Bob, ****************** If "Bob" is the assertion, then why can't it be: 1) *Bob, isn't he? ****************** With the "Bob, isn't it" example, it's possible that the speaker and hearer simply use knowledge of likely speech purposes given the context -- in other words, given the situation, a speaker is much more likely to make a guess about the source of the phenomenon than the identity of a person, even if a person is the source of the phenomenon. Note that "Bob, isn't it," in an ellipsis account could be shorthand for variety of "underlying" expressions, like "That noise is being made by Bob, isn't it?" or "It's Bob out there in the hallway, isn't it?" They're not really assertions about Bob's identity; they're assertions about the probable cause of the noise (and as a side note, the fact that there ARE multiple possibilities would render this a problem case for ellipsis as well -- most analysts constrain ellipsis by confining it to those cases in which everyone would fill in exactly the same missing material). "Bob, isn't he" *would* in fact work in some situations -- you just need a context in which someone's identity is being confirmed. While "isn't it" will work in the following, I think the "he" version will too (I don't know any plays with characters named Bob, so I'm having to shift to Hamlet): [Context: Cletus and Bocephus are watching a play being performed by the local theater club. ] Cletus: Can't figure out the guy on the left. Bocephus: Hamlet, isn't he? If that doesn't work for you, the corresponding positive tag to indicate skepticism might: Cletus: Can't figure out the one on the left. Bocephus: Oh, *Hamlet*, is he? That's not a good Hamlet. ************************** 2) *We are canceling the play in which the lead actor is sick, isn't he? If there is not concept of "sentence" why is (2) not a possible sentence? ************************** I mentioned in the previous post that restrictive relatives seem to be backgrounded to the point where they can't be "checked" or "contradicted"; this is a good example of that. I brought up subordinate clauses because they're the type of thing that most grammars class as dependent, and which we certainly have to *punctuate* as if they're dependent, but which may be foregroundable to the point where they act like independent assertions rather than subsidiary ones. Note that even a main clause (or what's treated as one in traditional grammar, at least) can be backgroundable if it's the kind of thing that can't easily come into play in the discourse: Bocephus: I think the play starts at 6:00 Cletus: No it doesn't. ?? No you don't. Again, I have no problem with the notion that people frequently communicate with clause clusters. What I do have trouble with is the notion that, given any abstract sequence of clauses like [A B C D E F], there's a single unambiguous boundary-procedure that will produce (for example) [ [A B C] [D] [E F] ], and that the groupings thus produced are "real" in some sense. The tradition of punctuation we've inherited ********************** The point is that for corpus linguistics to have ANY results requires underlying knowledge of the language (1) to know what to search for and (2) to evaluate the examples for relevancy. *********************** I would argue that there's a crucial distinction between what is entailed by the phrase "underlying knowledge of the language," and what is entailed by "competence." The second (if taken in terms of its use by Chomsky) has a very specific meaning. Competence is fully deterministic, is not controlled by context, and is grounded in a "mental faculty" that is unlike any others people have. It's Saussure's "langue" as if interpreted by Plato. If language isn't fully deterministic, is *inherently* controlled by context, and represents one application of general cognitive functions that subserve other, nonlinguistic, abilities, we can still have knowledge of it, but that knowledge won't be "competence" according to Chomsky's definition. Of *course* corpus linguists use "underlying knowledge of language" -- any given item one searches for is structurally polyvalent; it's the status of the item as part of a *construction* that's usually of interest. Acknowledging that constructions exist, and acknowledging that we can talk about them, does not require us to accept OR reject Chomsky's notion of competence. Bill Spruiell Dept. of English Central Michigan University To join or leave this LISTSERV list, please visit the list's web interface at: http://listserv.muohio.edu/archives/ateg.html and select "Join or leave the list" Visit ATEG's web site at http://ateg.org/