My understanding is that it is a requirement for multi-term analysis to only normalize tokens, and not eg. remove tokens (stop filter) or add tokens (by tokenizing or adding synonyms). Yet ASCIIFoldingFilterFactory.getMultiTermComponent will return a factory that emits synonyms if preserveOriginal is set to true on the original filter.
This looks like a bug to me but I'm not entirely sure how to fix it. Should the multi-term analysis component do ascii folding or not if the original factory has preserveOriginal set to true?