In section 10 of the target article, Vaesen reviews some of the approaches that link increased complexity in tool use to language evolution. For example, several proposals have seen in the mode of production of Oldowan choppers (2.6 mya) and/or Acheulean hand axes (1.7 mya) the kind of structural complexity that characterizes human language: that the kind of finer motor control involved in tool use facilitated speech control (Calvin Reference Calvin, Gibson and Ingold1993); or that hand motor control for tool use was rather instrumental in the appearance of a gestural mode of communication that is supposed to predate oral language (Arbib Reference Arbib2005; Rizzolatti & Arbib Reference Rizzolatti and Arbib1998); and that such tools illustrate the kind of hierarchical structure that is characteristic of language (Greenfield Reference Greenfield1991).
These approaches take for granted an early scenario for language evolution; that is, that the structural capacity required for grammar is already present in the abilities manifested in early hominin tools. Late scenarios for language evolution, on the contrary, contend that language is connected to the transition to Homo sapiens and its diaspora from Africa, which started about 120,000 years ago. In other words, it is in the behavioral complexity and cultural explosion of the past 100,000 years of human evolution where the critical changes that account for human uniqueness are to be looked for. In so doing, a stronger evolutionary discontinuity is established between human and non-human primates. Given Vaesen's concern in the target paper to argue for a cognitive discontinuity between human and non-human primate tool use, he should be sympathetic to such late scenarios and pay attention to the structural complexity involved in the tools of this age, rather than just the early ones he discusses.
It is from this standpoint that Upper Paleolithic knotting technologies (Adovasio et al. Reference Adovasio, Sofer and Klima1996; Sofer et al. Reference Sofer, Adovasio, Illingworth, Amirkhanov, Praslov and Street2000) are relevant in this context. Knotting technologies are involved in basketry, nets, and textile weaving. Archeological remains of such crafts consist of clay imprints, dated circa 30,000 years ago. Binding by knotting can also be inferred in the case of spears, harpoons, and arrow heads and their corresponding bows (c. 70,000 years ago), which had to be strongly attached to their shafts (older spears were glued to their bases; see Wadley Reference Wadley2005). Perforated ornaments, such as necklaces, bracelets, and wristbands, also were tied. Two features of these tools need to be underlined: They appear just in the last 70,000 years, and they are associated with Homo sapiens sites (ornaments also appear in late Neanderthal sites in Europe, but after they got in contact with the newly arrived H. sapiens).
It has been suggested (Camps & Uriagereka Reference Camps, Uriagereka, Rosselló and Martín2006) that the formal structure of knotting is similar in complexity to a context-sensitive grammar, such as that required to capture recursivity in human language. Relying on Chomsky's (Reference Chomsky1959) formal hierarchy, Camps and Uriagereka claim that knots also could involve a context-sensitive generative procedure. They observe that the procedure for tying a knot cannot be specified as an iterative sequence of steps (a finite automata), because each step in the procedure has to have access to previous stages and the material context of physical forces the knot is made to resist. The kernel of their reasoning is the claim that the knotting procedure involves higher-level units (“phrases”), which are deployed according to the material context. This is obvious when one thinks of the sophisticated ways of sailor knots. In general, though, the two hands have to mutually coordinate along the procedure, rather than doing one thing after another. Context sensitivity is linked to operational memory requirements, to keep track of each operation, taking the previous and next ones into account at the same time. From this, Camps and Uriagereka claim that the archeological evidence of knots – even if indirect and inferential – provides the best indication available of a cognitive complexity equivalent to that required by human language.
We think that their case can be strengthened in two directions: First, whereas in projectiles and perforated ornaments a single knot may be required to fix two elements together or to string them around the body, in textiles, nets, and basketry a series of knots is involved, within a general constructive plan (therefore, a more complex, context-sensitive, generative procedure). Second, such a series of knots is in the service of recursive patterns, which can be transformed, following distinct axes of symmetry, for example. Simple iterative processes are clearly not enough to generate such complex structures, where each single operation is conditional on the state of the rest of the fabric and the physical forces the knot is supposed to resist.
Besides, knotting cannot be accounted for in terms of Greenfield's “action grammar,” which is equivalent to a finite-state automata. She contented that the structural complexity of language can be also found in hierarchical organization of action. The kinds of actions she paid attention to, though, such as “Russian dolls” inclusion, are developmentally easier than knot tying is, and they are also within the reach of non-human primates (whereas knots are beyond the capacities of chimpanzees, according to Josep Call, personal communication). Therefore, the attempt to view recursion in terms of Greenfield's “action grammar,” as it has been recently suggested (Fujita Reference Fujita2009), does not pay proper attention to the context sensitivity of recursion.
On the other hand, the proposed connection between knotting and language entails that the program proposed by Hauser et al. (Reference Hauser, Chomsky and Fitch2002) got it right that recursion is uniquely human, but wrong that it is a uniquely linguistic capacity, even if the evidence is still not enough to decide how it came about: It could be a general capacity, deployed in different domains, or a domain-specific one that was exapted in others (Barceló-Coblijn, in press). It also offers a plausible hypothesis to set apart the linguistic capacities of sapiens and neanderthal, given that both species cannot be distinguished at the speech level (Barceló-Coblijn Reference Barceló-Coblijn2011). In summary, context-sensitive rules offer a principled mark of modern humanity, beyond the typical lists of modern behaviors that can be found in archeology (Henshilwood & Marean Reference Henshilwood and Marean2003).
In section 10 of the target article, Vaesen reviews some of the approaches that link increased complexity in tool use to language evolution. For example, several proposals have seen in the mode of production of Oldowan choppers (2.6 mya) and/or Acheulean hand axes (1.7 mya) the kind of structural complexity that characterizes human language: that the kind of finer motor control involved in tool use facilitated speech control (Calvin Reference Calvin, Gibson and Ingold1993); or that hand motor control for tool use was rather instrumental in the appearance of a gestural mode of communication that is supposed to predate oral language (Arbib Reference Arbib2005; Rizzolatti & Arbib Reference Rizzolatti and Arbib1998); and that such tools illustrate the kind of hierarchical structure that is characteristic of language (Greenfield Reference Greenfield1991).
These approaches take for granted an early scenario for language evolution; that is, that the structural capacity required for grammar is already present in the abilities manifested in early hominin tools. Late scenarios for language evolution, on the contrary, contend that language is connected to the transition to Homo sapiens and its diaspora from Africa, which started about 120,000 years ago. In other words, it is in the behavioral complexity and cultural explosion of the past 100,000 years of human evolution where the critical changes that account for human uniqueness are to be looked for. In so doing, a stronger evolutionary discontinuity is established between human and non-human primates. Given Vaesen's concern in the target paper to argue for a cognitive discontinuity between human and non-human primate tool use, he should be sympathetic to such late scenarios and pay attention to the structural complexity involved in the tools of this age, rather than just the early ones he discusses.
It is from this standpoint that Upper Paleolithic knotting technologies (Adovasio et al. Reference Adovasio, Sofer and Klima1996; Sofer et al. Reference Sofer, Adovasio, Illingworth, Amirkhanov, Praslov and Street2000) are relevant in this context. Knotting technologies are involved in basketry, nets, and textile weaving. Archeological remains of such crafts consist of clay imprints, dated circa 30,000 years ago. Binding by knotting can also be inferred in the case of spears, harpoons, and arrow heads and their corresponding bows (c. 70,000 years ago), which had to be strongly attached to their shafts (older spears were glued to their bases; see Wadley Reference Wadley2005). Perforated ornaments, such as necklaces, bracelets, and wristbands, also were tied. Two features of these tools need to be underlined: They appear just in the last 70,000 years, and they are associated with Homo sapiens sites (ornaments also appear in late Neanderthal sites in Europe, but after they got in contact with the newly arrived H. sapiens).
It has been suggested (Camps & Uriagereka Reference Camps, Uriagereka, Rosselló and Martín2006) that the formal structure of knotting is similar in complexity to a context-sensitive grammar, such as that required to capture recursivity in human language. Relying on Chomsky's (Reference Chomsky1959) formal hierarchy, Camps and Uriagereka claim that knots also could involve a context-sensitive generative procedure. They observe that the procedure for tying a knot cannot be specified as an iterative sequence of steps (a finite automata), because each step in the procedure has to have access to previous stages and the material context of physical forces the knot is made to resist. The kernel of their reasoning is the claim that the knotting procedure involves higher-level units (“phrases”), which are deployed according to the material context. This is obvious when one thinks of the sophisticated ways of sailor knots. In general, though, the two hands have to mutually coordinate along the procedure, rather than doing one thing after another. Context sensitivity is linked to operational memory requirements, to keep track of each operation, taking the previous and next ones into account at the same time. From this, Camps and Uriagereka claim that the archeological evidence of knots – even if indirect and inferential – provides the best indication available of a cognitive complexity equivalent to that required by human language.
We think that their case can be strengthened in two directions: First, whereas in projectiles and perforated ornaments a single knot may be required to fix two elements together or to string them around the body, in textiles, nets, and basketry a series of knots is involved, within a general constructive plan (therefore, a more complex, context-sensitive, generative procedure). Second, such a series of knots is in the service of recursive patterns, which can be transformed, following distinct axes of symmetry, for example. Simple iterative processes are clearly not enough to generate such complex structures, where each single operation is conditional on the state of the rest of the fabric and the physical forces the knot is supposed to resist.
Besides, knotting cannot be accounted for in terms of Greenfield's “action grammar,” which is equivalent to a finite-state automata. She contented that the structural complexity of language can be also found in hierarchical organization of action. The kinds of actions she paid attention to, though, such as “Russian dolls” inclusion, are developmentally easier than knot tying is, and they are also within the reach of non-human primates (whereas knots are beyond the capacities of chimpanzees, according to Josep Call, personal communication). Therefore, the attempt to view recursion in terms of Greenfield's “action grammar,” as it has been recently suggested (Fujita Reference Fujita2009), does not pay proper attention to the context sensitivity of recursion.
On the other hand, the proposed connection between knotting and language entails that the program proposed by Hauser et al. (Reference Hauser, Chomsky and Fitch2002) got it right that recursion is uniquely human, but wrong that it is a uniquely linguistic capacity, even if the evidence is still not enough to decide how it came about: It could be a general capacity, deployed in different domains, or a domain-specific one that was exapted in others (Barceló-Coblijn, in press). It also offers a plausible hypothesis to set apart the linguistic capacities of sapiens and neanderthal, given that both species cannot be distinguished at the speech level (Barceló-Coblijn Reference Barceló-Coblijn2011). In summary, context-sensitive rules offer a principled mark of modern humanity, beyond the typical lists of modern behaviors that can be found in archeology (Henshilwood & Marean Reference Henshilwood and Marean2003).