Just How Old Are Noun Phrases?

Scholars and apes have different interests, but how similar is the form of their thoughts?

The thinking that supports a sentence’s basic, two-part structure of topic and comment may be much older than either humanity, the great apes, or even primates, suggests James R. Hurford in a paper in the March issue of Lingua (abstract here, paper here).

Over 40 years ago, as part of a search for linguistic universals, Charles Hocket wrote that

Every human language has a common clause type with bipartite structure in which the constituents can be reasonably be termed ‘topic’ and ‘comment.’

In English, topics and comments are generally united in a sentence that divides into a noun phrase (NP) and a verb phrase (VP); e.g., The president [topic] lied [comment]; General Sherman, on his March to the Sea, [topic] occasionally showed a little mercy [comment]. A few languages, Hurford cites the Polynesian language of Tonga, do not distinguish between nouns and verbs and therefore cannot have NPs and VPs, but they too use the topic + comment structure.

Hurford’s paper builds on the observation that a full sentence can be turned into a topic: e.g., The lying president…, Sherman’s occasional mercy on his March to the Sea…. In these phrases the topic + comment of the earlier sentence has turned into a solitary topic. Hurford examines the reason that

language has evolved in such a way that in actual English [The president lied] exists alongside [The lying president…].

That existence of two ways of stating the same proposition is the central issue of Hurford’s paper and in the course of proposing an explanation he breaks with a very old philosophical and theological tradition that says the function of language is to state the truth.

I must confess that the issue of sentences and noun phrases did not grip me when I first skimmed Hurford’s not-so-easy-to-read paper. Language routinely permits many ways of saying the same thing (e.g., He hit the ball/The ball was hit by him; the trunk of the car/the car’s trunk; Jane arose/Up she stood.) But last week's post got me thinking and I read the paper more closely.

Hurford's essay is a response to an argument published almost ten years ago (in 1999) by Andrew Carstairs-McCarthy who, says Hurford, does lean toward the truth function and who asked why all (!?) languages make a clear syntactic distinction between whole sentences and noun phrases. Why do we have two different syntactic structures for stating the same proposition? If I can state a truth one way, why state it a different way?

The quick answer is that Hurford believes sentences are more than propositions and speech is not primarily oriented toward stating truths. Sentences commonly make assertions that are divided into mention of something already known (the topic) plus an assertion (the comment). Noun phrases serve to mention something already known.

—“Look out! That grizzly bear is charging us.” There’s a sentence that mentions what is already known [the existence of grizzly bears] and an assertion that is news [here comes one].

—The next sentence might be, “Whew! Joe shot that charging grizzer bear.” The play-by-play nature of this conversation makes it a little weird, but the second sentence shows how a noun phrase [that charging grizzer bear] can be used to advance thought by turning old discoveries into new topics.

Having rejected speech’s stating-truth function, Hurford feels obliged to give some alternative account of what speech is about. He sees it as the latest evolutionary development in the long effort to know things outside oneself. In two previous papers (here and here) Hurford argued that a visual pathway in the brain, the dorsal stream, mediates attentional reference in humans and many other species.

Mentally referring to an object, in the sense of directing focal attention to it, was and is distinct from forming true or false judgments about it.

This ‘referring’ is not a linguistic reference, but a mental one based on attention. Hurford's neurological studies indicate that pre-linguistic attention can use four pointers (variables). I'll call the pointers it, this, that, and t'other because these words serve speech as pointers with variable meanings. Here's an example of them at work in speech that depends on abundant pointing:

Look! It is a house and that is a door. You can see t'other door off to the side. This is a key. This opens that and t'other, so you can enter it.

To follow that last sentence you needed to keep four variables straight. The variables are assigned their temporary meaning by focusing attention on something. Hurford maintains that animals can juggle up to four variables at a time. So suppose it is not a person but an elk that sees the charging bear. The elk could know it's coming, it stopped, and still have a few other thoughts to spare.

Hurford suggests an

evolutionary sequence, starting with simple creatures who can notice things about the world, but not necessarily make any long-term cognitive use of such impressions. … [Later came] more advanced creatures for whom some distinction between focused-on information and incidental (or accidental) information may have been useful in a kind of private internal predecessor of public communication … Finally [came] public communications between individuals

This is the anti-Whorfian hypothesis with a vengeance. Not only can we think without words, but Hurford says that by focusing attention to define mental variables the animal kingdom has been thinking non-linguistically for millions of years.

Hurford’s example of a creature who notices things without thinking further is the snail. “It registers darkness and humidity and ventures out from its rock.” That’s a long way from muttering It’s hot in here, but we had to start somewhere. Interjecting my own thoughts here, I suspect that this level of behavior takes up through at least the insect world. This level of “intellect” is strong enough to enable an animal to respond to general conditions, but not to the individuality of a particular here-and-now.

Jumping ahead some many millions of years to the level of a predatory bear who spots potential prey, attends to it and stalks it.Now, even as the prey disappears from the senses it stays in mind and the stalking can continue. Eventually the prey may be close enough to be catchable. The bear attends to the new situation,  it(catchable), and charges. Hurford says:

Although communication has played no role in this tale, and the mental processes involved are all completely private to one animal, the [mental] functions … are somewhat like the functions of Topic and Comment in a sentence. — One kind of mental predicate serves as given background information about the thing concerned [e.g., prey], while the other represents new information about it [e.g., catchable]. … I do not claim that [the terms Topic and Comment] exactly fit the prelinguistic case. — I will coin the cumbersome terms ‘Already-in-Mind’ and ‘New-to-Mind’ to distinguish the roles that predicates play in a cognitively advanced non-linguistic creature.

Hurford proposes a three stage evolutionary process:

• Simple detection: animal discovers condition(x) and acts. (Note: Hurford calls this “simple perception” but because I doubt that snail or insect actions arise from sensations, I’m balking at the term perception.)
• Cognitive operations: animal discovers condition(x) and acts appropriately, then alters its behavior upon recognizing new-condition(x).
• Public communication: speaker and hearer voice stable topic(x) and less stable comment(x).

Hurford’s central thesis is that a basic syntactic structure of most languages (the noun phrase) and a universal pragmatic structure (the two-part topic/comment structure of sentences) is very much older than language, although language has sharpened it. Thus, when thought first “went public” the earliest utterances had this two-part structure. Nothing new was needed for evolution to move our ancestors to understanding or using this structure. Once begun, our ability to speak continued to evolve and language took on many other features, yet the distinction between topic and comment, has “never eroded away” and supports a universal structure found in languages today.

Comments

[Michael]

Thanks for this very illuminating post.
Do you plan to extend the discussion of Hurford's paper to further posts? I think his viewpoint resonates nicley with what you wrote about Speech as a means of directing and sharing attention and conscious awareness in several fo your earlier posts. I think it would be interesting if you further explored the similarities between Hurford's proposals, your own concept of the function of speech, and Michael Tomasello's (and others) theory of "shared intentionality".

Thanks,

Michael.
--------------------------
BLOGGER: It is very unusual for me to have plans for this blog. I have been posting to this site for a little over a year now, and when I started I had no idea how it would go. I was quite unsure that there would be enough material to keep going. Yet I have been able to plug along and certain themes have emerged, one of the most important being that language works by directing a listener’s attention hither and yon. Since Hurford seems to say that mammals have been thinking for millions of years by privately directing their attention to this or that phenomenon, I assume I will have further occasion to mention it.

[gorgo marchetti]

What Bolles reports here is very intriguing. “One kind of mental predicate serves as given background information about the thing concerned, while the other represents new information about it”. This two-part structure reminds me of what I have always held to be the one of the fundamental basis of thought (if not “the” fundamental one).

Think about (one of) the simplest instances of thought (by thought I mean here the general possibility of relating two elements): noun+adjective (e.g., “red flower”). To be able to form this very simple thought, you need a “background” or “topic” (call it whatever you want, anyway, something on which you can mentally, perceptually build something else). Then, on this basis you can perform all the (mental, perceptual, etc.) operations you want.

In my opinion, the necessary operations for this two-part structure are:

a) focusing on something (say, X, that is, the topic)
b) keeping mentally present, by means of memory, X so that
c) the next (attentional, perceptual, mental, etc.) operations are performed using X as their basis or boundary

In this way you get not only the noun-adjective relation, but also some other very general and common form of relations such as subject-verb, verb-object.

According to my analyses, the fundamental mechanism that makes this two-part structure possible is a sensory-motor one.

Giorgio Marchetti

[Robin Shannon]

Firstly, as I understand it (though I do not, myself, deal very much with syntax) is the evidence for the claim that nouns and verbs are not universal is equivocal at best. Certainly there have been several alternate analyses of the evidence put forward which seem to be more generally accepted than the original claims.

Secondly the idea that language is not strictly propositional is hardly new or controversial. The work of Giles Fauconnier may be of particular interest in terms of the new information-old information division you are talking about here (although many other research paridigms also make great use of this eg. LFG, DA, DFG, etc, etc.)

[anni]

wat a stupid website