1. Target Article Author: James R. Hurford

　 2. Word counts: Abstract: 55

　　　　　　　　　　　　　　　　　　　　 　Main Text: 960

　　　　　　　　　　　　　　　　　　　　 　References: 222

　　　　　　　　　　　　　　　　　　　　 　Total: 1237

　 3. Commentary Title: Word-sentences and an interaction-based account of language evolution

　 4. Commentary Author Name: Bipin Indurkhya

　 5. Department of Mechanical Systems Engineering, Tokyo University of Agriculture and Technology, 2-24-16 Nakacho, Koganei, Tokyo 184-8588

　 6. Telephone Number: +81-42-388-7096

　 7. Email Address: bipin@cc.tuat.ac.jp

　 8. URL: Not available

　 9. ABSTRACT

Considerations from an interaction-based approach to the evolution of language and the role of word-sentences therein show that the object-attribute ontology is arrived at a much later stage. Therefore, Hurford’s arguments, by focusing on the predicate-argument structure, seem to miss out on the most of the interesting aspects of the early stages in language evolution.

　10. MAIN TEXT

Although Hurford presents persuasive arguments to provide a neural grounding for the object-predicate distinction, they nevertheless constitute a big leap across the landscape of language evolution that seems to miss many of its interesting aspects. In my comments below, I attempt to show a glimpse of some of these skipped-over sights.

For instance, when Hurford distinguishes his account with the proposal of Rizzolatti and Arbib (pp. 14 – 15), he argues that a structure like “grasp-A(raisin)”, corresponding to a canonical F5 neuron as proposed by Rizzolatti and Arbib, is really a shorthand for the two-stage process corresponding to “raisin(x)” and grasp-A(x)”. The crux of Rizolatti and Arbib’s proposal, however, is to highlight this evolutionary stage of language where the action is inseparably linked to the object: the stage where the cognitive agent can be aware of something as a raisin only by being aware of how to grasp it. It is only in the later stages of evolution that the action and the object become differentiated corresponding to the two-stage process of Hurford.

Indeed, it has long been recognized that the early stages of language evolution are characterized by word-sentences. (See Aichinson 1996; Barrett 1995; Bloom 1973; Cassirer 1955; Langer 1957; Piaget 1962; Wegener 1885, Wray 1998, 2000). In a word-sentence, the predicate and the object ― and various other qualifiers ― are fused together, so that the child’s first words refer to complete situations. It is only through numerous later interactions that the various objects, attributes, and actions occurring in a situation get gradually differentiated and the child is able to refer to them individually. Even then, the holistic language continues to play a major communicative role in adult language (Wray and Perkins 2000).

This long evolutionary road leading to the predicate-argument structure can be further highlighted by considering another example mentioned by Hurford (p. 21): in going from “If it moves, grab it” to “If it catches your attention, inspect it carefully and figure out what to do with it.” Along the way, there may be a stage of “If it moves, I can grab it,” where the action is contemplated without being carried out. Then, once the action, “grab-it,” and the object, “it-moves,” are differentiated, there may be a stage where each of these schemas is exercised in order to increase familiarity with it. That is, “grab-it” action is applied to various objects, so that everything becomes something to be grabbed. In this process, the “grab-it” action may become more refined and more detached from the original object. Similarly, other actions are applied to the “it-moves” object, and from these interactions a more comprehensive representation corresponding to the “moving-object” may result. (See Piaget 1953; 1962; for an account of action-oriented interaction.)

This interaction-based approach to language evolution implicitly recognizes the role of a top-down mechanism. The scheme shown in Fig. 2 by Hurford suggests a bottom-up mechanism, in which it is the stimuli from the object that evokes the appropriate predicates, attributes, or actions, which may reflect the original affinity of the action and the object, or a habitual association. In the top-down mechanism, on the other hand, an action is applied to an unrelated object, as, for instance, when a child tries to grab whatever objects he or she may come across. In assimilating an object to an action, novel features of the objects may be discovered (Indurkhya 1998). For example, a child executing the action “grabbing and bringing to the mouth to suck” may notice that the object is visually interesting too. This, in turn, may lead to the related action “grabbing and bringing in front of the eyes to look at” (Piaget 1953).

Incidentally, it has been proposed that this top-down mechanism forms the corner stone of metaphors, because it determines the results of the interaction when a predicate is metaphorically applied to an object (Indurkhya 1992). It has even been suggested that the process of noticing perceptual similarities, and thereby generating perceptual metaphors, starts working when the predicates and objects are still in a fused state, and can actually underlie the discovery of relations and attributes which hastens their separation (Piaget 1962). In other words, it is in comparing two holistic situations, and finding that they have some features in common, that the independent status of an object or attribute originates. Notice that this account assumes an evolution of memory that is orthogonal to the language evolution in that the cognitive agent, in order to carry out the comparison, is able to memorize and recall holistic situations before the objects and attributes acquire an autonomous status. (This is in contrast with the view implicit in Hurford’s account, where the memory module presupposes object-attribute ontology.)

Finally, it is only at a much, much later stage that a cognitive agent gets to the point where an object is “inspected carefully in order to figure out what to do with it.” Notice that in “inspecting carefully,” all those actions that are habitually associated with the object may be activated via the bottom-up mechanism. At the same time, in trying to “figure out what to do with it,” actions motivated by the goals and the desires of the agent may be activated via the top-down mechanism. A neural account of any of these steps, if possible, would be quite fascinating.

To conclude, I feel that Hurford’s arguments to provide a neural grounding for the object-predicate distinction are a bit like speeding to one’s destination in such a hurry as to miss most of the interesting scenery along the way. However, it is the scenery along the way, some glimpses of which I have tried to provide in this commentary, which may well provide the most illuminating insights into the evolution of language and the neural correlates of its different stages.

　11. REFERENCES

Aitchison, J. (1996). The seeds of speech. Cambridge University Press.

Barrett, M. (1995). Early lexical development. In P. Fletcher & B. MacWhinney (Eds.), The handbook of child language. Oxford: Blackwell.

Bloom, L. (1973). One word at a time. The Hague: Morton.

Cassirer, E. (1955). The philosophy of symbolic forms, Vol. 1: Language. (Trans. by R. Manheim). Yale University Press.

Indurkhya, B. (1992). Metaphor and cognition. Dordrecht: Kluwer Academic Publishers.

Indurkhya, B. (1998). On creation of features and change of representation. Journal of Japanese Cognitive Science Society 5: 2, 43-56.

Langer, S.K. (1957). Philosophy in a new key. (3^rd ed.) Harvard University Press.

Piaget, J. (1953). The origin of intelligence in the child. (Trans. by M. Cook). Routledge & Kegan Paul. (Penguin: 1977).

Piaget, J. (1962). Play, dreams and imitation in childhood. (Trans. by C. Gattegno and F.M. Hodgson). W.W. Norton & Co.

Wegener, P. (1885). The life of speech. In P. Wegener and D.W. Abse, Speech and reason. University of Virginia Press: 1971.

Wray, A. (1998). Protolanguage as a holistic system for social interaction. Language & Communication 18, 47-67.

Wray, A. (2000). A protolanguage with no declaratives and no names. Proceedings of the 3^rd International Conference on the Evolution of Language. Paris, France.

Wray, A. and M.R. Perkins (2000). The functions of formulaic language: an integrated model. Language & Communication 20, 1-28.