Towards integrated microplanning of language and iconic gesture for multimodal output