Embeddings of words and concepts capture syntactic and semantic regularities\nof language; however, they have seen limited use as tools to study\ncharacteristics of different corpora and how they relate to one another. We\nintroduce TextEssence, an interactive system designed to enable comparative\nanalysis of corpora using embeddings. TextEssence includes visual,\nneighbor-based, and similarity-based modes of embedding analysis in a\nlightweight, web-based interface. We further propose a new measure of embedding\nconfidence based on nearest neighborhood overlap, to assist in identifying\nhigh-quality embeddings for corpus analysis. A case study on COVID-19\nscientific literature illustrates the utility of the system. TextEssence is\navailable from https://github.com/drgriffis/text-essence.\n