|
This is the page where I will put questions people ask me and the answers to those questions. It might be useful to check this page if a question arises; it is quite likely that somebody else has the same question as you. They will be listed with the most recent first. It may also be worth checking the forum on the Courseinfo site for this class, too. Dec 1: What is the structure of Bill is a genius? This is the structure that I would have given for this sentence. My reasoning is that genius is the main predicate of this sentence, actually, and that it must be assigning the theta-role to Bill. Because of that, Bill needs to be within its maximal projection (because theta-roles can only be assigned locally). So: Bill starts out in SpecNP, gets its theta-role from genius, and then moves to SpecTP to get Case, as indicated in the DS and SS trees below: DS ...
TP
|
T'
/ \
T VP
[pres] |
V'
/ \
V DP
be |
D'
/ \
D NP
a / \
DP N'
Bill |
N
genius
SS ...
TP
/ \
DPj T'
Bill / \
T+Vi VP
is |
V'
/ \
ti DP
|
D'
/ \
D NP
a / \
tj N'
|
N
genius
Let me add one final note about this: Part of why I think of a genius this way is that there are many languages where you can have a sentence with no verb, just something like Bill doctor (meaning "Bill is a doctor") and in those cases it is clear that Bill gets its theta role from doctor. However, things quickly become very, very tricky. If we broaden our horizons to things like The man in the cowboy hat is the President of the United States, where it really seems like be has taken on the role of predicate, equating the two DPs. A real question left open at this point is whether The man in the cowboy hat is a politician has the same structure as The man in the cowboy hat is the President. What I suggested in the trees above says that they're different: in the politician sentence, politician is the predicate and in the the President sentence, is is the predicate. It is not completely clear to me what the best analysis is going to be in the end. We've entered the realm here of asking What's the best analysis? -- there are several possibilities, and the next step would be to see which of the two analyses runs into less trouble with other sentences. The analysis I propose above has some strange side effects I think. For fun, you might ponder what you think the structure of I consider Bill my friend might be (I said "for fun", but actually, this sentence is guaranteed to give you a headache). Very thorny questions arise, and half the fun of syntax is trying to think of a clever analysis that can accommodate all of the facts. Usually this is the point where people start inventing things like AgrOP and so forth, based on examples that a simpler theory doesn't seem able to capture. We will not be able to come up with a fully satisfactory theory of DPs and be this semester I think. Nov 30: On the class handout pertaining to VP shells, you show an arrow from the lower verb to the light verb. Is this a movement, does it occur in SS? Does the light verb assigns the theta-role to the subject at DS? All exactly right. The light verb assigns the theta role to the subject, and the the lower verb assigns the theta roles to the other two arguments. The movement of the lower verb to the light verb happens between DS and SS. This means that the lower verb does not assign the theta role to the subject, by the way; the light verb does that. Nov 30: Homework 8, problem 1. Where does a genius get case from? Ah, good question. It must be that the Case is assigned by be, there's really no other option. When be is an auxiliary we don't usually have it assigning Case to anything, but when be is used like this, to support a non-verbal predicate like genius, it must be able to, assuming that a genius is a DP and thus needs Case. Nov 30: Homework 8, problem 1. What causes the relevant movement? Another very good question, and one that I don't have a particularly good answer for. We can see that the movement happens based on the word order, and I would suggest thinking about this as the same kind of thing as what makes the verb move to T in French. There's something about T that (can, in some languages) make V move to it. Here, there's something about light verbs that makes the main verb move to it. Chomsky's recent interpretation of these kind of VP shell structures actually is one in which the main verb always in every language necessarily moves to the light verb, whatever kind of light verb it may be. Sorry there's no better answer for this, but a very interesting question. Nov 30: Homework 8, problem 2. Does part C affect the answer for B? This was a difficult question for me to word clearly, but I did intend problem C to be a hint. The answer to problem B is reasonably obvious, and the hint in problem C is not meant to make you doubt your answer to problem B. Rather, think about the question and answer three before this one ("On the class handout pertaining to VP shells, ...") and think about how the theta roles are assigned in a VP shell structure. In particular, the lower verb does not assign the theta role to the subject. The idea in part C is related to that. Maybe that much of a hint will help you see what I was trying to elicit for this problem. Nov 30: Homework 8, problem 2, part C: So if break is unaccusative, where does the vase get Case? Very tricky, excellent question. For the moment, I can't give you a good answer to that. The answer is going to be roughly related to Burzio's Generalization, but the way we know Burzio's Generalization right now, combined with the structure you'll be coming up with, you'll find that it doesn't all quite fit together. I can talk about this more in class on Tuesday, but it will require an extra step that we haven't talked about yet. Nov 30: Homework 8, problem 3: So where does a letter get Case from? The same place as it gets Case in Bill sent Mary a letter -- that is, we don't know. I avoided this in class, because it is quite a tricky question. Richard Larson in the article where this analysis was proposed had a story about how the V' which is the sister to a letter can assign it Case, but it is not as easy argument to make. Maybe we'll talk about this more in class -- it's kind of cool, but also kind of strange. Nov 30: In Bill is a genius, I remember Bill originating in the DP (a genius), but I can't remember why. Could you shed some light on this? Right.. What I was suggesting was that since "genius" is actually the main predicate of the sentence, so it should actually be assigning the theta role to Bill, meaning that it needs to start within the NP (my inclination is to assume that it is in SpecNP at DS). So, just like as if it were a VP, only it's an NP. Nov 18: On homework #7, problem 2. I am very much confused with question #2. In (1), I believe himself is bound by the subject. But how do I show that in the DS? I am not sure when Roberts asks: where does Principle A apply (in the DS)? How do I show where one of the principles applies? Think about Principle A (anaphors must be bound) and what binding means (c-commanding and coindexed with). Wherever Principle A applies, it must be that the binder (which by the meaning must be the subject) c-commands and is coindexed with himself, since the sentence is grammatical. So, if you look at one of the trees and find that the subject does not c-command the anaphor, then it must not be at that level that Principle A applies. Think about this a little and see if that helps set you on the right path. The important thing is that the binder (the subject) has to c-command the anaphor in order for Principle A to be satisfied. Nov 18: On homework #7, problem 2, how do we draw to have been bought? To have been bought is a T and then two auxiliaries and then the main VP, so it's just like "to be eaten" except with an extra auxiliary, like: ...
/ \
T VP
to |
V'
/ \
V VP
have |
V'
/ \
V VP
been /__\
bought
(I'll leave you to figure out the structure of the lowest VP -- there are some traces etc. down there as well). Nov 10: In homework problem 11.2, how do we draw find out? Treat find out as if it were an unanalyzed verb for now; you can replace it with discover, if you'd like. Nov 9: In homework problem 10.2(d), we have Rosemary tried to get a new car. I do not understand how to make a tree for tried to get. I do not understand how to make a tree diagram for two verbs (try and to get). Right. Well, what you have here is a verb with an embedded clause (not two verbs, but one verb per clause), so the relevant part of the tree looks like: ...
V'
/ \
V TP
try / \
to get a new car
I've omitted any PROs that may or may not be there, but that's how the embedding works. So, try has two theta roles to assign: one to the person trying (Agent) and one (Proposition) to the proposition they were attempting to bring about (the TP). Nov 9: I do not understand is where is the PRO and the different types of PRO. Could you shed some light on the different types of PRO? There are basically two kinds of PRO: controlled and arbitrary. The meaning of a controlled PRO is tied to the meaning of one of the previous DPs in the sentence, and the meaning of an arbitrary PRO is more like "someone/anyone". So "[PRO to leave] would be stupid" means "for anyone to leave would be stupid"-- that's an arbitrary PRO. The controlled PRO can be either controlled by the subject of the higher clause or the object (depending on the verb in the higher clause). So "I persuaded John [PRO to leave]" has a PRO which is controlled by the object (PRO has to refer to John, it is John who is leaving), and "I am happy [PRO to leave]" has PRO controlled by the subject (it is me who is leaving). Nov 9: How does Who is to say what to do? work? That looks like a lot of nonfiniteness, is there a Case problem? Is there a PRO? So, Who is to say what to do? is a fine (but rhetorical) question. Here's what I would give as the structure for this. We have to treat be as a raising verb here and do it like this: (DS) CP
|
C'
/ \
C TP
[+Q] |
T'
/ \
T VP
[pres] |
V'
/ \
V TP
be |
T'
/ \
T VP
to / \
DP V' <-- say assigns Agent and
who / \ Proposition theta-roles
V CP
say |
C'
/ \
C TP
[+Q] |
T'
/ \
T VP
to / \
DP V' <-- do assigns Agent and
PRO / \ Theme theta-roles
V DP
do what
And then: PRO moves to SpecTP, what moves to SpecCP, who moves to SpecTP, be moves to T, and then who moves to SpecTP, T moves to C, and then finally who moves to SpecCP. Who will then have gotten case from the topmost finite T, what will have gotten Case from do, back when it was sister to the verb do, and I think everything is fine. (SS) CP
/ \
DP3 C'
who / \
/ \
C+[V4+T]5 TP
is / \
t3" T'
/ \
t5 VP
|
V'
/ \
t4 TP
/ \
t3' T'
/ \
T VP
to / \
t3 V'
/ \
V CP
say / \
DP2 C'
what / \
C TP
[+Q] / \
DP1 T'
PRO / \
T VP
to / \
t1 V'
/ \
V t2
do
It'll probably take you as long to process that as it took me to draw it, but I think that works. Nov 9: So how about John encouraged me to leave. Is there a PRO? The trick to finding PRO is believing deep in your heart that there is no way that one thing can get two theta-roles. Then, assuming that this is true, if you are ever confronted by something that looks like it requires two theta roles to be assigned to the same argument, there must be a PRO to absorb one of them. In John encouraged me to leave, we have basically the same three theta roles for encourage that we had for persuade -- the encourager (Agent), the encouraged (say Goal), and the proposition the goal is encouraged to effect (Theme). Meanwhile, leave still has its Agent theta role to assign to the leaver. The problem here is that it seems that in John encouraged me to leave, me is both the leaver and the one encouraged -- and that can't be, since that's two theta roles assigned to the same thing. So there has to be a PRO. Since PRO can only be in nonfinite subject positions, it must be: John encouraged me [PRO to leave]. Encourage, like persuade, is an object control verb because the reference of PRO is tied to the object (me) and not the subject. Nov 9: Fine, what about I am encouraged to leave? This one is just a passive of the previous one -- so the Agent theta role gets suppressed, the verb (encourage) can no longer assign accusative Case and so the DS object (I) moves to subject position and gets nominative Case. It seems like I "wants" to get 2 theta roles, in the sense that we seem to be assigning it two different functions in the sentence, but since we can't really be doing that, there must be a PRO taking up one of the roles, its reference controlled by I. Oct 20: Errors on the original practice midterm key detected! This email sent to the class covers the changes. A revised version has been posted to this web site. Oct 20: On the practice midterm, last part, sentence (i), how did you arrive at the answer you have in the key? The answer I have in the key made a particular assumption about (i) which probably wasn't fully justified; the answer I gave in the key would better describe the problem with John does not be winner. The assumption I was making was that there was a finite tense/agreement affix in T and then the problem was that the auxiliary didn't move there. Alternatively, however, you could consider the sentence to be nonfinite, in which case what's wrong with (i) is pretty much exactly the same thing that is wrong with (ii). Had this been the real midterm and you'd given the answer I wrote for (ii) as your answer to both (i) and (ii), that would have been correct. Oct 19: I have posted the email sent to the class covering issues raised by homework #5. Oct 19: On the practice midterm, problem 1, sentence (iii), does it matter which of the two possible meanings we draw? I actually hadn't seen the structural ambiguity before it was pointed out to me, but it is true that this sentence could mean that the chair of the department drives, of his/her many speedboats, the one that is kept on the lake. You could draw either that meaning, or the (to me) more natural meaning that the speedboat is being driven, and the driving is happening on the lake. Oct 19: On the first sentence on the practice midterm, if the t of believe assigns Nom to the president, does it also assign Acc to the CP? And if not, what does? Ah, maybe I never stressed this and I probably should have. Not only do all DPs need to get Case, but only DPs need to get Case. So CPs don't need to get Case from anything. So, nothing assigns Acc to the CP. Oct 19: You've got the indices backwards on the example on the practice midterm, right? Right, sorry about that... Oct 19: On the practice midterm key, problem 1, sentence (i), surface structure, it looks like you have a V node (believe) whose mother is also a V. Shouldn't that be a V-bar? Ah, actually, no. What it should look like is: V
//\
V T
believe [pres]
That's actually supposed to be a complex head, formed by head-movement. For some reason, my little "double bars" didn't show up very well on the keys. If you zoom way in on the PDF file you can tell there are double bars, but it's too small to see on the printout. Oct 18: How are theta roles assigned in subordinate clauses? There's no real difference between matrix and embedded clauses with respect to how theta roles are assigned. Remember that a theta role is a formal construct covering the intuitive idea of "participant" or "player" in the event. So, if you have an embedded clause like Mary left in I think that Mary left, Mary is the participant in the leaving (and in fact is the agent who initiated the leaving event), so Mary is going to get a theta role from leave. Perhaps another way to say this is that nothing changes with respect to the theta roles a verb (perhaps in a particular sense) assigns whether it is a main verb or an embedded verb. Leave always assigns an Agent theta role, whether it is in Mary left, John thinks that Mary left or John wants Mary to leave. Oct 18: In your trees, you have inconsistencies in modal and auxiliary verb placement: sometimes you have them in a separate verb phrase, and sometimes you just put them directly into T position. Which is correct? And if the first one is correct, is another T needed to tense the main verb? You didn't show another TP in the tree on page 2 of the homework #5 key. Ok, the short answer is: the first one is correct. Auxiliary verbs are verbs, and as such should head a VP. There's a bit at the end of handout #3 (X-bar theory) about this, and I was trying to be consistent from then on. The textbook often draws the auxiliary as if it starts in T, but I think that is an oversimplification. The tree in Homework #5 (for (e)) is correct as shown, but I think I know what you have in mind. My guess is that your question is: "Why is hated in the past tense? Where's that tense coming from?" And the answer is: Pretend the verb is eat and you'll see. The sentence would come out Have you always eaten peanut butter? which doesn't even look like it has past tense on the main verb. The problem is the same old English problem that the perfect participle often tends to sound the same as the past tense, but the difference is that we're drawing the perfect participle (eaten, written, etc.) as just a regular main verb, introduced by an auxiliary verb. Oct 18: From the practice exam, part 3, #1, you have that be is an auxiliary verb. In most cases I know that this is true, but in this case, is is the the main verb, even though it should undergo auxiliary verb raising. Ah, interesting point. Yes and no. Be here (in John is not the winner) is still really an auxiliary verb; it actually doesn't have any theta roles or contribution to make to the meaning. It turns out that in this case the main predicate of the sentence is really winner. It just happens that in English, we are not allowed to have sentences whose main predicate is nonverbal without also having be (John is a doctor, John is happy, John is on the boat), although there are plenty of languages where things like this come out as, roughly, John doctors, John happys, John ons the boat). I think it is probably accurate to say that be is just plain never a main verb, actually. Oct 18: Also from the practice exam, part 3, #4, do we need to know that this is Principle A of the binding theory, or can we just explain why herself does not fit in the sentence, using the argument of gender agreement in reflexive pronouns? Well, I think saying "Principle A" would be crucial to get all the points on that problem, yes. The fact that the pronoun doesn't agree means that there's no antecedent, meaning that it violates Principle A. Oct 18: I disagree with the way you are drawing the genitive 's-- I think that it should be on the DP of the noun it modifies, rather than on the main noun of the clause (e.g. it should go on the D for Mary rather than on the D for purse in Mary's purse). Can you explain why it is one and not the other? You'd think the genitive 's belonged to the possessor, wouldn't you? That certainly feels right intuitively. One argument comes from the fact that you can attach 's to seemingly anything (not necessarily the head noun in the possessor DP): The man from Australia's hat, or the horse you rode in on's saddle, or the woman that I met's briefcase, etc. It really seems like it is attaching to a whole phrase, that it is coming between a DP and an NP, and what better place to put it than in the head of the (main) DP? Moreover, another reason to think that it is in the head of the DP is that it cannot co-occur with other things that are in the head of the DP like *The man from Australia's the hat (which is supposed to mean 'The hat belonging to the man from Australia') or *The man from Australia's a hat (intended: 'A hat belonging to the man from Australia'). Oct 8: On homework #5, problem 8.1, we're asked to draw a tree for a sentence that has a dummy do (do-support) in it. How do we draw that? Right, I never included any examples on the handouts about this, sorry about that. The way you should draw this is as a regular head-adjunction structure. That would look like this: ...
/
T
/ \\
V T
do [past]
And just indicate that the do is inserted by Do-insertion. You could also get away with just writing it in the "+" notation I was using at the beginning of the previous class, like: ...
/
T
do+[past]
I suppose. Either format would be fine for this problem. Oct 6: At the end of chapter 8, it mentions that if main verbs undergo T to C movement, then the language has V to T movement. Can you clarify? First of all, when it says the language has V to T movement, it means for main verbs, like French. It's true that English "has V to T movement" in the sense that it moves auxiliaries to T, but it isn't a language-wide phenomenon for all verbs including main verbs. Now, the rest of the clarification. When something which started out below T winds up in C, that means that it must have stopped in T (this is a result of the head movement constraint -- you can't skip over T on the way to C). Both main verbs and auxiliary verbs are generated (at DS) below T, so if you see either kind of verb in C at SS it must have moved to T first (and then to C). However, if we are trying to determine whether the language has V to T movement generally (that is, for both auxiliaries and main verbs), we only consider the behavior of main verbs. We consider main verbs because auxiliaries are "easier" to move to T than main verbs are (crosslinguistically) -- so if you see a language moving auxiliary verbs to C, you can conclude that those auxiliary verbs moved to T before moving to C, but you cannot conclude that all verbs in the language move to T. You would only know that auxiliary verbs do (Main verbs: There is basically just one main verb per clause, but there can in addition be several auxiliaries. It is always the last verb that is the main verb, and any other preceding verbs are auxiliaries). As an extra aside, one warning about adverbs when you're trying to create new sentences to test: One thing that we haven't considered is that English not only allows adverbs to be attached to VP, it also allows adverbs to be attached higher in the tree (attached to TP); so you can also grammatically say I always have eaten ice cream I think (maybe it sounds a little weirder, but I can see myself saying it), but this isn't evidence that the auxiliary didn't raise -- because the adverb is already higher than T in the structure, moving V to T wouldn't change the word order between the verb and the adverb, making those sentences irrelevant for the purposes of testing for V to T movement. Different adverbs have different restrictions on where they can appear in the tree -- Some adverbs are a little more limited to the VP, so that's why I was using sloppily (an adverb showing the manner of, say, eating) at one point in class. Time-related adverbs (like never, always, even often sometimes) all seem to have the option of attaching in the TP instead of in the VP, while manner adverbs (like sloppily or completely) are usually more anchored to VP and less able to be attached to TP, so they make better test cases. Sep 30: On homework #4, problem 7.2: Is building a verb with theta roles to assign? For the purposes of this problem, think of building as a verb whose Theme is the house. Recall that nouns too have theta roles to assign sometimes (Like the city's destruction/the destruction of the city or a similar example that I gave in the handout), usually nouns derived from verbs. The Irish literal translation is roughly "To me, building the house is good". Sep 27: On the Theta Theory handout, you show a tree for an is likely sentence and you seem to have made use of a rule Adj' : Adj CP that we haven't seen before. What's happening? Well, the idea is that CP is acting as the argument of likely -- it's very parallel to the verb say, really. One of the big things about X-bar theory is the idea that syntactic category really turns out not to matter in a lot of instances. It's sort of a separate fact that things of the "proposition" type are always going to be CPs, and so if a verb (say) or an adjective (likely) has a "proposition" theta-role to assign, it will kind of indirectly require that the complement (the thing getting the theta role) is a CP. I think the question you're asking, if I'm not wrong, is basically just about the fact that we've never seen a complement of an adjective before, and certainly never seen it being a CP. You can take the existence of the tree on the handout as being an implicit extension of our system for AdjP; Adj is allowed to take a CP complement. Actually, the idea behind theta theory is really to replace the information in the rules like the Adj' rules with more general statements about the "argument structure" of the predicates, which seem a little more like what we'd expect an explanation of what's going on to look like (as opposed to a description of what's going on). Perhaps this is getting too far afield, but at any rate, I hope this helps ease your mind over the CP complement of Adj on the handout. Sep 27: On homework #4, problem 7.1, how is step (2) different from step (1)? Don't they have the same information? Mm, not exactly. Direction 1) is to say something sentence-independent about the verbs we're seeing, while direction 2) is to relate the sentence-independent statement about what theta roles the verb has to assign to the sentence at hand. So, if the verb were hit and the sentence were John hit the baseball, then the 1) part would involve drawing up a theta grid for (at least this use of) hit (regardless of what sentence it appears in) which has two theta roles, an Agent (external) and a Theme (internal). If you draw it like the textbook does, you'd have an "i" and a "j" assigned to them, respectively. Then, part 2) says take that, and now looking at the actual sentence "John hit the baseball", indicate where the theta roles are being assigned. One way you could indicate that is to just draw in the "i" and the "j" (either as subscripts or superscripts) on the DP that gets each theta role. Sep 20: Where else can I look for information about syntax besides our textbook? There are a couple of other textbooks that you might find helpful if you want to hear this stuff in different words (and sometimes from a slightly different perspective); click on the "Bibliography" tab on the left for more information on them. Sep 20: What's going on with NP and DP again? And why didn't we just start with the DP rule instead of spending two weeks with NPs we'll never use again? There seems to have been some confusion about what we were doing there when we suddenly came up with the DP structure to use instead of the NP, and what it means, where we'll use it and so forth. Let me see if I can justify the methods and clarify the point (I will return to this in the next class as well). The whole enterprise we are embarked upon is one of scientific investigation and discovery; as you might recall from the first class, the scientific method goes like this: We observe data, we draw generalizations and form hypotheses, and then we collect more data to see if the predictions of the hypotheses are met. To the extent that the predictions are met, we rejoice, but to the extent that the predictions are not met, we have to modify our hypotheses to reflect this. To that end, we started out observing that things like the student seem to have as their most central element the noun, student, and we progressively worked through some hypotheses about what a noun phrase can contain. Moving to binary branching "X-bar" structures is just one step further in this process; noticing that some parts (and not other parts) of the noun phrase can be replaced by one, we determined that we needed to have (nested) constituents inside the noun phrase, which we named N-bar. To come back to the point, the way to look at this is not as though we had a rule for how to form NPs which you were to learn, but which was then immediately discarded (wasting your time)—rather, we have had several hypotheses, each one hopefully a closer approximation to the truth than the previous one, but each revision made on the basis of evidence we had found that the previous hypothesis did not correctly predict the facts. To simply start with the "final rule" would allow you to have the benefit of the many years of research in syntax and might perhaps save a little bit of time, but it would miss one of the most important parts of subject: the scientific method for coming up with the rule in the first place. As for DP, we initially supposed that a noun phrase like the subject the students contained the determiner, but further data (particularly data from possessive constructions) indicated that we are closer to the truth if we assume that 's, the, and a are determiners which are outside the projection headed by the noun. This doesn't change the fact that the students is still perfectly good as the subject of a sentence, it's just that when we originally called it an "NP" we had applied the wrong label to it, since the head of the phrase the students is in fact the (students being an NP which is the complement of the D the). This means that, for example, in our "TP rule", where we allowed an "NP" to be the subject in the specifier, we need to apply the newly corrected label to the subject and instead allow a DP to be the subject in the specifier. Note that there still is such thing as an NP. There are nouns, and they are heads of an NP which contains both an N and an N-bar node. Moreover, most of what we discovered about them is still true; the facts haven't changed, after all. The only real change is that the places where we thought NPs occurred in the structure are actually places where DPs occur in the structure. The only place we find real NPs under this new and improved view is as the complement of a head D. The subject of a sentence, the object of verb, or the object of a preposition are all DPs. Sep 20: Isn't there something funny about DP and the "Golden Rule of Modifiers"? Well, kind of. The Golden Rule of Modifiers says that a modifier and the thing modified have to be in the same phrase, and in class we basically extended that early on to say that pretty much everything in a phrase (except the head) modifies the head of the phrase. The "funny thing" about DP here is that there's kind of an intuitive feeling that the determiner in some sense modifies (or "specifies" might be a better term) the noun, making us suspect that the determiner should be inside the NP after all. And, in fact, I have no doubt that this has a lot to do with why people didn't realize at the very outset of generative syntax that the student is headed by the, instead of student. But we have pretty good evidence for this now (and there's even better evidence out there in the linguistic literature which we aren't really prepared to understand yet), and so we might just have to accept this as a failure of the Golden Rule of Modifiers. Honestly, though, this use of The Golden Rule of Modifiers is taking a bit beyond what it was designed to do anyway; The Golden Rule of Modifiers really works for nouns and verbs and adjectives—for the "open class" items (categories in which we can readily create new words). For the "closed class" items (also known as "functional categories" sometimes), our intuitions about what modifies what sort of disappear. Think about PPs, for example (P being very nearly a closed class—if you don't believe that, your homework is to invent a new preposition and introduce it into the language); we know that a P like in takes a DP complement like the car to form the P-bar and ultimately the PP. There's no question about this; but how strong is your intuition that the car is somehow modifying in? Not very strong, I'd wager. D is just like P in that it is a closed-class, functional category, and we aren't going to be able to rely on an intuitive feeling of what modifies it to determine what is in its phrase with it—instead, we'll have to rely on other evidence, like that reviewed in class. Sep 20: Didn't you say there's only one specifier? In John's father's brother's roommate, I count three. Yes. Look at any single head (say, a single 's, a D); that head will be the head of its own phrase, a DP, and that DP has at most one specifier. Nothing prevents a different phrase from having a specifier, there just can't be more than one specifier in a single DP. So, in John's father's brother's roomate, the DP John is in the specifier of the DP headed by the 's that appears after John; that DP is John's father, which is in turn a specifier of a different DP, the one headed by the 's that appears after father; that DP is John's father's brother, which is in turn a specifier of a yet another DP, headed by the 's that appears after brother. So, there are four DPs, one being John, and each except the last (the DP that includes the whole thing) being the specifier of the next DP. Sep 19: Who do you think that finds this question to be ungrammatical? On homework #1, one of the problems was to assign prescriptive vs. descriptive to ungrammatical sentences, including Who do you think that saw Bill? The answer for this question was descriptive, and the reasoning was that, first of all, nobody taught us this, and second of all, it doesn't sound grammatical. The idea behind this reasoning is that prescriptive rules tell you which sentences that otherwise would sound fine should nevertheless not be used (that is, you have to be taught not to do it), whereas decriptive rules characterize sentences that sound ungrammatical, that you wouldn't have been tempted to use in the first place. Hence, because Who do you think that saw Bill? sounds bad, in the context of this assignment it must be due to a descriptive rule... but wait! What if you don't think Who do you think that saw Bill? sounds bad? Ah, there's the rub. It turns out that this kind of sentence (sometimes called a "that-trace violation" for reasons that will become clearer later) sounds ungrammatical to a lot of people, but not to everybody. Although I am one of the people for whom it sounds terrible, others will swear up and down that it is a perfectly grammatical question. Neither of us is really "right" or "wrong", just reporting our intuitions. Given this, if you find yourself in the category of people who think it is good, you will have had trouble answering this question, since if it sounds good and yet is supposed to be ungrammatical, logic would dictate that what makes it "ungrammatical" is actually prescriptive, taught. If you happened to have lost a point on this one, let me know, and you can have it back. Sep 18: Can we turn in handwritten trees? Yes, yes! I would actually encourage handing in handwritten trees, even if it means it is not possible to email your homework. It is a lot of work to typeset trees, particularly when they start getting big later on in the semester, and there's no need to do all of that extra work (which is unrelated to syntax). Sep 15: The examples from Bambara in the homework; I don't know how to do this. Let me try to get you started, anyway. First, it's important to look primarily at the words in the original language, using the glosses as your guide. So, if you are trying to decide, say, if there is a gender distinction, you'd look at the examples 'he cried' and 'she went' and see what difference there is between the two in terms of the subject (since they semantically differ in gender) and in terms of the marking on the verb. You'll notice that in the first two Bambara examples, you find that 'he' and 'she' come out, in Bambara, as A in each case. You don't have any real evidence about the verb marking, but there is nothing obvious that suggests that the verb is marked differently when there is a masculine subject (he) vs. a feminine subject (she). So, it appears that gender isn't marked in Bambara (that is to say, there appears to be no gender distinction). As for the rules like the "VP" rules, these are the rules that were developed in class (and in the chapter) of the form: "VP: (AdvP+) V (NP) (PP+) (AdvP+)". The idea is to look at the evidence that you have and see what "VP" is made of, roughly guided by our rules for English and following the same kind of procedure. Of course, the data we have for Bambara and Tzotzil are not as extensive as the data we can come up with in English, so the rules are going to necessarily be preliminary (we won't have any real evidence about whether, where, and how many adverb phrases can be included in VP, for example, so we won't include them in our rules). So, the English PP rule looks like "PP: P NP" (or maybe "PP: P (NP)"), and so we might expect other languages' PP rules to be similar (but, importantly, they could be like "PP: NP P" if the language has "post-positions", where the preposition like 'in' or 'on' follows the object instead of preceding it). So, Japanese and Korean are languages with postpositions, for example, and so would have a PP rule like: "PP: NP P". All of these rules should have a head (a PP should contain at least one obligatory element, a P; a VP should contain at least one obligatory element, a V; and so forth). Sep 15: "Head of lettuce": Is that N, P, N, or is that just one combined noun? This is indeed N, P, N. Only in special cases do you want to consider something with multiple words like this to be a single noun; primarily, these would be cases like names, such as New York or Baton Rouge. We know head of lettuce is made up of parts because you can replace them (head of cattle, leaf of lettuce) to form new noun phrases. This doesn't work so well with names (Improved York, Baton Jaune). Sep 15: I don't understand "exhaustive dominance" even after reading the definition in the book. We actually won't make much use of the concept of exhaustive dominance in this class, but let me attempt to just say it again in my own words and maybe that will help make sense of it. First, we start with dominance: nodes which dominate a node X are roughly speaking the nodes "above" X, the nodes which you'd cross as you draw a line along the branches up from X to the root node (and including the root node) at the top of the tree. So, for the following tree: A / | \ B C D A dominates B, A dominates C, and A dominates D. Now, if you have a list of nodes, for example {B, C}, you can ask if they are exhaustively dominated by A. You would then check:
The only little wrinkle in the definition of exhaustive dominance given in the book is that it is really only talking about immediate dominance, so if you had a tree A / \ B C / \ D E then, the set of nodes exhaustively dominated by A is {B, C} (and not {B, C, D, E}, even though D and E are dominated by A -- they are not immediately dominated by A). So, "exhaustive" just means "complete" -- the nodes which a node X exhaustively dominates are all and only those nodes which X immediately dominates. Sep 14: If there is an embedded clause without a complementizer (e.g., John said Mary left), should we assume that there is a C there anyway, and treat the embedded clause as a CP? Yes, exactly. Until you hear otherwise (there are special cases which are otherwise), any embedded clause at all is a CP. This means that in English we have to assume that there is a complementizer (a C) which has no pronunciation, and that in general the choice between the C pronounced as that and the C that is unpronounced is basically free. So, to reiterate, if you have a sentence like John said Mary left, treat it exactly the same way you would have treated John said that Mary left. Sep. 7: Is punctuation and capitalization included among the "prescriptive" rules? Yes. Generally, when we're talking about syntax, we're talking about spoken language rather than written language (despite the fact that we write out all of the examples). This isn't to say that there isn't anything interesting or universal about writing systems, just that we won't be dealing with them in this class. Sep 7: The textbook doesn't give distributional criteria for prepositions, determiners, or conjunctions. Are there some? Prepositions, determiners, and conjunctions are all closed class items (that is, you can't really make up new ones--give it a try), so you can get away with just a list of them. Determiners include the, a(n), and things like each, every, some, most, ... Prepositions include the standard in, of, on, through, under, above, etc. (plus the weirder ones like aboard), ... "Conjunctions" are really just and, or, but (and I guess nor). For things I didn't list, you could use substitution as a pretty reliable test (e.g., beyond is a preposition because you can substitute it for another preposition). I think the fact that they're closed-class is probably why there weren't any distributional criteria given. Note about logic: Perhaps "coordinators" would be a better term for the category of and, since in logic, "conjunction" means and (which is different from or--or creates a disjunction). Nevertheless, we'll just continue to refer to both and and or as "conjunctions", understanding that this isn't strictly logician-speak. Sep. 4: Your trees aren't binary branching; weren't we taught before that all trees are binary branching? Yes, I forgot to mention this as I was presenting it, but here's the deal: The "Truth" is that all trees are binary branching, but at the moment we don't have evidence that supports this. All we know is that "NPs" are made up of several things, like "AdjP" and "N" and "PP". We only know that these are constituents of NP and for all we know now they are constituents of equal standing. We will, however, very shortly see evidence for splitting some of these constituents apart into subconstituents, the result being that all trees are binary branching. Within two weeks, we'll have strictly binary branching trees, but we'll also know why we do too. If you ever teach a yodeling class, probably the hardest thing is to keep the students from just trying to yodel right off. You see, we build to that. —Jack Handey, Deep Thoughts.
|