Evaluating Automatic Spelling Correction Tools on German Primary School Children’s Misspellings

Abstract

Most existing spellcheckers have been developed for adults and it is yet understudied how well children’s texts can be automatically spellchecked, e.g. to build tools that assist them in spelling acquisition. This paper presents a detailed evaluation of six tools for automatic spelling correction on texts produced by German primary school children between grades 2 and 4. We find that popular off-the-shelf tools only achieve a correction accuracy of up to 46% even when local word context is taken into account. For many misspellings, the desired correction is not even among the suggested candidates. A noisy-channel model that we trained on similar errors, in contrast, achieves a correction accuracy of up to 69%. Further analyses show that this approach is very successful at candidate generation and that a better re-ranking of correction candidates could lead to a correction accuracy of ~90 %. Most of the remaining misspellings are so distorted that they are hard to correct without broader context. Furthermore, we analyze how the tools perform at different grade levels and for misspellings with different edit distances.