If you add language that says "reasonably interpreted to be threatening" you are getting closer to an objective test. But that is not what was proposed here. The problem occurs when you assign to one person the responsibility for another person's feelings, irrespective of the context or reasonableness.