Ostensive definitions don't work?

I was listening to an ABC Radio National philosophy podcast (the philosopher's zone), and the host and guest talked about Quine's criticism of ostensive definitions. The guest relayed this anecdote about a westerner pointing to a rabbit and asking an aborigine what he called it in his language. The aborigine responded, "gavagai." Quine noted that it's unclear whether the aborigine meant "gavagai" to refer to just part of the rabbit, the rabbit as it stands now, or maybe even a superstition that the rabbit gives rise to. Quine concluded that the meanings of ostensive definitions alone are indeterminate.

At first glance, this seems to pose a problem for Objectivism. Rand viewed ostensive definitions as the simplist and basest part of each conceptual chain. Source. If ostensive definitions don't elucidate meanings, it seems these chains won't much get off the ground. Rand also viewed irreducible primaries (i.e., axioms) as defined only ostensively. Source. It wouldn't be good if axioms were of indeterminate meaning because axioms are at the core of Objectivism.

I hadn't heard Quine's critique before, so I'm putting it out there for civil discussion.

Jordan

Post 1

Monday, February 2, 2009 - 5:23pm

No, this is not a problem for Objectivism.

Rand addresses the directly and extensively. There is the principle of parsimony. One does not multiply concepts beyond necessity. Developmentally, concepts for perceptual level types of entities come first - dog, tree, rabbit, ball, and so on. This is universal.

Also, extensive studies of the naive classification systems of "primitive" cultures show that classifications correspond in very large and detailed degree to the classifications of biologists, usually species for species. Some higher level classifications like "bug" or "worm" or "critter" or "fish" in the broad sense like "shellfish" do not correspond to real biological clades, but they still usually fit some biological grouping just as biologists use such terms as "herps" and "invertebrates" which also do not correspond to real biological clades.

Quine's argument itself depends upon the clear and unambiguous meaning of the basic concept rabbit which he takes for granted. He cannot point to any real world example of a language which has basic concepts for parts of animals but not animals themselves. He is relying on our ignorance of and suspicion of exotic foreign languages to make such an hypothesis seems plausible.

Steven Pinker (who buys into a lot of linguistic and philosophical nonsense by default) is still smart and honest enough to bring up this exact example in his book The Language Instinct and to discard it as a false problem.

I wonder how long it would take Quine to learn the difference between punch, slap, pinch, and kick if we used him as the subject for an ostensive demonstration of the meaning of those words?

Post 2

Monday, February 2, 2009 - 5:29pm

I'm not the least bit swayed by Quine's argument. Firstly, it's about learning the meaning of a word in a highly unfamiliar language. Secondly, it is an isolated incident. When children learn their native language, they pick up word meanings by hearing them many times. They often get feedback when they try to use a new word. Other words already known are used to delimit meaning.

Post 3

Monday, February 2, 2009 - 6:55pm

Ted: I have to express my admiration of your education, insight and the ability to express it.

So there.

Sam

Post 4

Monday, February 2, 2009 - 7:25pm

Sam, I agree. Holy crap. (sorry, all I could come up with)

Post 5

Monday, February 2, 2009 - 8:53pm

There are many detailed studies of folk taxonomy as compared to the scientific taxonomy of biologists. Folk taxonomies are bottom up affairs. That is, there will normally be a term for every unique taxon in a culture's environment. There will often be genera for species which are obviously related like black bear and grizzly bear. There may be mid-level terms such as bush and tree or bug that encompass more abstract groups. But there may not be higher level groups. The groups that do exis are always formed hierarchically, and they become less and less likely the more abstract they are.

Take "biblical" english for example. We have mention in Genesis of fish and fowl and beasts, but no mention of mammals or vertebrates or even really animals in the technical biological sense. The word beast is limited to four legged land animals, and does not include birds, so it is not synonymous with our word animal. Indeed, the word beast itself is not even native to English, it comes from French.

There was no basic un-analyzable native root word for English plant. That word, like animal, comes from Latin. Indeed, the ancient Indo-European word for tree, deru meant both "tree" and "oak." There was a folk word in English for tree, weed, grass, flower (originally bloom), and so forth. If pressed, we might say all the beasts and fish and birds and things that creep upon the earth, or all the trees and flowers and green things that grow upon the earth. One could say living thing if one had to. But there was no simple root for the abstract concept. So, if gavagai were a simple root, what are the chances it would stand for some bizarrely abstract concept?

And even now, what is the singular native word for one bovine or one human not distinguished by sex? We have man and woman and child. We have cow and bull and calf. We have men and cattle. But we have to use the Latin human or bovine to speak of an individual of these species without regard to sex - and even then, terms like human or person or individual have special technical and learned connotations that a word like dog which refers to one canine does not.

As for "gavagai", the word is made up. It is not an example from the real world that actually supports Quine's thesis.

English is what is called an isolating,/i> language. That means that most words have one base form that does not vary or inflect. We have minimal inflection of verbs for tense and person, and of nouns for number. Synthetic languages like Latin inflect almost every word, noun, verb or adjective, and the forms are so complex that one word can make an entire English sentence. While most English verbs have five forms (speak, speaks, spoke, spoken, speaking) and some have as few as 3 (fit) or as many as 8 (to be) Latin verbs normally have over 140 forms each. (For instance, caveat means "let-(he she or it)-beware." In fully isolating languages like Chinese, there are no real endings at all.

So let us assume that "gavagai" is a word in a complicated tricky synthetic language like Latin. Lots of languages require a noun to be marked for possessor. So let's assume that the prefix ga- means "my." And many languages mark nouns for gender and number. In Latin, filius means "son" but it really means fili-u-s "child-masculine-singular." (And "fi-l-i" means "suck-l-ing" if you look at the development of Latin, so we really have "teat-verb-little-masculine-singular," but we needn't go that far.) So what if ga-vag-a-i means my-rabbit-feminine-singular or "[That's] my doe rabbit"?

Confusing?

Well, not really - no more confusing to you than "thatsmydoerabbit" would be to a speaker of Quinese. Eventually, if our native Quinan pointed to his sister's two baby male rabbits and said la-vag-o-r and then offered you soup while saying y-ab-da ga-vag-mish-e (You-try-please my-rabbit-bisque-dir.object) then you would figure out what "vag" meant. And just as you would likely simplify your speach and say "rah-bit" rather than "thatsarabbitrightthere" to a foreigner, the Quinan would likely say "VAG-ai" (A rabbit) to make things simple for you.

Unless he was a professional academic philosopher.

Thankssamnmike!

(Edited by Ted Keer on 2/03, 9:23am)

Post 6

Tuesday, February 3, 2009 - 3:04am

What Sam said. Damn, Ted!

(Still waiting for that "Favorite Post" button)

Post 7

Tuesday, February 3, 2009 - 6:49am

Now you're showing off. :-)

Sam

Post 8

Tuesday, February 3, 2009 - 7:38am

Good discussion, Ted. You said:

I wonder how long it would take Quine to learn the difference between punch, slap, pinch, and kick if we used him as the subject for an ostensive demonstration of the meaning of those words?

That might be difficult, given that he died in 2000. Or, more precisely, he ceased to be "the value of a variable".

BTW, if anyone is interested in a discussion of Quine from an objectivist point of view, I recommend B. John Bayer's talk, from one of the ARI summer seminars, entitled "Understanding 20th-Century Philosophy: The Case of Quine".

Thanks,
Glenn

Post 9

Tuesday, February 3, 2009 - 9:28am

Glenn,

That might be difficult, given that he died in 2000. Or, more precisely, he ceased to be "the value of a variable".

That's real funny. Not that he died, though (as communicated by pointing to his casket); I mean the other part.

Ed

(Edited by Ed Thompson on 2/03, 9:29am)

Post 10

Tuesday, February 3, 2009 - 9:59am

We had an opinion poll here recently that exhibited an interestingly Quinical folk taxonomy of coin styles:

8% Miss Liberty in different styles and poses
4% Allegories like the Panama-Pacific commemorative.
17% Famous Capitalists
12% Just the weight and fineness.
58% The government should not mint coins.

Let us imagine we wanted to break up the color spectrum, but we didn't already have a set of basic color words to do this. We could follow Quine and the above pollster, and, using a little creativity, come up with the following categories:

Pink
Hot Pink
Peach
the Orange color of a can of Off bug spray
Charcoal Grey
Colorless

this would be somewhat like Quine's type of putative "basic" concepts there-is-rabbiting-going-on, "the visible upper left hand surface of a rabbit" and "source of lucky charms."

But do normal people or real languages split the world up in this fashion? We don't need to resort to the armchair. We have real world evidence.

One of the most famous studies of all time of language and cognitive psychology was Berlin and Kay's Basic Color Terms of 1969.

It turns out that while English has eleven basic color terms (red, orange, yellow, green, blue, purple, brown, pink, white, grey and black, with grey being an old borrowing from French and purple (a dye) orange (a fruit) and pink (a flower) being the most recent additions) many other languages lack some of the distinctions we make (Spanish has no basic term for brown, in Japanese aoi means blue or green) and others make distinctions we don't (Russian has separate basic terms for light blue and dark blue).

But, it turns out that there is a hierarchy of basic colors.

If a language has only two color terms, they will be dark and bright. Black and white will be the "exemplary" colors to match those terms. If a language has three color terms, it will distinguish "dark" "bright" and "colorful." The shade that will exemplify "colorful" will be what we consider pure red. But the term will cover a spectrum including pink and orange and purple, while yellow, white and light blue will be considered bright, and so on. (The Spanish word "colorado" means, broadly, red and corresponds to this "colorful" term.) As a language adds color terms, it will always adopt them in a hierarchical order:

Stage I: Dark-cool and light-warm (this covers a larger set of colors than English "black" and "white".)
Stage II: Red
Stage III: Either green or yellow
Stage IV: Both green and yellow
Stage V: Blue
Stage VI: Brown
Stage VII: Purple, pink, orange or grey

So we will not normally have people divide up the world in such bizarre ways as Quine or our pollster above would have us do.

(Edited by Ted Keer on 2/03, 12:10pm)

Post 11

Tuesday, February 3, 2009 - 10:34am

First, I echo Sam, Mike and Teresa's statements: It is a joy to watch a first class mind in action! Go, Ted!
--------------

On Quine, I see it as an attempt to discredit meaning. He had several phrases that attacked the ability to know:

"The indeterminacy of translation" - claiming that one cannot have any certainty as to what the meaning of an ostensively defined word is
"The inscrutability of reference" - a word can't be pinned down as to what particular properties or even objects are being referred to,
"the museum myth of meaning" - words are not labels for 'meaning objects' - this last may have been more of an attack on a Platonic view of meanings having an external reality - living in a world of ideals.

He ends up contradicting himself by attempting to use words to mean that there is no meaning. He attempts to say that because any single pointing at a rabbit cannot guarantee a correct understanding of 'rabbit' that it can't be acquired. But, if his readers didn't have a 'good enough' understanding of what 'rabbit' means, his thesis would be meaningless. The flaws in his reasoning seem to be in attempting to create an artificial limit to the learning effort leading to understanding (it might take many instances of rabbit pointing to get the job done, and it might take iterative attempts to refine the understanding. He is also setting his artificial limits on what makes an ostensive definition, while setting no upper limits on what can be known about the subject at hand (a rabbit in this case). He calls for the impossible (a perfect understanding of all there is to know about rabbit, and a perfect correspondence between what all parties know about rabbit, and that it happen with a single pointing. He doesn't accept that a valid meaning can exist to start with, or that it is adequate for a given context, or that it can improve.

Post 12

Tuesday, February 3, 2009 - 12:58pm