I take both views to be inconsistent with cognitive neuroscience. Or, at a more abstract, simplified, level, with Yudkwosky's suggestion of how to think about conceptual clusters, http://lesswrong.com/lw/nl/the_cluster_structure_of_thingspace/

There is also an important distinction between "Language" and "Languages" made by Lewis, relating to how people think about language as within, or separate from the world. I'm unable to find source though.