So there is a contradiction with tools like this that has always bothered me. LanguageTool is great. It is so useful, and I couldn’t imagine writing anything online without it. However, that is a problem. See, LanguageTool, and other services of its kind, have zero functions to aid you in internalizing the corrections. This, specifically bothers me for spelling mistakes. Grammar can be subjective and subject to the style of writing. Spelling, however, is mostly a regional difference. You are either spelling the word correctly, or you’re spelling the word incorrectly. It is a binary that can be solved for.
LanguageTool does not log specific words that you misspell most frequently. It does track how often you correct a misspelling, but that is the limit. Having a tool like LanguageTool shoulder the burden of knowing how to spell means that the end user does not learn. The system automatically replaces the word.
Ideally, LanguageTool would log your most frequently misspelled words and then offer you a kind of typing test that asks you to type the word out correctly, and through repetition the user would learn how to spell the word correctly.
I’m not confident that a feature like this would ever be implemented, however, because of the above contradiction. If LanguageTool actually improved my ability to use correct spelling and even grammar, then I would have less need for the service as a whole.
Since LanguageTool is open source, however, that doesn’t exclude the user base from creating such a tool that interfaces with LanguageTool. What would need to be done is that there would need to exist an API proxy, one that sits between the user and the LanguageTool servers and logs misspelled words and the selected corrections, as well as the frequency of misspellings (with a datetime stamp). This could allow a tool to know if a misspelling is trending down or up and offer spelling/typing challenges to the user based on their most frequently misspelled words.
Has anything like this been attempted previously? Is an API proxy something that could be implemented? The typing challenge interface would be comparably simple to implement.