What is African-American English? (Part III: Trees)

I’m going to start a bit afield of the topic suggested by the title today, with trees. I don’t remember which I saw first in school, a tree depicting the history of life, or a tree depicting the history of the so-called Indo-European languages. But I remember being fascinated by both. I remember being fascinated by how such a simple and elegant image could capture so much.

The fascination with the tree has been felt by many, including, apparently, Charles Darwin. In On the Origin of Species, Darwin conjectured that a strictly diverging tree model represented “natural genealogies” and a “plan of creation.” The patterns of the tree were driven by laws akin to the “fixed law of gravity,” and one was free to imagine, as many did, that these laws were in fact written by the hand of God.

The arboreal glorification perhaps never died down and in any case was alive again in the late 20th century. As Richard Dawkins argued in The Blind Watchmaker, “it follows from the idea of evolution that there is one uniquely correct branching family tree of all living things.”

At best, Dawkins’ statement is misleading. The idea of evolution, in involving variation and natural selection, does not in itself require a tree, nor does it necessarily suggest a tree, unless one ignores the possibility of phenotypic variation arising by means other than mutation, as it indeed can, for instance, through sexual reproduction. And even without sex, and without cellular division, bacteria can share short passages of genetic material with each other. Because of this, the history of bacterial life is like a complex web, and it is often not clear how to classify individual bacteria into species [1]. And, even in the animal kingdom, so-called hybrid species exist, and there is even even an insect whose genome is apparently part-fungus in origin.

Because of facts like these – to say nothing of the quandaries for classification that biotechnology may present – while one is free to believe that there exists a “plan of creation” or “one uniquely correct branching family tree of all living things,” we will never have any certainty in knowing what this uniquely correct structure is, unless by proclaiming – tacitly or explicitly – to hold some supernatural power of ascertaining how nature is classified from its own (or from God’s) perspective.

To be fair, despite the rhetoric in Origin of Species which would seem to advance the idea that absolute truth is accessible to a scientist who can thereby distill it into a diagram, Darwin also appears to have appreciated some of the quandaries nature presents to objective classification, as demonstrated by the following passage:

In short, we shall have to treat species in the same manner as those naturalists treat genera, who admit that genera are merely artificial combinations made for convenience. This may not be a cheering prospect; but we shall at least be freed from the vain search for the undiscovered and undiscoverable essence of the term species.

There are a number of other tensions in the book, but I would like to shift to one in The Descent of Man, shifting also the focus from biological tree models to linguistic trees. First, Darwin says:

Languages, like organic beings, can be classed in groups under groups; and they can be classed either naturally according to descent, or artificially by other characters.

Then, a few moments later, he mentions:

Distinct languages may be crossed or blended together.

If languages can be crossed or blended together, as they indeed can, then languages are not necessarily going to fit “naturally” into a tree diagram without some decisions being made. For instance, we may classify English as “Germanic,” but if somebody wanted to emphasize the preponderance of contemporary English vocabulary which derived from Latin and French, one might instead classify English as “Romantic.” There is no purely objective way to decide between these two options, and in any case, the classification chosen will emphasize one aspect of history while ignoring others.

Of course, none of this means that Darwin’s two statements above amount to a paradox. Indeed, languages, like organic beings, can be fixed to a strictly hierarchical scheme. Moreover, it is possible to use knowledge of history to make decisions regarding classification, and this may have been all that Darwin meant by his use of the word “natural.” To clarify, both “Germanic” and “Romantic” would be natural classifications for English, whereas “Subject-Verb-Object,” would be artificial, based on characteristics taken as independent from history.

None of this would imply that a tree diagram necessarily amounts to anything more than a convenience, though as a convenience, the model is quite efficient for certain purposes. In the real world, trees are in fact all over the place, in governmental structures, corporate structures, course catalogs, old-fashioned card-catalogues, etc. But then if a network were not an efficient model for certain purposes, why Facebook, why the hypertext in Wikipedia, why roads?

It is thus no surprise that some linguists, such as Bill Labov, have used both trees and networks. Still, the specialness of the tree has been reflected, it seems, by the specialized usage of the word “genetic” among linguists. That is, for a linguist, “genetic” means “phylogenetic,” conforming to the tree model.

This practice could simply be terminological, one might imagine. However, it appears to have at least colored theoretical discussions in intimating that phenomena within languages (or dialects, etc.) are somehow of a different nature from those which transcend the boundaries.

In this light, it is to Labov’s credit that in a recent paper [2] he has made the distinction explicit with reference to the real, social world by proposing that the tree model corresponds to parent-to-child learning, specifically, by a process he calls transmission. Transmission takes place within “speech communities” and is contrasted with diffusion, which takes place among adults between communities.

Before turning to evidence, the logical consequences of Labov’s definitions should be examined. In particular, it should be noted that if all you are dealing with is what can happen within a speech community, and presumably within a dialect or language, then the result necessarily fits the family tree. A dialect or language, in other words, cannot converge into itself; it is already a unity. Therefore, it does not appear as though Labov has set up an empirical test of the connection of the tree model to different types of language learning, given his definitions.

However, Labov takes the tree model as representing also a gradualness of change, and it is this aspect he is most concerned with. In this regard, Labov proposes a process of “incrementation,” in which successive generations of children continue a change in the same direction. For my purposes, it doesn’t matter whether or not he’s on to something here, although it would seem to me that incrementation should be thought of first as a result, not a process.[3] Regardless, my concern is not with graduality but rather with the tree model as a hierarchy, so I will leave this aside.

As it stands, Labov provides no evidence that the applicability of a hierarchy has anything to do with the differences between child and adult learning. If adults are disrupting the hierarchy and kids aren’t, the simplest explanation (with regard to his discussion) would be that adults are more capable of getting in cars, boats, and so on, and are thereby more capable of transcending the boundaries of their native speech communities.[4]

In short, there is no evidence that the tree model represents the work of an underlying natural process.[5]

Now, to find our way back to Black English by way of natural groups, I would like to address the following question: Why, since the 19th century, have scholars repeatedly naturalized the tree diagram? The possible motivations are myriad, so I’ll touch on just a few here.

First, there is the actual success of the tree diagram with respect to the linguistic and biological record. However, it should be noted that the success of the tree diagram does not depend on there actually being a corresponding tree inscribed in natural processes. Its success depends on its ability to help researchers organize their data to construct hypotheses, and how reasonable these hypotheses are depends in part on historical knowledge.

A second possible motivation is that a naturalization of the tree model expresses the belief that there is a way to objectively trace back through history by algorithm, by laws akin to the “fixed law of gravity.” This might connect to a more general desire for certainty, among other things. In any case, a thorough analysis of this belief, or of a notion of elegance, for instance, could be a rather complex endeavor, so I’ll set such possibilities aside.

Perhaps more straightforward is a notion that the tree is familiar. In one sense, a tree is like a real-life tree, as they have sometimes been drawn.[6] In another sense, and for the final possibility I’ll explore, the family tree model itself was familiar when Darwin published Origin of Species, found commonly in the backs of Bibles, among other places.

What does it take, then, to make a family tree in genealogy? First, let us start with a small snapshot of the human pedigree, showing only sex, male or female:


Note that there is no purely objective way of organizing the members of this pedigree into two or more groups (i.e., beyond sex). Nevertheless, there is an easy way to organize them. Just give names to the oldest fathers, and have their children and their children’s mothers take the fathers names, and keep doing this (yes, somebody has thought of this before):


From this simple act of creating a patrilineal nomenclature, the “families” which have been created have no choice but to conform to a strictly hierarchical scheme (barring such possibilities as women bearing children with multiple fathers). That is, if a family has one son, the family continues, if it has two or more sons, the family divides, and if it has no sons, the family dies. In no case does a classification based on this naming system reflect convergences.

As the common practice of assigning family names yields distinct groups with tree-like histories, it is possible that people have been biased to think of families in this way. To take a step further, perhaps these notions associated with the family were extended to the concept of nation, and then they were carried from nation to language, and from language to dialect, etc., or something like that.[7]

This brings me back to the topic of African-American English, which I will pick up in a later post.

Meanwhile, thanks for reading...

notes

[1] See also Doolittle, 1999, “Phylogenetic classification and the universal tree,” in Science, 284, 2124-8, and Arnold, 2006, Evolution through Genetic Exchange.

[2] Labov, 2007, “Transmission and diffusion,” in Language, 83, 344-87.

[3] See also critical discussions of linearity and language change in Lass, 2006, “The end of linear narrative?” in Language & History: integrationist perspectives, ed. by Love; and Janda, 2001, “Beyond ‘pathways’ and ‘unidirectionality,’” in Language Sciences, 23, 265-340.

[4] Compare also Labov, 1966, The Social Stratification of English in NYC, p. 553.

[5] This not to suggest that the tree has no relationship to reality whatsoever. In any case, the tree, since it allows for divergences but not convergences, will be more apt when there are forces favoring the separation – physical or social – of people, and a few things should be kept in mind with regard to this point. First, the forces favoring separation (over togetheration) have not been constant throughout human history. Second, each separation has its own history, and separations are not necessarily absolute, not under the definitions used to assign membership into (or exclusion from) the groups of the tree model, nor otherwise. Finally, the applicability of the tree model to a given history depends in part on classificatory and goodness-of-fit decisions which are not reducible to purely objective criteria.

Also, because there is no evidence that language change involves distinctly different natural processes within languages as opposed to between them, I believe that the word “genetic,” as used among linguists, is misleading. I’ll return to this in a later post.

[6] As, for instance, Ernst Haeckel sometimes drew them. See also Alter, 1999, Darwinism and the Linguistic Image, for a fascinating discussion of the relation of the biologist Haeckel to the linguist Schleicher, and a fascinating discussion of the relation of the two disciplines more broadly.

[7] This is obviously too simple, and I can’t possibly be the first person to have said something like this, and so I welcome good references. For one relevant to the relation of theories of language to nationalism in the 19th century, see Joseph, 2006, “The grammatical being called a nation,” in Language & History, ed. by Love.