Can Topic Modelling create new words (words not in the document vocab) based on semantics ?

I did try to understand it as a beginner and found that only lemmatization / stemming happens + removal of stop words. But couldn't find literature to check if LDA Topic Modelling can actually understand the text semantics and generate new words ( words within a topic).

So for example: If manual, publication , novel are frequently appearing words - It gets merged into something synonymous like "book"

languagetechnology modelling semantics words