Journal of Pragmatics 33, 129-135, 2001
".. brilliant .. original, well-informed, well-written and intellectually exciting." Deacon's main idea is that language is unique to humans not because it is complex, but because it is symbolic. My main criticism focusses on his theory of the cognitive demands posed by symbols. He thinks they have to be embedded in a system of symbols, which raises obvious questions about how the first symbol can be learned, or could have developed. I suggest instead that they involve mind-reading: I have to be able to work out that when you say cat you are thinking of a cat. This is the first hurdle that a child crosses; and the second is to generalise from several such examples to the relationship 'meaning'. This makes symbol-learning much faster and easier.