Comments Welcome
George W. Grace
University of Hawaii

Ethnolinguistic Notes

Series 4, Number 3

LINGUISTIC CHANGE

2. THE "ÉTAT DE LANGUE" OF ABSTAND LANGUAGES

The preceding Ethnolinguistic Note concluded that, since Ausbau languages (see Kloss 1967) are not natural, the succession of states of the Ausbau language called "English" (or of any other Ausbau language) can't teach us anything about natural processes of linguistic change. But now, what is the case with Abstand languages? To discuss this, I'll continue my hypothetical dialog with myself.

QUESTION: What is an Abstand language? Kloss tells us that it qualifies as a separate language by virtue of its linguistic distance from other languages, but just how is this distance measured?

ANSWER: As far as I'm aware, Kloss has not said much about this (although I should point out that I'm not really familiar with his work). However, although the designation ("Abstandsprache") is Kloss's, the concept has a long history in linguistics. That is, there has long been a general understanding that whether two varieties are (or belong to) different languages, rather than being just dialects of the same language, depends on the linguistic distance separating them. Probably the most-referred-to criterion is that of mutual intelligibility (to say that A and B are "mutually intelligible" means that A-speakers can understand B and B-speakers can understand A).(1) It has traditionally been said that varieties that are mutually intelligible belong to the same language. However, it is now generally acknowledged that intelligibility of one variety by speakers of another is a matter of degree, and there is no established way of measuring that degree.(2) In short, there is still no generally agreed-upon measure of linguistic distance between varieties.

QUESTION: For the moment, let's set aside these two problems and imagine that we do have a satisfactory way of measuring the linguistic distance between varieties and that we can agree on a particular critical point on the scale of linguistic distance--call it the "language limit" (to borrow a term from Dyen 1965)--such that any two varieties whose linguistic distance from each other is less than the language limit are defined as belonging to the same language. (For present purposes it doesn't matter whether the measure is based on intelligibility, on how many descriptive differences distinguish the two varieties, or on some other consideration).

Now supposing we have such a touchstone, how do we classify varieties into (Abstand) languages when there is more than one linguistic distance to be taken into account? For example, imagine three varieties, A, B, and C such that the distance between varieties A and B and that between B and C fall within the language limit, but that between A and C does not. How many languages are there, and which varieties belong to which language?

ANSWER: I think comparative linguists have generally agreed that in such a situation there's only one language because there is a chain of sufficiently-close (i.e., under the "language limit") links (i.e., the links A-B and B-C) connecting A to C.(3)

QUESTION: All right, I now have a fairly clear idea of what, in principle, an Abstand language is. In fact, the English language as an Abstand language--what we might refer to as "Abstand English"--clearly consists of such a chain: one with a very large number of links. Now we can return to our question, which is: How can we observe a change? What are the actual observations that attest to a linguistic change?

ANSWER: At one time there was the idea that "sound change" was too gradual to be observable at all, but that wasn't supposed to be necessarily true of other kinds of changes. In any case, however, with the work of Labov and others (e.g., the Milroys), it now seems that even ongoing sound change is observable. In fact, studies such as those by Labov and the Milroys finally provide a starting point for discussing the problem. They tell us a lot about what goes on--what kinds of things change in the actual utterances of speakers--when language change is in progress.

QUESTION: One question that remains from Labov's studies (e.g., Martha's Vineyard, New York City) or the Milroys' (several different Belfast neighborhoods) of ongoing changes is how one can tell when these changes have proceeded far enough that one can say that a change has been completed--that "the language" has changed.

ANSWER: By definition, the change would have been completed--the language would have changed--when one language state had been replaced by another. This can be interpreted as meaning that the description that would be made of the language by a linguist now would differ--in the feature representing the change--from the description that would have been made previously.

QUESTION: All right. Now a second question is: What is the language whose state is changing in Labov's and the Milroys' studies?

ANSWER: We're talking about Abstand languages here; the only possibility among Abstand languages is English.

QUESTION: Now we can return to the question (originally raised in the preceding Note) of how the evidence--that is, changes in the speech of native speakers--is interpreted by the linguist. The question is: what relationship do the features exhibited in the speech of particular individuals who are counted as being speakers of a particular language have to the état de langue of that language at that time--that is, to the description of the language that linguists might be expected to write?

Or, to return to the example of English, what kind of observations would be required to demonstrate that the English language had undergone a change: that one language state (état de langue) had been replaced by another? In fact, how do we ever decide what the state of an Abstand language like English is?

ANSWER: We must realize that English has an enormous amount of dialect diversity. It's hard to imagine what a description of its state that adequately took into account all of the usages of all of the speakers would look like. Although Hockett (1955: 18-22, and 1958: 332-37) did speak of description of the overall pattern of a language as a theoretical possibility, he didn't attempt to describe what such a description might look like. I also recall C-J. Bailey speaking of pan-dialectal grammars, but as far as I know, none was ever completed. I think the answer must be that there is really no very clear idea of what a linguistic description of a dialectally diverse Abstand language such as English would be like.(4)

QUESTION: Is English exceptional? Is the immense dialect diversity of this one language a special case? Or to put the question more generally: How much dialect diversity can there be in one language--how different can the dialects of a single language be?

ANSWER: Well, there's the rub. Consider again the definition we proposed for Abstand languages. It puts no limit on the length of the chain of mutually-intelligible dialects that may belong to the same language. It also puts no limit on the degree of difference between the most distant members of the chain. Therefore, theoretically it's possible to have two dialects of the same language that are as different from each other as are any two distinct languages in the world. All that's required is that the chain contain enough links to permit all of the differences to be added step by step.

QUESTION: It's easy to see that writing a single description for a language composed of such diverse dialects would be an imposing task, and that it wouldn't have anything like the self-consistent structure that we expect a language to have. In short, our definition of Abstand languages leads to what seem to be quite unsatisfactory results. It would sometimes classify varieties whose speakers couldn't understand one another at all into the same language, wouldn't it?

ANSWER: Yes.

QUESTION: But how can a language like that jibe with our other idea that a language is a single system, with a single grammar and vocabulary?

ANSWER: Not at all well as far as I can see.

QUESTION: Then why should we accept a definition of (Abstand) language that produces such awkward results?

ANSWER: Consider the alternatives. (Note by the way, that when we introduced the concept "language limit" above, we didn't specify what the criteria for the limit were. We can, if we want to, specify that the limit is the point beyond which different varieties can no longer be described as having a single grammar. Thus, by definition, any two varieties whose linguistic distance falls within the language limit have--i.e., are generated by--the same grammar, whereas any two whose linguistic distance falls outside that limit can't be generated by the same grammar).

Recall first the definition we've been using. We've been defining "a language" so that any two varieties whose linguistic distance falls within the language limit were to be assigned to the same language. But that definition led to the results that we're finding unacceptable.

The obvious alternative would be to define "a language" so that no two varieties whose linguistic distance does not fall within that limit can be assigned to the same language.

Let's consider this definition in relation to the hypothetical situation discussed above (that is, three varieties, A, B, and C such that the distance between A and C and that between B and C fall within the limit, whereas that between A and C does not. Our new definition has the consequence that A and C can't belong to the same language. But where do we draw the boundary: between A and B (thus giving us two languages, A and BC)? Between B and C (thus giving us AB and C)? Or both (giving us A and B and C)?

The last solution requires us to set up languages that are linguistically as close to one another as are the dialects of some other languages.(5) The first two solutions set up a language containing varieties that are no closer to each other than one of the varieties is to another language. And, of course, the choice between the first and second solution is completely arbitrary.

QUESTION: Doesn't all of this lead to the conclusion that Abstand languages are not entities that occur spontaneously in nature, but are artificial constructs set up for someone's (usually linguists') convenience?

ANSWER: Well, there are languages that consist entirely of varieties such that the distance between each member variety and each other member variety falls within the limit and that between any member and any non-member falls outside the limit. It might be argued that languages meeting these conditions are not artificial constructs.

QUESTION: But if we said that those are real languages while the others are artificial constructs, wouldn't that leave us in the position of saying that what some people speak are real languages and what others speak are not, and that there's no consistent linguistic difference between the two.

ANSWER: It would be more accurate to say that all languages are fictions; that they're the product of "as-if" reasoning.

The fiction of individual language systems was probably suggested by the development of standard languages (effectively, the Ausbausprachen that were the subject of the previous Note).

QUESTION: Are you saying that our strategy for understanding natural language is to try to see in it the properties of the standard languages that we have artificially designed?

ANSWER: Yes, except that none of the standard languages fully embody the principles that were considered to be ideal.

QUESTION: But isn't it a strange reversal of the natural relations to use an artificial creation such as a standard language as a model for trying to understand the nature of the natural phenomenon--ordinary human language--from which the standard language itself is ultimately derived?

ANSWER: Ironic in a way, perhaps, but not unusual. Humans' main strategy for understanding something new or previously not understood is by metaphor--by considering the new thing as if it were something that we do understand (or think we understand)--and what we often seem to understand best are precisely the things that we have ourselves invented.

And this strategy is often successful in large measure. For some purposes it works quite well to talk as if the reason why people understood each other is that there are natural languages largely analogous to the standard languages and that the people in question speak the same such language. And for some purposes it works well to talk as if the linguistic expressions they use in speaking were compositional, and as if these expressions were governed by a system of grammatical rules. And, finally, for some purposes it has proven quite fruitful to talk as if there really were self-consistent systems that evolve in a succession of discrete states.

However, although these fictions provide a convenient way of talking about many kinds of linguistic phenomena, they, like all fictions, can lead into serious difficulties if we allow ourselves to forget their fictive status--if we allow ourselves to reify them.

QUESTION: This seems to leave us in the awkward position of having to say (1) that we can't be specific about what kind of evidence from actual speakers of English would show that a linguistic change had occurred, and (2) that this is because we can't be specific about what the speech of actual speakers has to do at all with the état de langue of the language, and (3) that this is at least in part because the language itself is a fiction.(6)

ANSWER: I can't deny any of that. However, in defense of the existing strategy of linguistics, I should point out that just to be able to formulate questions that are at once precise enough and concrete enough to make it possible to design research that has a reasonable chance of answering them is a remarkable accomplishment. It isn't surprising that it may require resort to a variety of fictions.

QUESTION: We started out to answer the question: What is a linguistic change? We've said that a linguistic change was a change of language state, the replacement of one discrete language state by another. And we've seen that what was meant by a change of language state was in fact a change in the description that a linguist would write for the particular language. Then we saw that identifying the particular language wasn't as straightforward a task as might have been supposed, and wound up having to conclude that the individual languages that we speak of are actually not real entities at all.

Where does this leave us in our search for an answer to the question: What is a linguistic change? Are we to conclude that it's a meaningless question based on false premises?

ANSWER: Discrete languages are fictions, and so are discrete états de langue. However, human language exists and so does linguistic change. It's linguistic change that people like Labov and the Milroys have been studying. And so, in a different way, have the traditional historical linguists even when they operate with the fiction of états de langue.

However, if we're to avoid the dangers of reification, we must recognize that the scope of a change may not coincide with the boundaries of what we (either on the basis of Ausbau or Abstand criteria) have decided to call a single language. That is, it may not affect all of the utterances that would be said to be in that language (or all of the communities, or all of the speakers). In addition, it might simultaneously affect some or all relevant utterances attributed to another language.

We must also recognize that changes don't normally proceed by discrete steps.

NOTES

1. Another proposed measure of linguistic distance has been lexicostatistics, but it isn't generally accepted now as having much validity. Back up

2. There is another problem--there are cases of varieties A and B such that A-speakers can understand B significantly better than B-speakers can understand A. Back up

3. Given these assumptions, we may define a language (an Abstand language) as consisting of a set of linguistic varieties such that: (1) for each variety which is a member of the language there is at least one other member at a linguistic distance that falls within the language limit, while there is no non-member whose distance from any member falls within the language limit, and (2) any two varieties that are members of the language are connected to each other by a chain of members such that each such member is at a distance within the language limit to the next member in the chain. Back up

4. Incidentally, it would seem that an actual description of the overall pattern of a language would necessarily change (and therefore the état de langue would change) if one of the constituent varieties were to die out even though all of the surviving speakers continued to speak exactly as they had before. To require our theory of linguistic change to accommodate "changes" of this sort would hardly be an advantage. Back up

5. Another strange consequence of the third solution (that A, B, and C are all different languages) is that if C suddenly disappeared, A and B would, by the criteria that we are using, immediately be reinterpreted as dialects of the same language even though neither had undergone any change at all! And it would be a language whose immediate ancestors were distinct languages--something that is not supposed to happen. Back up

6. Actually, "the English language" is a confused fiction, since references to "English" rarely make clear whether the speaker has in mind the Ausbau or the Abstand version (or what). Back up