Obtaining referential word meanings from visual and distributional information: Experiments on object naming