IMHO, enter and blend (listen first), keep your own center (don't take it personally), connect to their center (see their positive intent), and extend ki (your own positive intention).
Very well put, I can't agree much more.
In my situations where verbal harmony can be considered, I usually:
1. Determine Maai (where each individual is "coming from", or their "stance"). In doing this, I usually become aware of the other individual's center, intentions, moods, etc..
2. Find a commonality between the two stances (blending). This can sometimes be very difficult, especially on the fly, and that blended harmony may not always be what is considered externally as "peaceful". Similar to an Aikidoka blending with the movement of an agressor, to remove them both from harm, it may require/result in some form of manifested conflict, ie; bodies colliding to an extent, or, in a verbal situation, concepts carried by words may collide.
3. Apply the most efficient technique (in this case, words and tone) to manifest the commonality, and thereby avoid conflict. This involves, IMO, empathy and love (intentional understanding, designed to create unity in some form between the two entities)
That's all sort of vague, as the concept can get pretty complex. But basically, analyze maai, enter then blend, and in so doing redirect the motion/intent of the agressor into a peaceful form.
Make sense? Hope so