We describe a series of preliminary experiments undertaken to investigate the relationship between complicated tool-making and the presence or absence of language in its communicative role. The experiments involved teaching two groups of university students how to make Levallois flakes by either verbal or non-verbal demonstration.The rates and mean times of acquisition of the Levallois technique and of successful flake production were compared. They did not differ significantly between the two groups. From these results, we infer that spoken language was not indispensable for Levallois flake production in the Middle Palaeolithic.