Cognitive Reframing of Negative Thoughts through Human-Language Model Interaction