Distributional semantics is a theory in computational linguistics that posits words with similar meanings will appear in similar contexts. It forms the basis for many natural language processing techniques by representing word meanings through vectors derived from their distributional properties in large text corpora.