Bob,



I'm repeating your example sentence below, along with a description of how I'd approach it. First, though, I'd like to distinguish among three different "stances" I sometimes adopt, since I want to ensure that I don't end up trying to present a very narrow personal position as being representative of all functionalists, etc. :



(1)	Approaching a phenomenon from what I think may be an "ecumenical descriptive" point of view; i.e., not making any claims about psychological reality, but advancing a description based on consistency, consensus, and pedagogic utility. "Consensus" here doesn't refer to consensus among functionalists, but rather to consensus among as wide a sample of linguists/grammarians as is possible. 



(2)	Approaching the phenomenon from a "general functionalist" point of view.



(3)	Approaching the phenomenon from my own particular positions on linguistic structures.



You're quite right that I frequently wax kvetchy about overuse of abstract structures, particularly null elements. When I do that, though, I'm either adopting stance #3 (and I hope I'm explicit about that), or I'm objecting to someone else's imposition of *their* stance #3 within what I think is a stance #1 situation ("textbooks should all say X because X is how language is," when X is actually still under debate, etc. -- the equivalent of saying "Textbooks should all acknowledge the fact that <fill in name of political party of choice> are right about economic policy"). 



Functionalist approaches vary in their approaches to abstraction, however, so positions like "as few null elements as possible" don't apply to all of them. I do think there's a tendency for functionalists to be a bit more suspicious of claimed entities that have no realizations (e.g. empty nodes, empty "landing sites"), but the degree to which these are rejected varies, as does the degree to which the particular theory claims to make statements about psychological reality. When I'm adopting stance #3, I do want to make statements about psychological reality, if I can back them up with evidence from psycholinguistic research -- but I try to keep that out of the pool of things I claim there's wide consensus on among linguists/grammarians. 



Now for your student's sentence: 



	They are not agree with the Input Hypothesis.



Your discussion precisely pins down the main problem with "error analysis" -- what we observe as an "error" frequently doesn't lead us to precisely one possible target that's being missed. There's even a third possible target for this one, albeit one that is archaic: "They are not agreed on this issue." And, of course, the student is highly unlikely to have heard "agree" as an adjective from a native speaker -- but (as far as we can tell) is using an interlanguage in which the sentence is believed to be legitimate English. Here's a parallel conjecture: based on the student's prior language experience, s/he is quite likely to have found that the following operating principles work fairly well:



(A) English-speakers tend to use "X is Y" constructions to talk about conditions that X is/are in ("Bjarki is tall / Bjarki is an engineer / Bjarki is ill"). 



(B) Words can be used in more than one way (particularly since this is a Chinese speaker; Chinese either has a lot of different words that sound exactly alike and are closely related in meaning -- a description I predictably dislike (stance 3!)-- or has a ton of verbo-noun-jectives). 



(C) The basic meaning of "agree" has something to do with sharing an opinion or position.



(D1) There's tons of weird stuff going on in English about the fiddly bits at the ends of words, but...



(D2) people can still frequently understand you if you mess up the fiddly bits, although ETS will never acknowledge this. Ever.





All the student has to do is think s/he's talking about a condition rather than an act, and these would favor the student's observed output without it being directly modeled on any one sentence ("X BE Y" is pretty schematic). At this point, I'd like to know how Chinese-speakers normally talk about this kind of situation in Chinese -- do people perform an act of agreeing, or are they in harmony? Assumptions derived from the native language don't *determine* a given interlanguage, but they're certainly one of the factors affecting it. Learners probably don't adopt a general approach of thinking that any structure is possible unless one is given counterevidence, but their ideas about which structures are *probable* could be based on a wide variety of experiences. 



Sincerely,



Bill Spruiell