Error correction is proposed, wherein, on the basis of a data word, a syndrome calculation is carried out with a matrix M on the basis of a matrix H of a code, and, if the result of the syndrome calculation reveals that the data word is erroneous, the result of the syndrome calculation is transformed by means of a linear mapping. Next, an error vector is determined on the basis of the result of the linear mapping by means of an efficient error correction algorithm and the erroneous data word is corrected on the basis of the error vector.