Chapter 8 analyses and appraises a modern version of an inference about Neanderthal language drawn from data about the experimental making of stone tools by modern humans. This knapping inference looks skeletally as follows, the arrows depicting inferential steps: Modern humans knap Palaeolithic stone tools by means of action sequences structured in terms of hierarchies and recursion → Like modern humans, Neanderthals knapped stone tools by means of action sequences structured in terms of hierarchies and recursion → Like modern humans, Neanderthals processed language by means of action sequences structured in terms of hierarchies and recursion. Probing the submerged components of this inference, Chapter 8 advances two reasons for doubting its soundness. First, the two inferential steps are poorly warranted, involving weak analologies that disregard important differences between Neanderthals and modern humans. Second, the first inferential step lacks appropriate empirical grounding. That is, the sequences involved in the experimental knapping of stone tools by modern humans are not characterised by the hierarchies and recursion used by the syntax of human language.

Chapter 9 focuses on an inference about Neanderthal language drawn from data about the experimental teaching of stone-tool making to modern humans. Comprising three inferential steps, the knapping-pedagogy inference looks as follows in outline: Experimentally gathered data about the transmission of Oldowan technology to modern humans → Verbal language enhances the transmission of Oldowan technology to modern humans → Verbal language originated hundreds of thousands years ago as a prerequisite for Acheulean technology → Some Neanderthals had verbal language. There are various reasons for doubting the soundness of this inference. First, its empirical grounding is contentious: some experimental studies have found verbal interaction to be unnecessary for teaching knapping skills, even causing underperformance by modern learners. Second, Chapter 9 finds the inferential steps to lack the required logical force. For instance, it cannot be inferred that verbal language was a prerequisite for teaching knapping technology to modern humans from data that it only enhances the teaching of such technology to modern humans. This logical flaw leaves the final inferential step ungrounded.