If we find language growing more and more metaphorical, the further back we go into the past, what possible justification can there be for assuming a still earlier time when it was not metaphorical at all? Thus, Max Müller postulated a 'metaphorical period', during which the progress from literal to metaphorical meanings must have taken place. But, what is this but a purely arbitrary surmise? And is it not highly improbable?
Owen Barfield, The Owen Barfield Reader. Wesleyan University Press (London: 1999) p. 108.