In the post Phenomenology of Names and Twin Earth, I said that if one accepts the position that common nouns are based on awareness of multitude of things (real or assumed) which share some similarity, the consequence is that in the Twin Earth scenario, Oscar and Toscar mean same thing by “water” before they figure out that water is H2O, and that twater is XYZ. That is, because the similarity on which their words are grounded are shared across both water and twater.
In the comments of that post, I compared two scenarios to back up that claim:
1 – Martin sees a bunch of elm trees. He becomes aware of the gestalt similarity of this multitude and names those “trees”
2 – Later, on some other place, Martin sees beech trees and says – “Ah, there are more trees here”
1′ – Martin sees bunch of elm trees. He becomes aware of the gestalt similarity of this multitude and names those “trees”
1a’ – Martin further puts attention to the form of their leafs, branches, roots and so on. He gets more knowledge of the extension that he is acquainted with (and which he calls “trees”).
2′ – Later, on some other place, Martin sees beech trees. But in this case he notices the difference between beech trees, and what he called “trees”. So he says – “Ah, there are those things here that are similar to the trees, but those are not trees”.
If you find the second scenario sounds wrong to you, just change the word “trees” with the word “elms”. Nothing substantial changes, by changing the word that is used.
Both scenarios seem normal to me, but in the first scenario word “tree” ends up meaning tree, while in the second scenario “tree” ends up meaning elm. However everything is same in both scenarios, so there has to be some change of meanings which corresponds with the additional knowledge that Martin gained in second scenario in step 1a’.
Richard in the comments said that if the additional knowledge changes the meanings of the words, then communication is not possible. One can point here to two things:
That it is a fact that this kind of differences of meaning appear in the conceptual development. Frank C. Keil in his book Concepts, Kinds and Cognitive Development does different tests of the development of different concepts in kids.
One example is where he asks children if some kind of transformation would change certain thing (animal, mineral, artifact) from one type to another. Here is an example of the story about tiger/lion “transformation”:
The doctors took a big tiger that looked like this. They used special fur bleach to take away its stripes, and they sewed on a huge mane so that it ended up looking like this. Was this animal after the operation a tiger or a lion?
The question was asked to children of different age, and the results were as in the following graph:
The X axis represents the kids’ grades K (5 to 6 year olds), 2 (7-8 year olds) and 4 (9 to 10 year olds). The Y axis represents the answers that were given, where 1 = judgment that transformation changed kind type, 2 = judgment indicating indecision on that issue and 3 = judgment that operation did not change kind type.
So, it seems that there is some change of what kids mean by “tiger” or “lion” through their development and I think that the graph goes nicely with the idea that first kids become aware of multitude of things that show gestalt perceptual similarity, and that only later they become aware of other kind of similarities that hold between certain multitudes. (This is surely a oversimplification, but I think it could be analyzed in more details if needed).
But what to do with objections like Fodor’s from Thought and Language, when talking about possibility that children and adults might mean different things by the words, he says (citation also taken from Keil’s book):
They must misunderstand each other essentially; and, insofar as they appear to communicate, he appearances must be misleading. Nothing less than this is entailed by the view that word meanings evolve.
I want to point here, that as long as the two meanings (in the sense of awareness of some similarity) are grounded on the same extension, the communication might go without bigger problems, as in both cases it is finally the extension that one is aware of (i.e. the multiplicity). So to say, as long the things that show the gestalt perceptual similarity which is ground for the usage of the common noun of certain person (e.g. “tree” in the example) are the same things which show some other kind of similarity that is ground for the usage of the same word for another person, those two persons will agree in lot of cases on the use of the noun. For example Martin-scenario1(after 1) and Martin-scenario2 (after 1a’), would not have lot of problems of communication until such things as 2/2′ happens.