Обработка слов с частотными орфографическими ошибками (исследование на базе учебного корпуса английского языка)

Вестник Санкт-Петербургского университета. Язык и литература. 2023. Т. 20. № 4. С. 824–837.

Klimova M., Viklova A., Overnikova D.

The article presents an experimental study of the influence of the frequency of spelling errors in a word on its representation in mental lexicon. The hypothesis that frequently misspelled words cause difficulties in reading even if they are written correctly has been proved for native speakers of Russian and English. This paper aims to check the hypothesis on the basis of the learner corpus REALEC (Russian Error-Annotated Learner English Corpus), comprising texts by Russian L1 learners of English. The most frequently misspelled words collected from the corpus were used for the experiment that consisted in recognising correct and incorrect spellings. We analysed the influence of the following factors: the frequency of spelling errors in a word, its frequency in the corpus, entropy (the measure reflecting the amount of effort needed to choose between the variants of spelling), type of error. The results demonstrate the significance of entropy and frequency in the corpus, which corresponds to the findings of previous studies. A particular error type, substitution, has been found to be significant. Its special role corresponds both to the greatest difficulties this error type caused during the experiment and the results of previous research into written speech production of L2 English speakers, according to which substitution was considered the most frequent error type. The lower significance of the frequency of errors factor in comparison with the corresponding studies of L1 English can be explained by differences in language environments, in which learners of English are less exposed to incorrect spellings.

Keywords: ментальный лексикон, spelling errors, учебный корпус, learner corpus, mental lexicon, орфографические ошибки, word processing, обработка слов

Publication based on the results of: Second-language acquisition modelling within different frameworks of existing theories on the basis of learner corpora platforms for experiments and computer tools (2023)