I argue that assertoric content functions as a means forus to track the responsibilities undertaken by commu-nicators, and that distinctively assertoric commitmentsare distinguished by being generated directly in virtueof the words the speaker uses. This raises two ques-tions: (a) Why are speakers responsible for the contentthus generated? (b) Why is it important for us to distin-guish between commitments in terms of their mannerof generation? I answer the first question by developinga novel responsibility based metasemantics. I answerthe second by reference to the conflicting pressuresgoverning the resources we have available for apprais-ing speech.