To investigate whether the impact of phoneme merging on recognition rate can be predicted, different measures to quantify the relationship between two phonemes a and b were compared: (1) the functional load of their opposition, (2) the bigram type preservation, (3) their information radius, (4) their distance within an information gain tree induced from a distinctive feature matrix, and (5) the symmetric Kullback-Leibler divergence. For each of 25 phoneme pairs we trained a speech recognizer on data in which the respective pair was merged. Based on correlation analyses and predictor selection in stepwise regression modelling we found that the impact of phoneme merging on accuracy can tentatively be captured in terms of functional load and tree distance between the merged phonemes.