The fabric of our metaphor is itself evolving. As in Quine, the vocables in our metaphor are related to experience, but the nature of the relationship has more to do with population dynamics than semantics. In fact most of what we call meaning is the result of exploiting behaviors, behaviors that are reinforced in a social setting, some of which have evolved into linguistic interactions, others have not. Our fabric evolves because of human interactions giving rise to linguistic behavior from non-linguistic behavior. But the emergence of a category such as linguistic behavior is reflected, later on in the history of language, in the ever-growing number of linguistic categories that emerge from strictly linguistic interactions. Moreover, we can assume that present linguistic categories are less related to non-linguistic behavior than linguistic categories were in, say, proto-languages, and in some cases not related at all. For example, in the case of logicalized connectives, such as or and and; the vocabulary has lexical roots, but the truth functional has emerged purely from the dynamics of linguistic interactions and the ability of linguistic transactions to coordinate human activity.
One could say that, at the periphery, the distinction between language and non-language is minimal.
Quine's model follows physics, unconstrained but consistent with interactions in the core with facts (or experience, in this case) defining the behavior of the edge. Quine describes a semantic field, the edge of which is supported in a certain way by experience. An incidental consequence of such an image is that it imposes an actual boundary (one that we may presume him to have implicitly acknowledged) between language and non-language, and so between linguistic and non-linguistic behavior. From this perceived constraint Quine defines semantic consequences that are derived from the interaction at this edge. Quine's notion of bounded field and resulting behaviors borrows indirectly from physics. To assume a boundary is to stipulate fixed conditions for some of the constituents of the system (those that constitute the boundary) that distinguishes them from those of the interior of the theoretical system. Through the defined interactions of the system, the behavior of the boundary constituents affect the behavior of the system as a whole. In Quine's metaphor, experience constitutes a dynamic but functionally fixed condition. The semantic content of sentences at the edge may change (because sentences may come and go), but they have a semantic status which is constant because of their constant relationship to experience.
The nature of the fabric (core, periphery, dependencies) is laid out. Elsewhere in the fabric and increasingly as we approach the centre, semantic status is also relatively unaltered, but, again, as we approach the centre, semantic content is also increasingly unchanging. If the vision of language that underlies the metaphor has sentences of mathematics at the centre, then one might say that the semantic content of the sentences at the centre is fixed because it is null or nearly so. It is relatively impervious to the influence of changes at the periphery, but not absolutely so. And again, changes at the periphery are relatively independent of one another. Changes nearer the centre more fundamentally disturb the structure of the fabric as a whole.
Quine's metaphor is, of course not a metaphor principally about language, but about human understanding as it expresses itself in language. The relative peripherality or centrality of positions within the fabric represent the status of sentences that are taken to be true, that, taken as a whole, constitute a view of how things are.
Our fabric is distinguishable from Quine's by virtue of the elements described in it; in our case they are positional elements referring to the relationship between vocable uses. Our fabric or field is not semantic as Quine would have it. Only comparatively and indirectly is it connected with the truth values of sentences. Our fabric registers positions of vocabulary, and therefore of sentences in which they occur, independently of whether those sentences are true or false. It records movements of vocabulary from uses reflecting the particularities of experience that give rise to them to the roles they retain when those particularities have long disappeared from the collective memory of the speakers of the language. Thus it records a loss of information. But it is not primarily synaptic either, though Deacon's account [18] of three kinds of meanings (iconic, indexical, and symbolic) could plausibly be mapped onto some such structure; iconic meanings being the elements at the periphery, then indexical meanings in intermediate position and symbolic meanings at the center. The notion that a progression in evolutionary time from an early stage of language development towards abstraction should be paralleled in developments of synaptic function in individuals is by no means incompatible with our account; however our point is a different one: the shared nature of language in a population shapes linguistic transactions in ways that are independent of the synaptic effects that occur in individuals involved in the transactions.
Our fabric also assigns a different understanding of centre and edge. In fact our fabric has no edge. Movement is from the centre, but what is recorded is attenuation. As we move from the centre, we encounter no boundary, only a gradual diminution of a semantic value, a motion registrable in the closed unit interval. At the centre that value is 1; but it drifts inexorably toward 0.
The fabric is woven from relationships between vocable uses and their diachronic account. This fabric grows with succeeding generations of users, adding and subtracting uses, but its internal constitution is shaped by the collective memory of the whole body of its users. In fact, we may say that the memory of the fabric is the collective memory of its users. What we mean is: Uses of elements of the language are determined partly by previous uses and partly by whatever initiates change. But for successive generations of users of the language, there is less and less contact with earlier uses. For elements near the centre, use is determined by relatively short-term memory; for those toward the periphery, likewise. But for elements at the centre, short-term memory has the ingredients for semantic understanding based in relatively non-linguistic material. In later stages, experience and relatively earlier uses of elements play a diminishing role in the regulation of new uses. This is mostly because the history of uses is slowly lost over time, often forgotten as successive generations inherit the vocabulary from previous generations. The loss is, however, relatively smooth and slow, since the changes, though discrete, are discrete on the scale of individual births and deaths, that is, well below the scale of observable general linguistic changes or trends. As we have previously mentioned, the attenuation of uses is statistically unidirectional, the direction being that toward the growth of the fabric. By contrast, in Quine's metaphor, re-evaluations of the truth-values of sentences can occur, although against varying degrees of resistance, anywhere throughout the fabric. Moreover, it is not clear that in the Quinean interconnectedness of meanings there is any concept of temporality, entropy, equilibrium that govern the area from the periphery to the center. On our account, elements acquire new uses as they are further away from the centre and in most cases the interconnectivity of uses generate unidirectional attenuation towards the virtual periphery. As elements drift away from the centre there is also a gradual loss of non-idiomatic compositionality. In functionalized uses of auxiliary verbs, for example, we do not consult their surviving lexical uses in constructing tensed forms of verbs. But, for example, by the time the verb have was given an auxiliary role, its non-auxiliary uses had already lost their capacity to occasion very specific neural responses independently of linguistic environment. Much the same could be said of the auxiliary do. A sentence composed from elements that have already moved toward the periphery cannot generate any specific inferential effects unless additional information from the periphery is added.
In fact the notion of periphery is ill-applied here. We illustrate the movement of vocable uses towards a horizon, with specific examples. We might better speak of an horizon, since beyond it, elements of language cease to have effects of a specifically linguistic nature. So for example at the present point in the development of English, wycg (horse), absent very strong contextual clues amounting virtually to definition, would have auditory effects, but is essentially foreign. Like much of the rest of the Anglo-Saxon lexicon, it has passed the periphery into complete disuse. Some of this vocabulary is superseded by other lexical vocabulary, but some has simply disappeared with the need for it. Other vocabulary approaches this horizon essentially through our ceasing to require a semantic grounding for its use. Some of this vocabulary approaches the horizon through functionalization. Auxiliary verbs (have, is, do) and some connective vocabulary (or, but, if) are examples of this. But still other vocabulary approaches it simply through a loss of connection with its more concretely grounded origins. Into this category falls much of the philosophically charged vocabulary, (intention, mind, good, right, truth, fact), but other vocabulary as well (as well, indeed, very, mere, even, just and so on.)
The dynamics are so gradual as to allow the system to be in quasi-equilibrium on timescales most reasonable for several generations of language users to interact successfully. Our claim is that movement of vocabulary toward a horizon is an inexorable feature of language-change. It is sufficiently demonstrated by the observation that, whatever the rate of change of a language, (and that itself may change) it must eventually be so altered that its practitioners can learn the earlier form only as a second language - Anglo-Saxon for example. Nevertheless, the change is a gradual one; every generation of users inherits a language in which nearly all stages of this process are represented. To be sure, some vocabulary may pass from innovation to extinction with few recognizable intervening stages. But for much of the vocabulary of a language, the process is a gradual one involving many degrees of semantic attenuation.