Static word embeddings that represent words by a single vector cannot capture
the variability of word meaning in different linguistic and extralinguistic
contexts. Building on prior work on contextualized and dynamic word embeddings,
we introduce dynamic contextualized word embeddings