For manually imported items, would it be possible to introduce an alternative marker to delimit words/phrases/sentences or to provide an option for users to indicate that the text is already parsed and parsing should be skipped during the import process.
If there were an option to skip parsing that would work. Otherwise I have used ChatGPT to break down specific sentences if I am not sure how and it has done a decent job, but then it gets parsed on import. Since spaces are ignored by the parser it wouldn’t be difficult to request ChatGPT to substitute a symbol like ‘§’ or even consider a character substitution such as ‘%20,’ which is commonly used in HTML to represent spaces (a custom substitution may work better). If MeCab is being used to parse, then LingQ could feed the items between the markers into the MeCab API separately.