At the end of the English-syntax course I co-taught last semester, my colleague and I set a number of examination questions designed to test students ability to argue points about syntactic structure. This one will serve as an example:
|Although the following two sentences exhibit a superficial similarity, they contrast sharply in syntactic terms:
 I saw Jane with her new boyfriend in the bar.
 I saw Jane and her new boyfriend in the bar.
Show that these two sentences have radically different syntactic structures, using at least two different syntactic arguments.
Getting students to answer questions of that kind involves changing the whole way they think about language. People who have never studied linguistics always think meaning is the key to everything. In fact, they find it hard to imagine that one’s interest in language could lie anywhere else. Undergraduates not fully accustomed to syntactic thinking are subject to the same temptation, and will often meander off into writing whole pages about how seeing Jane with her new boyfriend in the bar means actually spotting them in each other’s company, whereas merely seeing Jane and her new boyfriend in the bar could be accomplished by seeing her in the bar alone at 2 p.m. and seeing him there at 5 p.m. with someone else… and so on. None of this is false; but it can gets no credit as part of an answer to the question above.
The question asks for syntactic arguments. What that means is that we syntax instructors are looking for an answer along these lines:
|I will give four arguments for drawing a structural distinction between  and . They all stem from the fact that with her new boyfriend is a preposition phrase (PP), while and her new boyfriend is a noun phrase (NP) prefixed with a coordinator, functioning as the second component of the coordination Jane and her new boyfriend.
Argument I. Jane is the object of saw in , but in  it is the first element in a coordination, so from  we can derive a passive counterpart with Jane as subject like , but from  we cannot form an analog like :
 Jane was seen with her new boyfriend in the bar.
 *Jane was seen and her new boyfriend in the bar.
Argument II. The comitative adjunct PP in  can be positioned after the temporal adjunct PP in the bar instead of before it, but parts of coordinations cannot be similarly reordered, so we see this contrast:
 I saw Jane in the bar with her new boyfriend.
 *I saw Jane in the bar and her new boyfriend.
Argument III. Coordinate structures do not allow disruption by such operations as preposing of interrogative words. Thus we find this contrast:
 Guess who I saw with her new boyfriend in the bar.
 *Guess who I saw and her new boyfriend in the bar.
Who can be associated with the gap where the direct object of saw would have been, as in , but in  who is not the object of saw; rather, it is the first component of a coordination. And coordinations are universally barred from having gaps as components.
Argument IV. English prepositions can be optionally stranded (separated syntactically from their complements), but coordinators like and have no such privilege. Thus we find this contrast:
 It was her new boyfriend that I saw Jane with in the bar.
 *It was her new boyfriend that I saw Jane and in the bar.
English allows the object of a preposition like with to be relocated in the way seen in , but not an element that follows a coordinator as in .
All of these contrasts provide arguments that the structure of  is very different from the structure of .
In our dreams, and in the work of our best students, we get answers like that. But I still worry about the students who drift off into wasting crucial final-exam time in doomed and irrelevant discussions of what the example sentences mean, what it means to be “with” someone, etc. I have never found a reliable way to forestall this. Some students (like all nonlinguists) seem to be just drawn to the meaning conveyed by a sentence, like moths to a flame, even when they have been explicitly instructed to set that topic aside and investigate its syntactic structure.