knife

Comments welcome
George W. Grace
University of Hawaii

Reflections on the evolution of human language:

3. A “Swiss-Army-Knife” Conception of Early Language Evolution*

1. Language as an Instrument of Reality Construction
I am proposing here that the role of language in communication first evolved as a side effect of its basic role in the construction of reality.
Harry J. Jerison (1976: 101)

What I want to propose here is that knowledge of language, as it evolved and even to a great extent as it exists today, consists in a number of somewhat independent systems. This might mean, for example, that each of the “intelligences” proposed by Steven Mithen (1996) for the early members of the genus Homo might have had its own dedicated communication system that was developing more or less independently of the others. For example, one can imagine his social, technical, and natural history intelligences--each reflecting growing specialized knowledge and each accompanied by specialized communicative resources to support this knowledge growth. The image suggested in my title is that of the blades of a Swiss army knife, each with its own entirely independent form and function. Although the image is striking, I should acknowledge at once that it exaggerates the degree of independence of the systems in question; they are certainly less independent in the language of today and have most likely been from the beginning.

Now for this proposal to make any sense requires a recognition of the role of language in constructing, maintaining, and disseminating the realities in which we effectively live. Therefore, what I’ll try to do here first is make clear what I mean by saying that language is an instrument for reality construction. Then I will explain what I mean by the awkward expression “ways of talking about things”, and finally I will present my scenario for the early evolution of language.

Although I would hesitate to say that either of the roles mentioned by Jerison was only a “side effect” of the other, the importance of the latter has certainly not received sufficient attention from linguistic theory. As I tried to show in my book The linguistic construction of reality (Grace 1987), ignoring the importance of this role creates serious obstacles to any attempt to understand how language actually works.

It is this capacity of language for reality construction that makes possible our enormously enriched knowledge of the world and how it works. In my view the primary selective advantage provided by most of the earliest steps toward language must have been its role as a system for recording and communicating knowledge. And the two go hand-in-hand: the accumulation of knowledge would have become a community affair once new information acquired by any individual could be stored in transmissible form. It could then be transmitted to others, already in storable form. I would suggest that the evolutionary process took off in earnest from the earliest stage in which an effective means to record and communicate new knowledge existed.

However, there are some qualifications that need to be added to the above picture of reality construction. First, it is important to remember that what we think of as “knowledge” of the world is something less exact than knowledge. The fact is that we think, act, and talk in terms of a model--or actually multiple models--of reality. I’m not suggesting that there is not an actual reality “out there” or that our modeling of it is not to a great extent informed by our experience in relating to it. But our access to this actual reality is limited.

To begin with, our perceptions are filtered by the limitations of our senses. Our ears are only sensitive to frequencies within a particular range, our eyes only to a narrow segment of the electromagnetic spectrum, and the information provided by each of the other senses is shaped by its own kind of filtering. Then there remains the further task of sensory integration--of combining the various streams of incoming sensory information in such a way as to construct a single snapshot. A snapshot that evolves in real time, that is.

But for such a snapshot to be of any use to us--for it to “make sense”--we must already have more general models in terms of which it can be interpreted. Its implications can only be drawn if we can relate it (or various parts of it) to other experiences--to other things we “know”. These general models have been constructed, and continue to be further developed and revised, in large part through the integration of such experiences--such sensory “snapshots”.

However, each individual’s reality models are further informed by what others have communicated to him/her of their own realities. And that is a particular advantage that language, and presumably its antecedents from a fairly early stage--afforded. It permits reality models to be shared among the speakers of a community, and this permits the innovations of each to be available to all. And over time of course, each useful innovation to the shared model provides a more sophisticated base from which the next innovation can emerge. Consequently from very early on such communication-rooted models must have rapidly become much richer than the pre-communicative reality models available to any individual. And any communication-rooted model that survived would presumably have been advantageous to the survival and reproductive success of those who adopted it.

Nowadays the communication of such reality models takes place in many ways. Of course, some of it is direct--people tell us “facts”. That is one of the roles of caregivers of children and one of the functions of formal education. But much more is communicated by what we can infer, in particular what we need to infer of people’s beliefs and assumptions if we are to make sense of what they say and do.

It is probably important to recognize how different these reality models are from what might accurately be described simply as “knowledge of the world”. For one conspicuous example, consider the realities (i.e., the personal models) that attribute important roles to beings that others regard as fictitious--e.g., unicorns, mermaids, or gods, goddesses, or other supernatural beings. In the end, the degree to which any reality model possessed by any individual corresponds to the reality “out there” simply can’t be determined exactly.

What is of interest to us here is that language is the principal means by which these realities are constructed, preserved, and remembered. But how does a communication system (how do its users) go about the task of constructing a reality? The answer is: by devising (and revising) ways to talk about it. Of course discussions of the nature of reality as a whole are extremely rare, and it must surely not have been a major topic of communication among our early ancestors. On the contrary, the reality construction must necessarily have proceeded piecemeal as our ancestors slowly developed ways to talk about one subject of interest after another and in greater and greater detail. (It’s convenient to call it talking, even though the use of this term may be somewhat premature at the stage we’re concerned with here).

2. Ways of Talking as Mini-Systems

But what is a way of talking (henceforth, WT)? Of course, the concept originally occurred to me in thinking about present-day language. No doubt a major factor in suggesting the concept was the experience of the “Chomskyan revolution”, which some have argued was a true scientific revolution in the Kuhnian sense. The contrast between the WT of the new generative grammarians and that of the structuralists (especially those of the American structuralist tradition), and the constant misunderstandings that this contrast led to, made quite an impression on me. Although each of these WTs talked about an object that they called “language”, the objects bearing this name were actually not the same--each had its own characteristics. That is what I mean by saying that the different WTs had constructed somewhat different realities.

And Kuhn’s work itself, especially the postscript to his second edition (Kuhn 1970), was also a factor. And certainly the works of Michel Foucault were a factor. I’ve always felt trepidations about citing his work, because I’m never sure whether what I take him to be saying is what he really meant, but I’m sure that reading his work, especially Foucault 1976, contributed to my concept. A WT certainly has much in common with a Foucaultian “discursive formation”. Furthermore, Whorf’s “fashions of speaking” (e.g., Carroll 1956: 108) should certainly be given some acknowledgement, although he doesn’t bring out what for me is the key point that a language comprises not one, but multiple, such mini-systems.

It’s true of course, that today’s language can hardly be assumed to be representative in all respects of its fledgling antecedents, but it must necessarily still serve as one main source of clues on earlier evolutionary stages--clues as to what it was that was evolving.

I must admit that the term “way of talking” is not very satisfactory, but I haven’t found a workable substitute. Aside from the questionable appropriateness of the word “talking” for the earlier evolutionary stages, the term also suffers from an inherent ambiguity. For example, we could say that everyone has his/her own individual way of talking defined by voice quality, mannerisms, and other aspects of his/her use of language. However, it is shared WTs that interest us here. There will always be recognizable similarities between any one individual’s WT and those of at least some others, and these shared similarities themselves constitute a WT.

Such shared WTs are presumably always a sign of some other similarities connecting those who share the WT. Moreover it is often possible to recognize particular WTs as characterizing the members of particular groups, and inversely, recognize group members by their characteristic WT (at least as it appears in the appropriate contexts). In fact, any time that we’re able to match different utterances (or written texts) by their form or content with authors either individually or by schools or traditions of any kind (or by ongoing discourses in the Foucaultian sense), we’re defining them, broadly, as a distinctive WT. But to say this is to suggest that a WT is simply what has been labeled a “style”, and of course linguistics has always acknowledged the existence of stylistic differences. At the same time, however, stylistic differences have been regarded by many linguists as nothing more than different ways of saying the same things and therefore quite peripheral to their understanding of how language works.

But there is much more than that to those WTs that concern us here. As far as reality construction is concerned, what is of particular interest is the association of some WTs with particular subject matter areas. In fact, the discussion in Grace 1987 was mainly concerned with what I called “ways of talking about things”. Andrew Pawley has used the term “subject matter code” to refer to a similar concept. In Pawley 1991 he offers an excellent description of the quite elaborate code devoted to the game of cricket.

The extent to which WTs permeate contemporary language is seldom recognized. It has always seemed to me that Condillac was quite ahead of his time (if not of ours!) to speak of scientific disciplines as “well made languages”. It’s seemed to me that learning a new discipline is as much a matter of learning its WT as of learning factual information. Anyone who doesn’t talk the talk doesn’t come across as a professional and anyone who does, does. It’s worth noting that Thomas Kuhn in his 1969 postscript to the second edition (Kuhn 1970) of his The Structure of scientific revolutions talks at some length about the communication problems associated with such revolutions. He points out that one of their characteristics is that the pre-revolutionaries cannot understand the WT of the revolutionaries: as he describes it, “they talk through each other” (Kuhn 1970: 109).

But all of the discussion of such WTs so far has been focused on today’s language, and today’s highly developed WTs are a far remove from the situation at the dawn of language. In fact, the examples from today’s language such as the cricket code and the discourse traditions of science might seem to suggest that such WTs are no more than secondary specializations within an already mature language. I would argue, to the contrary, that such narrowly focused systems were what constituted language (or its forerunners) in the early stages of its evolution. In fact, as early as Grace 1987 I proposed that the most significant development in the evolution of language(1) was the creation of new ways of talking developed for and around different kinds of subject matter.

Moreover, it’s important to emphasize that cricket and scientific fields are not the only parts of language that involve their own particular WTs: on the contrary, they permeate today’s language. But this isn’t the place to expand further on that point.(2)

Before going on, there is one more thing that I particularly need to emphasize: namely that WTs should not be dismissed as “just vocabulary”. Linguistics has often promoted a view of language that has a grammatical core as its essence, with vocabulary reduced to a kind of periphery. The vocabulary is seen as an assemblage of items that can freely come and go without disturbing the continuity of the grammatical essence.

In the last half century, in particular, there has been insistence on the intricacy of the syntactic systems as considered from a mathematical perspective. It is claimed that in order to speak a language, one must know and apply (in some unconscious way) this mathematical system. I think this is quite wrong, and quite misrepresents the way in which people know their languages (cf. Grace 1995, 1997, 1998). Unlike syntax, WTs haven’t been much studied; I’ll make only these very general suggestions about what all they consist of.

For one thing, much recent work has shown that much of linguistic performance that has been explained as the application of syntactic rules is actually accomplished by the application of stored knowledge. In other words, much of the knowledge used to produce and interpret the syntax of constructions is holistic, rather than analytic, knowledge. See for example, Wray’s (2002) heteromorphic distributed lexicon. Also consider various approaches to linguistics that are labeled “cognitive” such as construction grammar (cf. http://www.constructiongrammar.org/).

For another thing, a full description of any WT would no doubt have to include such things as frequencies of particular words and collocations, association of wordings with external situations, and the like. However, it would be a mistake for me to speculate further at this point.

The point is that such a conception of what language is throws a quite new light on how it can have evolved--on what stages it is most likely to have passed through. What I want to propose here is the outline of a scenario for the early stages of its evolution.

3. Mini-Systems in the Evolution of Language?

According to this scenario, language developed in the service of various rather specific functions. The idea is that (pre)-language did not start out as a single system but principally as a number of independent mini-systems representing separate strategies for dealing with particular kinds of subject matters and probably other functions (with functions related to social relationships probably prominent among them). These systems presumably depended at least in some measure on the growth of cognitive capacities that already distinguished our particular ancestors from those of related species. And presumably each of the systems contributed--some more and some less--to the factors favoring the expansion of such capacities. In fact, one can see here a kind of feedback mechanism.

Before going on, I should enter a couple of qualifications. First in my conception, what was involved was a gradual evolutionary process, and I don’t want to make any commitment about exactly what point in the process will ultimately be judged worthy of the designation “language”. Therefore, I don’t want to be specific about whether what was evolving in my scenario should be called “language” or something like “pre-language” or some of each.

Second, I don’t want to claim that there were multiple systems at the earliest stage in which our ancestors’ communicative abilities began to separate themselves from those of other species. All that would have been required to set the process off would have been some system that provided selective advantage to neurological and/or physiological capacities that could support more effective systems. At the moment, at least, I can see no reason why a mini-system consisting of nothing more than elaborated “contact calls” among the members of a grooming clique might not have provided the initial springboard for further advances (see Dunbar 1996, Grace 2003). Possibly the earliest stages would have accomplished nothing more than “vocal grooming”. However, as Dunbar has pointed out, as the group sizes grew, the ability of allies to share information about particular individuals (the “gossip” function) would have become very valuable. This suggests that selection would have come to favor capacities that permitted the handling of factual information--or to put it more accurately, the construction of mini-realities. At this point, as the hypothesis goes, the emerging capacities would have come to be exploited in multiple mini-systems for a variety of subject matters.

(1) The mini-systems (which are the earliest stages of what later can be called WTs) develop for more or less narrowly focused purposes. The systems at this stage very possibly involve all manner of signals--auditory, visual, perhaps tactile and even olfactory--in isolation or in different combinations in different systems.

(2) The systems naturally adapt in such a way as to exploit the existing cognitive (and other) capabilities.

(3) Some (at least) of the systems prove to afford survival advantages to their users. These advantages select for further advances in the innate capacities that underlie the systems; those capacities that offer advantages to more than one system tend to benefit particularly.

(4) The systems themselves are continuously adapting to exploit possibilities provided by this evolving genetic endowment.

(5) Since all of the systems are adapting to the same capacities, they tend to make similar adaptations. Such similar adaptations lead naturally toward systemic similarities--increased prominence of the vocal medium no doubt being among them.

(6) Eventually such convergence becomes sufficiently complete that a kind of overarching supersystem can be recognized--the forerunner of the languages that we know today.

In fact, it would seem that some such sequence as this must be required to account for the evolution--anatomical as well as neurological--that resulted in the human speech apparatus. This evolution would appear to have involved a gradual progress, each step of which must have afforded selective advantages that led to further steps. In fact, I picture the steadily increasing role of voice as a major component in the process of convergence.

It was Alison Wray (p.c.) who pointed out the similarity of the idea of language as constituted of multiple WTs to Steven Mithen’s (1996) proposal that early humans had several separate specialized kinds of intelligence. Mithen observes that they demonstrated quite advanced knowledge and skills in certain areas alongside a number of puzzling failures to combine knowledge from different areas to take advantage of what, from our perspective, seem to have been conspicuous opportunities. He proposes the existence of four “intelligences” in this period--social, technical, natural history, linguistic--and attributes these puzzling gaps to cognitive barriers between the different “intelligences”.

The notion of multiple intelligences has led me to see the connection with work by psychologists that suggest that the mind is not analogous to a general purpose computer, but rather that it has evolved around a number of different specific functions--some have even cited the blades of a Swiss army knife as an analogy. However, just what the specific “modules”, “cognitive domains”, “intelligences” are and just how much or how little they are free to interact are both matters of wide disagreement.

The fact that others are finding evidence that the mind is not monolithic in structure seems to lend additional plausibility to the hypothesis proposed here. However, I have no immediate suggestions about how many mini-systems were involved or what they might have been. As to their interaction, what I am suggesting is that each was responding to a somewhat different demand, and that the biological capacity that underlay each such system evolved in such a way as to heighten the particular kind of survival advantages that that particular system afforded.

However, it also seems likely that mutations that benefited one mini-system sometimes must have proved useful in others, and that consequently the systems themselves tended to adapt so as to exploit the new biological capacities. It also seems likely that if, as I’m proposing, adaptation to the same capacities led to a convergence of the systems themselves, it would be reasonable to expect that as the systems became more commensurable, the obstacles to communication between them would gradually have diminished.

This would suggest that, however isolated from each other (“encapsulated”) the mini-systems might have been at the beginning, they would probably have tended to become steadily less so. Anyway, whatever isolation there was at the outset has been overcome sufficiently that people today find acceptable a constructed reality in which languages--and ultimately “natural” languages--are represented as unified systems.

NOTES

* My thanks to Alison Wray for a number of helpful suggestions.

1. To be completely forthcoming, what I said was that it was most significant kind of development in “the entire period subsequent to the first appearance of full-fledged language” (Grace 1987: 97). However, I was making some assumptions then about when and how full-fledged language appeared that I would no longer accept--in fact, when and how it appeared is precisely the problem I’m concerned with here.

2. One thing that should be pointed out is that these WTs are by their nature not very precisely defined. They are typical prototype-defined categories. Although prototypical utterances of such a WT may be clearly recognizable, there is likely to be no clear boundary between less-prototypical members and non-members.

Of course, the number of WTs that could potentially be distinguished is very large indeed. Which ones we choose to recognize is a matter of convenience. And of course there are hierarchies: more narrowly defined ones that are subsets of more general ones.

Furthermore, there can be a hierarchy of subject matters that can be attributed to a particular WT. From one perspective several different WTs might be said to represent different approaches to a common subject matter, whereas from another it would be said that the different WTs each create their own, different, subject matters. For example, different schools of linguistics are generally characterized by more or less different WTs. However, for some purposes it might be convenient to say that the subject of all is the same--i.e., “language”. On the other hand, there might equally well be purposes for which it is convenient to say they are talking about different subjects--more or less distinctive (e.g., more or less inclusive) kinds of thing that each calls “language”. In short, we may identify some WTs that are more general than others, and a more general one may include numerous more specific ones. Moreover, the boundaries between WTs for neighboring (whatever that may mean) subject matters are rarely if ever very sharp.

Such WTs characteristically depend heavily on metaphor. In fact, most of what is described as "literal" is just no-longer-live metaphor. A way of talking for a new subject seems to require a root metaphor. "A root metaphor represents one structured domain of experience (usually a more abstract one) as being like another (usually a more concrete one."(Grace 1987: 98). (The term “domain of experience” was taken from Lakoff and Johnson 1980). Not surprisingly, then, WTs often represent unstated assumptions about their subject matter.

Since the same root metaphors--encoding the same assumptions--may be used in different languages, the same WTs may be identified as existing simultaneously in such languages. This is exemplified by the discourses of particular sciences, or even science as a whole. A major part of the challenge of introducing Western science into Asian and other non-Western countries involved something close to a calquing of the WTs of their underlying discourses. Other prominent examples of WTs shared by many languages can be found in religious traditions such as that represented by the Roman Catholic Church and in international sports such as soccer or baseball.

It is important to note that any WT about a particular subject matter accounts for only a limited domain within an individual’s overall way of talking. Individual A might well use a recognizably “same” WT as B when discussing a professional topic--a WT that might even be entirely absent from C’s repertoire. But at the same time A and C, and not B, might use a recognizably same WT for a different subject matter area such as sports.