markh@csd4.milw.wisc.edu (Mark William Hopkins) (02/01/89)
This unsettling news item is shocking for two reasons actually. First, it reveals that a particular woman is in the unfortunate situation of having no hair of her own. Second, it quietly slips in the revelation that France is still a monarchy ruled in 1989 by a queen. The interpretation of this sentence reveals the underlying structure of its assertions: there exists a person x such that: * x is the Queen of France and * x is bald. Now, even in mathematical texts one will see such statements often abbreviated to forms such as the following: there exists a Queen of France x such that: * x is bald. or there exists a bald person x such that: * x is the Queen of France. where one or more assertions are placed in with the quantifier, and it is widely accepted that these all mean the same thing. The meaning does not change, but the emphasis does. These sentences put different facts in the foreground to be directly asserted or in the background to be assumed. The same underlying meaning still allows you the flexibility to shift different assumptions back and forth between foreground and background. This process is to some extenj continuous: A certain person is the current Queen of France and is bald. The person who is the current Queen of France is bald. The current Queen of France is bald. As we read from top to bottom the assertion "x is the Queen of France" gets put more and more into the background. How much we place assumptions in the background depends on how much we regard its assertion to already be an integral part of the network of assumptions that the speaker believes the hearer to share. It's very a convenient way of getting points across succinctly but can be easily abused, and usually is, by people when they argue dishonestly. Most people, in fact, will argue dishonestly by bringing extra assumptions in the "back door" by using unwarranted background assumptions. It works pretty well pretty often because many people do not know how to respond to assertions that have not been directly made. "Is the current Queen of France bald?" means "Is there a person who is the current Queen of France and is bald?" Their only difference is that the second makes explicit the assumption implicitly made in the first. So the answer is: "No, there is no queen of France."
bph@buengc.BU.EDU (Blair P. Houghton) (02/02/89)
In article <697@csd4.milw.wisc.edu> markh@csd4.milw.wisc.edu (Mark William Hopkins) writes: > "Is the current Queen of France bald?" means "Is there a person who is the >current Queen of France and is bald?" Their only difference is that the >second makes explicit the assumption implicitly made in the first. So the >answer is: "No, there is no queen of France." If so, then I think I'll go learn Esperanto; English will have been rendered useless. "Is the current Queen of France bald?" means 'You are to know that I assume that there is a current Queen of France, and that I wish to know the state of the pate on said sitting monarch, and that I expect the answer from you.' 'I wish to know if there is a current queen of France' never enters into it. The proper answer is: "non-sequitur; there is no current Queen of France." --Blair "...intruders will be destroyed; resistance is useless; exterminate; Ex-Ter-Min-Nate; EX-TER-MIN-NAATE!"
eric@snark.uu.net (Eric S. Raymond) (02/03/89)
In <697@csd4.milw.wisc.edu>, markh@csd4.milw.wisc.edu (Mark William Hopkins) writes: > How much we place assumptions in the background depends on how much we > regard its assertion to already be an integral part of the network of > assumptions that the speaker believes the hearer to share. Interesting. I know of one domain where the use of implied existentials (the kind of `backgrounded' assertion you describe) is not only superior technique but a positive good -- the writing of science fiction. Consider the number of background existentials asserted in the following: "As Koreth's aircar skimmed over the rise, the Forerunner city stood revealed in the light of the twin suns." By modern SF standards this is a well-constructed expository sentence. Robert Heinlein and other writers associated with John W. Campbell's _Astounding_ magazine developed this technique of exposition by implication in the late Thirties and early Forties. Yours is the first analysis of implied claims I have seen that suggests a formal characterization for it. One school of SF criticism, focusing on the importance of this kind of construction, has claimed that SF is a genre properly characterized not by its props or history but as a *style of reading*, one in which (among other things expository) text is deconstructed *primarily for (what you have described as) its background existentials*. Thus, to a person who is practiced in the SF style of reading, the above sentence is resonant with color and meaning; it implies an entire universe of exotic technologies, interstellar travel, and at least one long-vanished species of extraterrestrial (nor would an SF reader assume `Koreth' is human!). To a person unpracticed in this kind of deconstruction, on the other hand, the sentence is at best a sort of gaudy purple word-bauble and at worst simply an incomprehensible noise. I don't personally think this school tells the whole story; it is unable by itself to explain some other consistent features of SF (such as its tendencies towards technophilia, individualist political philosophies, and a sort of prosyletic zeal about rationality) but I think it certainly illuminates SF's characteristic prose constructions and explains why outsiders so frequently miss the point of them. Here's the tie back to AI. The process by which SF readers assemble the implied counterfactuals of a text to construct a consistent world-picture is a marked version of what any language-understanding system must accomplish. What makes the SF case interesting is that experienced readers have no apparent problem maintaining internal representations of *multiple worlds*, some with utterly bizarre premises. My own introspection suggests that this is handled by evolution of a set of `prototypical' worlds that work out the implications on traditional SF themes; new stories are either analyzed as variations on a type or become new major prototypes. SF writers have become very skilled and subtle at exploiting the analytical reading style of their readers; in fact, reading a new novel often has the aspect of a sort of intimate intellectual contest between reader and author, in which a large part of the reader's fun comes from reconstructing the author's world and anticipating what the author will do with it. Meanwhile the author tries to evoke a world without telegraphing the plot. Comprehension in this kind of context would make a wonderful `torture test' for knowledge-representation and language-understanding systems, because it emphasizes precisely the challenges in comprehension of `normal' prose that are most difficult to capture. Perhaps we should be trying to aim our machines at appreciating _The_Moon_Is_A_Harsh_Mistress_ rather than _Dick_And_Jane_! -- Eric S. Raymond (the mad mastermind of TMN-Netnews) Email: eric@snark.uu.net CompuServe: [72037,2306] Post: 22 S. Warren Avenue, Malvern, PA 19355 Phone: (215)-296-5718