Copernicus
Industrial Grade Linguist
I was once in a meeting of robotics researchers in which one participant said that we needed to develop theorem-proving systems that were better at resolving contradictions so that they could resolve ambiguity in their perceptions. Another participant pointed out rather forcefully that we needed to develop systems that could manage contradictions rather than eliminate them, because reasoning under uncertainty always involved contradictory information. Robots need to be able to change their minds, and they can't do so if they always resolve contradictory information.
Optical illusions are contradictory models of the same visual sensor data.
Consider a Necker cube:
We perceive it as a 3D image that faces in one of two directions. It is a visually ambiguous fiction, because it is really a 2-dimensional image with two overlapping squares connected with lines. We resolve the contradiction by thinking of one square in the foreground and the other in the background. So we can switch back and forth between the two, depending on the way we construe the context--different perspectives brought to bear on the same visual data. Deduction plays a role in that perception, because we deduce the sides of the cube on the basis of that shifting perspective.
Linguistic ambiguity is much more complex than a Necker cube, but the principles that govern linguistic disambiguation are much the same. That is, we build up different contexts in our minds, and we deduce other aspects of the context to arrive at the meaning of a linguistic expression. Linguistic meaning is always an illusion created from logical building blocks. As one of the great semanticists of the 20th century, Charles Fillmore, once put it to me--"Language is word-guided mental telepathy." That is, it enables us to see into each other's minds by evoking shared knowledge that we convey through language conventions.
Whether it is appropriate to say that the bag contains "2 oranges" depends on the context that is the key to understanding the expression.
Optical illusions are contradictory models of the same visual sensor data.
Consider a Necker cube:
We perceive it as a 3D image that faces in one of two directions. It is a visually ambiguous fiction, because it is really a 2-dimensional image with two overlapping squares connected with lines. We resolve the contradiction by thinking of one square in the foreground and the other in the background. So we can switch back and forth between the two, depending on the way we construe the context--different perspectives brought to bear on the same visual data. Deduction plays a role in that perception, because we deduce the sides of the cube on the basis of that shifting perspective.
Linguistic ambiguity is much more complex than a Necker cube, but the principles that govern linguistic disambiguation are much the same. That is, we build up different contexts in our minds, and we deduce other aspects of the context to arrive at the meaning of a linguistic expression. Linguistic meaning is always an illusion created from logical building blocks. As one of the great semanticists of the 20th century, Charles Fillmore, once put it to me--"Language is word-guided mental telepathy." That is, it enables us to see into each other's minds by evoking shared knowledge that we convey through language conventions.
Whether it is appropriate to say that the bag contains "2 oranges" depends on the context that is the key to understanding the expression.