Comments Welcome
George W. Grace
University of Hawaii

Ethnolinguistic Notes

Series 4, Number 7

LINGUISTIC CHANGE

6. THE INDIVIDUAL'S KNOWLEDGE AND THE TRADITIONAL NOTION OF LANGUAGES

QUESTION: In the last Note (Grace 1997) you argued that an individual's knowledge of language (KOL) takes the form of a large memory store of experiences in which language was used. All right, supposing that that is the form of the speaker's knowledge of language, what does it tell us about how linguistic change occurs?

ANSWER: It tells us that the KOL of each speaker is undergoing modification almost constantly, since presumably there's some slight effect from each encounter the individual has with language in use.

QUESTION: This seems to be taking us more and more off the track. If you remember, this whole discussion started out to make my feeling that some of the Austronesian languages had developed in unexpected ways--ways that didn't conform to our assumptions about how languages change and differentiate--more specific. That led us to this attempt to pin down just what linguistic change is. But now you're saying that the real locus of change is in the individual, that languages are fictions, and that change in the individual KOL is almost continuous.

Does this mean that my original question is meaningless, or is this discussion going to lead eventually to some progress toward finding an answer?

ANSWER: Well, the reason that we've gotten to this point is that we've been trying to figure out what actually goes on. The idea is that the better we understand what actually goes on, the better our chance of specifying what facts lie behind such things as your perception that some Austronesian languages have developed in unexpected ways.

QUESTION: We continue to talk about Austronesian "languages" even though we've agreed that languages are fictions. However, you promised earlier to attempt to give a better account of what lies behind this fiction. Isn't it time to do that?

ANSWER: I think it is, and I'll make the attempt now.

To begin with, the KOLs of different individuals may be more or less alike. Different individuals who have grown up in a sufficiently small and homogeneous community may actually have observed many of the same utterances--i.e., been present when the utterance took place. Therefore, an utterance made by one of these individuals--one derived from his/her memory store--may be readily recognized by another of the individuals as (employing a linguistic expression that is) part of his/her own store. It's customary to say that the latter individual "understands" the former (or the former's utterance), and to describe individuals who understand one another's utterances as speaking "the same language".

And then, of course, there would be other such communities sufficiently different from the first that the individuals who had grown up in them would be said to speak an entirely different language.

QUESTION: In sum, if the KOLs of two individuals are sufficiently similar, the two will be said to speak the same language, and if they're sufficiently dissimilar, they'll be said to speak different languages.

ANSWER: That's right.

QUESTION: And as an illustration you described the kind of similarity that might be found in the KOLs of individuals whose memory stores overlapped because they'd been present to hear some of the same utterances. But what accounts for the similarity of the KOLs of individuals who haven't been present on any of the same occasions?

ANSWER: Well, the KOLs of different individuals who have the same role models will be similar, even though those individuals' paths have never crossed. (Recall that we mentioned in the preceding Note [Grace 1997] that we regard some people as being more suitable role models for ourselves than are others, and that our memory stores will tend to give particular prominence to situations in which these role models found themselves and especially to their utterances in these situations.)(1)

Therefore, to the extent that there is cohesion in a community, to the extent that everyone shares the same role models, there will be an approximation to the ideal conception of a single language. That is, the utterances in the people's memory stores will be so similar that it will be approximately as if their speech were governed by a common set of rules.

QUESTION: You imply that the "ideal" conception of a single language is one in which everyone's speech is "governed by" a common set of rules. I take it that what you have in mind is what you call (for example, in the "Side Comment" to Grace 1997) the "governing system" conception of languages?

ANSWER: Yes. I'm thinking of that as the ideal language for linguistics in general (although the ideal language for historical linguistics specifically would probably emphasize somewhat different properties).

QUESTION: Anyway, you're saying that there could be quite large communities that are very linguistically homogeneous (which is to say that all of their KOLs are very similar) as long as everyone in them has accepted and had access to the same role models?

ANSWER: Yes.

QUESTION: But wouldn't such homogeneity in a large community be exceptional, nonetheless?

ANSWER: Very much so. You may recall that in the preceding Note (Grace 1997) we briefly mentioned the fact that the experiences in an individual's memory store generally may be expected to involve a considerable variety of different speakers with the language learner exposed to a range of speech variation. In such a case I assume that the individual comes to perceive some of the variation in speech as emblematic of social roles and identifies some speakers--and specifically their speech--as better modeling the roles to which she/he ultimately aspires.

QUESTION: And so different members of the community wind up with different role models?

ANSWER: Yes.

QUESTION: Does that imply multilingualism?

ANSWER: A multilingual (or multidialectal) society is one kind of case of a society with significant linguistic variation. If the role models sort out into distinct sets so that one set is associated with one social identity and the next with another and so on, then there'll be that many clearly defined sets of examples. In such a case, it should be easy for a linguist to classify the linguistic expressions uttered in the community into "languages" or "dialects". In fact, if the question were properly presented to them (and very likely without it's being so presented), the people themselves would probably be able spontaneously to identify those languages/dialects.

QUESTION: I believe that what you just said contains an implied answer (or the beginning of an answer) to the question that we'll eventually have to come to, namely, what's the relationship between the KOLs and role models that we're talking about here and the "languages" that actually figure in linguistic discourse (such as "the Austronesian languages" that we've alluded to from the start). Or maybe I should say you imply two answers: (1) that these languages (and dialects) may be entities that linguists create by the classification of linguistic expressions, and (2) that they may be already-existing categories in the sociocultural worldview of the people themselves.

ANSWER: You're right; at least that's the starting point for my answer.

QUESTION: We need to explore that answer in more detail, of course, but before we go on with that I have a question about another kind of case where not everyone shares the same role models. Sometimes what is generally agreed to be the same language is spoken in two or more communities--say, in different villages. Now, suppose a particular language is spoken in two villages which are far enough apart that language learners in one couldn't have access to any of the same role models as those in the other. How could the utterances in the two (and therefore, the KOLs of their speakers) remain enough alike for them to count as the same language?

ANSWER: I think the answer must be that they couldn't remain so for very long under the conditions you describe (since, as we noted above, KOLs are constantly changing). However, if there were frequent trade or other contacts between the two villages, the role models from the two villages (i.e., those who serve as role models for the language learners of the respective villages) might meet each other frequently and might serve as role models to each other. In that way, a common focus might be maintained for the speech of the two villages.

QUESTION: Can two role models serve as role models for each other?

ANSWER: Yes, language-learning continues throughout our lives (even if not at the same pace as in the early years), and--especially in our later years--we don't model our behavior exclusively on those who are senior to us or whom we regard as otherwise superior.

QUESTION: So in the scenario that you're proposing here, there are people in each community who serve as role models for the members of that community, and that includes serving as role models for each other. However, these role models of each community--call them the "elders"--are in frequent contact with the elders of the other community and the combined group of elders in effect serve as role models for one another. Therefore, the role models for each community effectively share the same set of role models. Is that right?

ANSWER: Yes.

QUESTION: However, it seems to me that the role models from the two villages would have to have an awful lot of contact with each other to prevent the KOLs of the villages from growing apart.

ANSWER: I'd agree. It's hard to imagine in the situation we're describing that enough contact could be maintained over a long period of time to prevent the gradual differentiation of two foci. However, the kind of contact we've described could certainly slow the rate of differentiation far beyond what would have occurred if there hadn't been any contact.

QUESTION: When you speak of the differentiation of two "foci", do you mean different sets of role models who in turn have different role models for themselves?

ANSWER: Yes, or maybe who have partially different role models.

QUESTION: Explain.

ANSWER: We can imagine a stage in the differentiation process where the elders of the two communities each in principle accept all of the combined set of elders as role models, but in practice have less contact with those of the other community. Therefore they are (i.e., their KOLs are) in fact less influenced by the elders of the other community than by those of their own.

Another possible reason for partially different role models would be selectivity, because a person's status as role model may be limited in particular ways. That is, one can be a model for some things and not for others. For example, people with specialized knowledge can serve as models for the area of their specialization and perhaps for nothing else, or there may be sex-marked vocabulary or stylistic differences which still don't prevent an elder of the opposite sex from serving as a model for a great deal of behavior.

Thus, the elders of the other community might have their status as role models limited in some particular ways--for example, there might be some conception of special competences or characteristics possessed only by the members of one community or the other, or there might be features that came to be recognized as emblematic of one or the other community.

QUESTION: You spoke of the differentiation of two foci, but couldn't there just as well be more than two?

ANSWER: Yes, of course, there could be. In fact, there are sometimes long chains of communities where the speech of each is intermediate between the speech of its neighbors on either side. Of course, there have been no analyses of such chains in terms of sharing of role models (at least as far as I know), but they suggest a situation where there is a partial sharing of role models between communities a and b, between b and c, between c and d, and so on.

QUESTION: I want to get back to discussion of the question of how we get from individual KOLs to languages, but before doing so I want to digress long enough to ask you why you haven't said anything about standards of correctness recognized in the community.

ANSWER: For one thing, the nature (and, apparently, even the existence) of such standards varies widely from one society to another.

However, I do believe that a society's whole view of the nature of language--its ideology of language--can have a profound effect from the earliest stages on how the language learner perceives and goes about his/her task. In Modern culture, to take the most ready-to-hand example, the ideology of language is all-pervasive. The behavior of caregivers toward infants is very much shaped by it from the beginning. Caregivers have very definite ideas about what's involved in language learning, about what there is to be learned, and they offer examples and reward behaviors in accordance with those ideas.

But precisely because most of us (i.e., of those most likely to read this) have so thoroughly assimilated this particular ideology into our thinking, it's especially important for us to guard against any tendency to assume that it's universal or natural rather than cultural.

I'm trying to proceed as if the individual formed and maintained his/her KOL in the absence of any cultural assumptions about how language is learned (or taught).

QUESTION: But it isn't an accurate assumption?

ANSWER: No.

QUESTION: Then how do you justify it?

ANSWER: The problem is that we're nowhere near having enough information to pursue what might be the best alternative. That would be to analyze the beliefs in all human societies about language and how it's learned and attempt to design some kind of common denominator--a hypothetical cultural environment that would be broadly representative of the whole human experience.

However, the distortion produced by assuming the absence of any applicable cultural assumptions may not be as serious as one might be inclined to expect.

QUESTION: Why not?

ANSWER: Because I think most societies' ideologies of language are much less elaborate--and therefore, much less of an intrusion in the language learning process--than those of Modern (Western) culture.

QUESTION: All right, let's get back to the question of how we get from individual KOLs to languages because we need to be able to talk in terms of languages, even though we know them to be fictions. What I understood you to say above was that there are two main reasons for something to be recognized as a "language" in linguistic discourse: that some are recognized because they are already categories in the sociocultural worldview of the peoples who speak them, and that others are in effect the result of the classification of linguistic expressions by linguists.

ANSWER: That's an overly simplistic statement, but with that reservation, it's essentially what I intended.

QUESTION: Let me ask first about "languages" that are recognized on the basis of the fact that the people who speak them recognize them. Are we justified in taking "languages" defined by such non-linguistic criteria seriously for scientific purposes?

ANSWER: That's a hard question to answer. The answer shouldn't just be an automatic "no" because whether the people of different communities believe that their speech represents the same or different languages isn't irrelevant. It can be a significant factor in the acceptability of members of the other community as role models and therefore of the subsequent evolution of their speech over time.

However, such languages often don't conform to the assumptions of historical linguistics very well.

QUESTION: Could linguists improve the situation by disregarding the popularly-recognized "languages" and, on scientific grounds, defining our own set of languages--e.g., drawing our own language boundaries--for scientific purposes?

ANSWER: In some cases, that might help. However, it wouldn't solve all of the problems. First, as we noted above, whether or not people think their speech represents the same language as that of their neighbors may well be a factor in its future evolution.

But aside from that, there's no way to divide everything--all of human language--up into units that satisfy the requirements of historical linguistics.

QUESTION: What requirements do you mean?

ANSWER: I've discussed the main requirements somewhat more fully in Grace 1987a, but I have in mind mainly the following two:

(1) that the language be sufficiently insulated from all other languages to preclude its merging with any of them--that is, that it be an entity that will continue to evolve independently (forever, unless it ceases to be spoken), and

(2) that it will behave diachronically as a unit.

QUESTION: Explain what you mean by saying that the language needs to behave diachronically as a unit.

ANSWER: English is an example of a language that includes such a range of diversity that it doesn't behave diachronically as a unit. For instance, different dialects of English have undergone different sound changes; the pair "fox/vixen" is a familiar example. Or consider the studies of sound change in progress by Labov, the Milroys, and others. All of these reveal changes in the speech of local communities that are different from what's going on in the rest of English.

If a change only affects, say, the English of Martha's Vineyard, then English as a whole has not behaved as a unit. The diachronic unit has been the English of Martha's Vineyard.

QUESTION: Explain why that's important.

ANSWER: Since historical linguistics seeks the explanation for changes in the antecedent state of the language that has changed, then the preceding state in which the explanation must be sought has to be that of the English of Martha's Vineyard. In other words, the relevant "language"--the language that meets requirement (2)--is the English of Martha's Vineyard.

However, the English of Martha's Vineyard doesn't meet requirement (1).

QUESTION: Explain.

ANSWER: I mean that it isn't sufficiently insulated from all other languages (in particular in this case, from all of the rest of English) to preclude its merging with any of them in the future. Where "languages" are so similar that there's a lot of overlap in the KOLs of their speakers (especially if some automatic adjustments for regular sound correspondences are allowed), they can easily influence one another. Also, it's easy to imagine that if political and cultural conditions change in the future, such neighboring "languages" could merge.

QUESTION: Would you say that in order to behave diachronically as a unit, a language must consist of highly similar KOLs?

ANSWER: That would presumably be a principal requirement, except that maybe we ought to make some allowance for specialization based on complementary roles within a single society.

QUESTION: As, for example, a difference between men's and women's speech?

ANSWER: Yes, or other differences by social role.

QUESTION: All right. Now when you say that a language is supposed to be sufficiently insulated from all other languages to preclude its merging with any of them, what do you mean by "insulated"?

ANSWER: Well, I think historical linguistics has traditionally made the assumption that languages that are sufficiently different can't merge--that difference alone, if there's enough of it, is a guarantee against any future merger. I don't think that's true, but I can't think of any other way to state the requirement. In any case, languages that are very different are very unlikely to merge, so the fact that what's being presupposed isn't entirely true will probably have no unfortunate consequences most of the time.

QUESTION: And we can interpret "languages that are sufficiently different" to mean that the KOLs of which they consist must be significantly different?

ANSWER: Yes.

QUESTION: Let me try to summarize.

As we said before, languages are fictions.

However, there are a large number of such fictive entities that we--linguists and the general public in the modern world--recognize (i.e., refer to as languages).

There are certain characteristics that we typically attribute to a language (although these differ somewhat as between descriptive/theoretical linguists and historical linguists, and as between linguists and the general public). For historical linguists in particular, the main characteristics in terms of KOLs are (1) likeness of the KOLs of any two individuals who count as speakers of the language and (2) a significant difference between the KOL of any speaker and that of any non-speaker.

Did I get anything wrong?

ANSWER: I don't think so.

QUESTION: Now, to what extent do the "languages" that historical linguists actually work with meet these requirements?

ANSWER: Well, there are many linguistic situations in the world where it would be impossible to define languages that meet the desired criteria. In such cases, the "languages" with which we work (i.e., whose states our linguistic descriptions describe) are necessarily more or less arbitrarily defined.

But on the other hand, as best I can judge many of the languages that historical linguistics works with (e.g., many Austronesian languages) do seem to offer a fairly good approximation to the desired conditions, or at least that seems to be so as far as one can judge from the synchronic situation. Of course, there's no way (at least that I can think of) to be certain that the same conditions have existed for the centuries and often millennia with which we're concerned.

However, the demonstrated success of historical linguistics is itself probably the best reassurance we'll have in the foreseeable future that these conditions are approximated to a sufficient degree with sufficient frequency to make the enterprise worthwhile.

QUESTION: So, in your opinion the concept of languages is a close enough approximation to truth to justify continuing to use it?

ANSWER: Yes, despite the fact that it can't be made to work as well in some cases as in others.

QUESTION: Maybe you should say a little more about what kinds of cases it doesn't work well in.

ANSWER: Well, there are dialect chains where the nearest thing to a locus of homogeneity is very small--a village, an island--but where none of these units represents anything like an independent tradition.

Then, there are cases like contemporary English. It also has some of the properties of a chain.

Also, there are cases of what Le Page and Tabouret-Keller (1985) call "diffusion", where there seems to be no locus of homogeneity (see also Gardner 1966 and Scollon and Scollon 1979).

QUESTION: If such cases are frequent enough, wouldn't they seem to raise a serious question about the whole enterprise of historical linguistics?

ANSWER: The point is that historical linguistics has had conspicuous success comparing linguistic descriptions, that there's the promise of much more success, and that we have no equally productive alternative.

QUESTION: Then the upshot of all of this discussion is that although we've been operating with some fundamentally incorrect assumptions, there's still nothing that should be changed?

ANSWER: Right now I can't see any likelihood that we'll find anything to replace linguistic descriptions as the basic data any time soon. However, I think we should look forward to some fundamental changes in our explanations. We need eventually to re-think our ideas about the regularity of sound change, for example, and genetic relationship. To what extent are sound changes really regular, and why? In what ways and in what measure is the fundamental assumption of historical linguistics that languages spoken today are the continuations of particular languages spoken in the past actually valid?

More immediately, we need to accept a much greater separation than we currently assume to exist between changes in linguistic descriptions and the processes of which those changes are consequences. Changes in a linguistic description are incidental by-products of the actual changes--the categories of the description don't participate directly in any way.

QUESTION: Then, how do the actual changes come about?

ANSWER: That's to be the subject of the next Note.

NOTE

(1) There's another point that is needed to complete the picture here. We've said that when we're in the process of deciding on an utterance, what we have to do is search in our memory of previous experiences and find one or more situations that have enough similarity to the present one to furnish a basis for designing our response.

What I avoided going into in this and the previous Note is that the situations in question must be conceived of as what I've called "subjective situations". I've referred to the notion of subjective situation in Grace 1987b, 1989a, and 1989b.

Although very much information about a speaker's subjective situation can usually be inferred from the external evidence (the "objective situation"), subjective situations are ultimately states of mind of the speaker. Therefore, a certain amount of empathy is necessary for arriving at an approximate understanding of the subjective situation that lies behind any particular utterance. And such understandings are never anything more than approximate.

And that brings me to the point that needs to be made here. It's that a person's perception of who s/he is--his/her self-image--is an important facet of whatever subjective situation s/he finds her/himself in. Consequently, in order for one to be able to perceive oneself as being in the same situation as the utterer of some previous utterance, one must be able to identify (to that extent) with that utterer.

The way I've been talking about this is to say that our memory stores give particular "prominence" to the utterances of the individuals with whom we identify (and I should no doubt add something like, "or aspire to identify"). I'm not very sure whether "giving prominence" to these utterances will turn out to be the best way of talking about it, but for the moment I can't think of anything better (probably because my understanding of how the whole thing works is so murky). (Back up)