Abstract Automatic spell checker systems aim to verify and correct erroneous words through a suggested set of words that are the nearest lexically to the erroneous ones. The performance of such corrective systems differs according to the correction mechanisms implemented. The lexis of the automatic spell checker tools represents a main disadvantage. In this paper, we address the lexicon insufficient problem used in the automatic spell checker and we treated the problem of vocabulary management and its relation to the performance of a system of automatic spelling correction. In the first place, we developed an approach that aims to correct the derived words, considering that the most Arabic words are derived ones by adjusting the Levenshtein algorithm to our need, then, we show a dictionary management that helps us to minimize the number of comparisons. In this context, two axis were studied, the organizational aspect and the morphological aspect. For the first axis, we proposed a vocabulary management allowing the optimization of the access number to the lexicon’s file. For the morphological level, we exploited the derivation notion to reduce the number of the words to compare avoiding redundant comparisons.