Technologies for emotional modulation of natural language responses include a computing device that receives natural language requests from a user. The computing device identifies emotional features of the request and estimates an emotional state of the request by comparing the emotional features to a vocal pattern database. The computing device generates a natural language response and modulates the emotional content of the natural language response based on the emotional state of the request and the vocal pattern database. The computing device may modulate the natural language response to mimic the emotional state of the request, or to oppose the emotional state of the request. Possible emotional states include urgency, certainty, and dominance. Possible emotional features include acoustic, prosodic, and linguistic characteristics of the user request. The computing device may update the vocal pattern database based on the user request to adapt to the user. Other embodiments are described and claimed.