The vocabulary effect is the retrieval advantage that content written by genuine subject matter experts has over keyword-optimized content. Experts naturally use the full semantic vocabulary of a concept — related terms, adjacent ideas, domain-specific language — producing a denser, more coherent embedding closer to the target cluster in vector space.
The vocabulary effect explains why content written to a keyword brief underperforms content written by someone who actually knows the subject. The expert is not optimizing for embeddings — they are just talking about the topic correctly, and correct talk about a topic uses the topic’s vocabulary. Keyword repetition does not replicate this; semantic vocabulary does.