Yesterday’s post (here) discussed an essay (abstract here) in the in the September issue of Trends in Cognitive Science that sets forth a theory called "Simpler Syntax" as a rival to the dominant, Chomsky-inspired theory of a syntax rich enough to capture all meanings not covered by the bare words themselves. Today I want to look at Simpler Syntax's implications for the evolutionary origins of syntax.
Standard theory (as illustrated at the recent Stellenbosch conference) imagines a preliminary protolanguage with one or more evolutionary jumps leading to syntactical speech. Protolanguage might consist of single words or a couple of words joined together in no particular order. In short, it’s what we see in children today. Around their first year they start speaking single words; half a year or so later they begin adding two words together. That last seems so natural that most parents don’t even notice when it begins. By age three they are lumping three or more words together in a loose order, when suddenly they begin speaking in true sentences.
There are two main views of the single-word phase. Either the children are just pointing something out, or they are speaking “holophrastically.” When a fourteen-month old girl says doggie is she just noting the dog, or does she mean something more along the lines of there’s a doggie or perhaps even oh, look a nice little doggie all cute and cuddly? How you answer that question seems like a matter of taste, certainly not one of evidence, although Occam’s razor argues against holophrastic claims. Even so, many linguists believe that children speak holophrastically.
The transition to multi-word utterances presents a problem. If children speak holophrastically to begin with, why do they ever bother to switch to actual phrases? If doggie means I see a doggie, why spend your wind on the whole sentence? On the other hand, if doggie merely identifies something present, why don’t children speak more fully from the beginning?
Attentional linguistics shaves off these problems. Single-word speech calls attention to a topic like doggie. The listener responds more fully, oh, that’s a big doggie, drawing the speaker’s attention to more about the topic. Eventually, the child is expanding its attention from the start, biggie doggie.
These kinds of interactions are widely observed during the one- and two-word phases of language acquisition, and there is no reason to suppose that early speech proceeded in any other way. Joint attention expands the horizons of both speaker and listener. Of course, in the early days no speaker was so much beyond the other as parents are with children today. The kind of rapid push that an eighteen-month old experiences today was not available to the early speakers, but so what? Over the past 2.4 million years there have been many generations and much time for improvement. Instead of protolanguage, we might want to call it tunnel attention. They did not take in whole scenes in a single sentence the way speakers do today.
But eventually children do start speaking more elaborately. They don’t just expand their attention; they add syntax and its benefits to their speech. Something similar must have happened during the evolution of speech. The standard explanation for this change, as demonstrated at Stellenbosch, is that one or more mutations came along, probably late in the day, creating a fundamentally different being from the one that existed back during the protolanguage/tunnel-attention days.
Back in the 1980s I contributed to a text book on child development (here) and that work made it plain that three-year-olds are profoundly different from two-year-olds, much more different than they are from four-year-olds. At three, children become interested in one another. Before that they were interested in their parents and older children, but not their age peers. Go to a pre-school. The two-year-olds are sitting by themselves, playing alone. When they interact, they are stealing each other’s toys. The three-year-olds are playing together and speaking together. But, according to the scenario being developed on this blog, that new sense of community cannot be the change that led to syntax when speech was first evolving.
Two or more million years ago, the human ancestral line underwent a selection-based change that produced a more community-oriented species, more interested in one another, less based on dominance and more on respect. These are the sorts of changes one sees in three-year-olds today and they occurred long before, possibly more than a million years before, modern syntax appeared.
The Simpler Syntax article offers a way to understand how to move from two words to syntax without requiring a new sort of being. It uses “pieces of structure,” that are managed like words. The article offers as an example, the idiomatic phrase, “kick the bucket.” Anybody learning English will not be able to understand the term by parsing it. It is a verb phrase with a unitary meaning, to die. You just have to learn it.
Culicover and Jackendoff, the article’s authors, write, “All languages contain thousands of such complex stored units.” The Oxford English Dictionary often follows its definitions with long lists of idiomatic usages for particular words. And indeed this kind of speech is very common among three-year-olds. Iwannaplay, itsmine, nofair. Children speak these language chunks all the time. They can be broken down into formal syntactic structures, but there is no evidence that the children understand them as anything but chunks of meaning just like any other long word. Once they have a chunk, however, it can be broken apart. “Kick the bucket” might be changed to “slammed into the bucket” for a car crash, or “jumped off the bucket” for a suicide. We are all one in our common mortality, “Hers was a very expensive, gold-plated bucket, but in the end she too kicked it.”
Breaking a chunk into words is not radically different from putting two words together, especially if you are already putting words together. If you can say, big doggie and lil doggie you can take a word like bigot and make it a lilot. So if you hear that your friend’s father is a big ot, you might suppose that your schoolyard friend is a lil ot. There is no such phrase, but what does that mean to somebody who never heard of a grammar book?
Experience at breaking words apart gives novice speakers access to whole range of syntactic structures whose formal structures are unknown. Culicover and Jackendoff cite numerous idiosyncratic syntactic structures that allow for some substitutions:
- Far be it from [Noun Phrase] to [Verb Phrase]: Far be it from me to disagree with you.
- [Prepositional Phrase] with [Noun Phrase]: Off with his head; Into the house with you.
- How about [any phrase]: How about a scotch; how about we talk.
- [Noun Phrase] and [Sentence]: One more beer and I’m leaving.
- The more [Sentence]: The more I read, the less I understand. (p. 416)
These chunks are described in formal terms, but we should not assume that users, especially linguistically naïve users, understand them that way. If I know the sentence Far be it from me to disagree and understand Far be it from to be some sort of statement of deference, I am unlikely to substitute an adjective like blue for me. That’s because Far be it from blue to disagree makes no sense, even in an already somewhat confusing context. Meaning is pushing the substitution, not syntax. At the same time, if I am socially skilled I won’t say seriously, Far be it from you to disagree either. There is no formal reason to debar such speech, but it is too much of a claim of dominance to be socially acceptable. Social relations are the constraint here, not syntax.
There is much more to syntax development than this, even in Simpler Syntax. This post is only a discussion of the implications of about a quarter of one column of text in the Culicover and Jackendoff article (p. 416, box 3). Readers familiar with Chomsky’s early writings will know that he was shaking his head 50 years ago, insisting that the kinds of substitutions listed above cannot possibly be the whole story. But they do suggest part of the story of how syntax arose:
- Tunnel attention focused on single things or experiences. Single words emerged. (e.g., fig accompanied perhaps by pointing in the direction of the fruit)
- As speakers’ tunnel attention broadened, two words were needed where one was previously sufficient. Whole phrases emerged as a single chunk of utterance. (e.g., goodfig)
- Sometimes a single word might be recognized in a string of sounds and replaced, (e.g., bad fig). It might look like formal syntax has emerged, but meaning is driving the changes, not syntactical knowledge.
- Over time, a series of chunks and partial replacements were remembered just as words were remembered. (e.g., Iwanna play Iwanna eat; its mine, its big; no fair, no good.
This kind of development is not going to give us somebody who can say When in the course of human events it becomes necessary…, but we’ve come a long way from single-word speech and we have done it without needing any special mutations favoring syntax.