seen, the performance constraints make tl1e results deviate markedly from a pattern with completely uniform preferences.
