Does language take cues from reality? Tallis says Yes; at least, that’s one kind of relation language has to reality.

His charming evidence: He notes that it’s more common to add “barking” to “dog” than to add other verbs. If language doesn’t take its cues from reality, “we must surely be at a loss to explain why the transition from ‘dog’ to ‘-is barking’ has a high frequency in observed speech while those from ‘dog’ to ‘-is quacking’ or from ‘dog’ to ‘-is reading Of Grammatology with pleasure and profit’ have much lower, or even negligible, frequencies.”