A LISP program to determine similarity relations in letter displays

Similarity relations between letters are important in displays used in tacbistoscopic recognition experiments. A LISP program is described that calculates similarity relations on the basis of an empirical letter-similarity matrix or on the basis of feature sharing in a feature-set definition supplied by the user. With either derivation method, there are three options for the type of average to be calculated for target display: (1) the average for the whole target, (2) the average for each letter in the target paired to all other letters in the target, and (3) the average of each letter in the target paired with its immediate neighbors.