Response to Pulvermüller: the syntax of actions and other metaphors

Let us first consider the notion of embedding in syntax with a simple example as in ‘the men [who he saw] were tall’. We have a phrase in square brackets [who he saw] with a specific structure and a specific distribution (i.e., it can appear in certain contexts but not in others); this phrase is embedded in another structure [the men were tall] where a dependency is established between ‘men’ and ‘were’. In fact, the verb ‘be’ cannot enter in a relationship with any other element of the sequence but ‘men’; if it did with the closer element ‘he’, as in ‘he was tall’, the sentence would be ungrammatical – ‘the men [who he saw] was tall’. Of course, not all ordered sequences involve embedding and dependency. Take for example the first letters of the alphabet, ‘abcdefg’. Although one can isolate a subpart of it – say the sequence [cde] – this subsequence is not endowed with a specific structure nor restrictions based on the context of distribution; moreover, there is clearly no dependence between the element and where the sequence has been embedded [ab…fg], unlike the case of the sentences above. The sequence could well be ‘abcdehg’.Let us now turn to actions, as in the case observed by Pulvermueller: [open the door [open the bottle close the bottle] close the door]. What we see here is analogous to the sequence of letters in the alphabet rather than the sequence of words in the sentence: the action [open the bottle close the bottle] does not have any special distribution and there is no dependency between [open the door] and [close the door], which is the structure where [open the bottle close the bottle] is ‘embedded’. In fact, there is no embedding at all; rather, there is just a sequence (although, of course, the actions may be connected in a causal way). Suppose that the door is built in the middle of a wall that stands alone in a garden. I can open the door and go through it without coming back and closing the door; indeed, [close the door] in the sequence under discussion can be substituted by a virtually infinite number of different types of action [[sing Aida backwards], [jump], [open another door]] yielding, for example, [open the door [open the bottle close the bottle] jump], unlike the case of verb agreement we observed. Moreover, in human languages dependencies can be nested as in ‘if [the men [who he saw] were tall] then it was raining’. The agreement dependency we observed before between ‘men’ and ‘tall’ is nested in another independent dependency between ‘if’ and ‘then’; that this is a dependency is trivially clear because ‘then’ cannot be substituted by, say, ‘or’, which would rather require a dependency with ‘either’. Notice that embedding goes on without bound in human language. Consider this simple example: let S be a sentence containing n nested dependencies and T a sentence; ‘if S then T’ is a grammatical sentence but ‘if S or T’ is not. All humans know this without explicit instructions or training and this has been proven to be mathematically beyond the bounds of finite automata, as observed 60 years ago [1xThree models for the description of grammar. Chomsky, N. IEEE Trans. Inf. Theory. 1956; 2: 113–124Crossref | Scopus (626)See all References[1]. Regarding questions related to memory limitations, note that it is well known that, without external aids, people can understand such examples with up to about seven nested dependencies [2xFinitary models of language users. Miller, G.A. et al. : 419–491See all References[2]. This is the behavioral correlate of short-term memory limitations. Finally, note that if we refer to ‘usage’ as a theoretical notion to uncover the capacity of the human mind to deal with a code and apply it to arithmetic, we are led to the unacceptable consequence that knowledge of arithmetic is finite, because without external aids people are simply not able to perform unbounded operations.In summary, disregarding the absence of an analog of functional words and locality conditions for actions (syntactic movement not being a special case at all), the idea of ‘a syntax of actions’ remains a metaphor if compared with the syntax of any human language. Perhaps, as noted in the paper, if we shift from actions to action planning some analogies with syntax could be explored, although it remains to be proved that this activity is not essentially parasitic on language [3xPlans and the Structure of Behavior. Miller, G.A. et al. CrossrefSee all References[3]. Given that animals seem to be able to plan actions, I still consider this separation between actions and syntax a welcome fact for any research on the human brain and evolution [4xEvolution, brain, and the nature of language. Berwick, R. et al. Trends Cogn. Sci. 2013; 17: 89–98Abstract | Full Text | Full Text PDF | PubMed | Scopus (57)See all References, 5xPrinciples of Neural Science. Kandel, E. et al. See all References, 6xThe Equilibrium of Human Syntax. Symmetries in the Brain. Moro, A. See all References, 7xThe Boundaries of Babel. The Brain and the Enigma of Impossible Languages. Moro, A. See all References]. Once more, no commitment appears to be empirically and theoretically legitimate, unless these preliminary issues have been clarified.