Based on this phenomena, we reduce multiple surface forms to one basic segment, a phoneme, where the related segments derive by applying a rule to the underlying phoneme without affecting the meaning of the word. In order to predict the output of a module one has to know what the input was, as some part of the data cannot be predicted with the rules in the underlying form. (Eg. ‘Compute’ or ‘Compude’ is not predictable). One has to look at the morphology of the language to understand its phonology as allophonic details can be predicted using the rules and