When users speak something that does not match with an utterance in the interaction model, Alexa will result in an error EXCEEDED_MAX_REPROMPTS. As a developer, it is not possible to know what did the user spoke which resulted in this error.
Developers need a way to know the user's input that results in a utterance mismatch and can update the utterance model accordingly.