Talk:Raising (linguistics)

From Wikipedia, the free encyclopedia
Jump to: navigation, search

Moved from article[edit]

Moved from article:

John seems [t to work hard]
The verb seem cannot assign the agent role; thus a non-expletive NP won't receive any theta-role which would violate the Theta Criterion (all arguments must be assigned a theta-role). So we have to suppose that the non-expletive NP raised from the Spec-IP position in the embedded clause, where it received the agent-role from the verb.

RuakhTALK 03:15, 24 May 2007 (UTC)

What's wrong with my explanation? Russky1802 04:19, 24 May 2007 (UTC)

This is an encyclopedia entry. This means that you need to explain what you're talking about. (On Wikipedia the need to explain things is slightly lessened by the ease of linking to related articles — you can linkify terms like theta role — but you still need to accommodate your average reader, who does not already understand these things.) A number of your terms require explanation, and some (such as "NP" and "Spec-IP") simply should not be used. Additionally, this is one single entry; insofar as it discusses multiple concepts, some sort of transition is necessary to relate these concepts to each other. (Also, your "explanation" is kind of begging the question, in that it simultaneously assumes that John has an agent role but that seems can't assign the agent role. These are both true statements, but their truth is justified by recognizing seems as a raising verb, not the other way around. Note that your "explanation" wouldn't explain a sentence like "John seems to be angry", where John has a theme role, unless you want to argue that seems can't assign the theme role, either?) —RuakhTALK 04:53, 24 May 2007 (UTC)

I agree with your first suggestion. However, I can't agree with the second one that the raising verb "seem" assigns the agent role to an external argument. Please compare the following sentences:

John fell. (unaccusative)

John is reading a book. (transitive)

John jumped over the fence (unergative)


*John seems.

*John seems a book.

*John seemed over the fence.

The asterisk * means that the sentence is ungrammatical. So the verb "seem" cannot take an NP-object (like to read), it cannot occur with the subject only (like to fall) and it doesn't take a PP-object, either (like to jump). THAT IS WHY linguists claim that the verb "seem" cannot assign the agent-role. However, if you can give me sentences of the following pattern

X seems. X seems Y. where X and Y are noun phrases all linguists (including me) will change our point of view.

Now your sentence, John seems to be angry. Its structure is as follows:

John seems [t to be angry], where t stands for the NP-trace, the original position of "John" (btw, adjectives as well as nouns do assign theta-roles, cf. John's criticism of the theory or We found John critical of the theory). Please remember that all overt non-expletive NPs must receive a theta-role. The reason why "John" moves in your example is because the tenseless I-head (the particle "to") cannot assign a case, so the Case Filter would be violated. Thus "John" moves to the Spec-IP in the matrix clause where it receives the Nominative case from the [-Pst] I-head.

I'm saying the things that are accepted by the majority of linguists working or who worked in the GB theory framework.

Good luck! Russky1802 06:15, 24 May 2007 (UTC)

I can't agree with the second one that the raising verb "seem" assigns the agent role to an external argument.
What? I said no such thing!
Also, the GB theory framework is by no means universally accepted. (This doesn't mean you can't use it in the article, but you have to say that you're doing so.)
RuakhTALK 06:19, 24 May 2007 (UTC)

quote your "explanation" is kind of begging the question, in that it simultaneously assumes that John has an agent role but that seems can't assign the agent role. These are both true statements, but their truth is justified by recognizing seems as a raising verb, not the other way around. unquote

We assume that "John" receives an agent role because of its position, Spec-IP. It's UTAH, proposed by Mark Baker. We assume that the verb "seem" cannot assign a theta-role to the external argument because of the reasons I mentioned in my previous message. Do you see how it works now? Russky1802 07:10, 24 May 2007 (UTC)

some clarification about object-raising[edit]

while expect is object-raising: in "She expected him to do that," him is syntactically the direct object of expected, but semantically the subject of to do, and the sentence could be rewritten as, "She expected that he would do that."

I believe you misinterpreted things in your explanation. It is a wide-known fact that the verb "expect" can be either an ECM verb which takes an IP-complement (She expected him to do that) or a subject-control verb (She expected to win the contest: She expected that he would do that) which, in its turn, takes a CP-complement.

However, there is NO subect-to-object raising in "She expected him to do that". Its structure would be as follows:

She expected [IP him to do that]

You may find a brief and a very good explanation in Ouhalla, J. (1999). Introducing Transformational Grammar (pp. 166-168 and pp. 197-199). I'm sorry but I have no time explaining the nitty-gritty of syntactic theory for every sentence.Russky1802 18:08, 24 May 2007 (UTC)

Is there any empirical difference between saying that something is an ECM verb and saying that it is a subject-to-object raising verb? I always thought that the invention of ECM was the result of being painted into a corner by other assumptions about the distribution of expletives and case assignment. Other linguistic frameworks that do not make these assumptions can analyze expect as an object raising verb without any difficulty. And for the purposes of this article, unless it becomes a much more detailed and high-level discussion of formal linguistics, it seems to me that saying (descriptively, pre-theoretically) that there are four types of verbs — subject/object × raising/control — is sensible. On the other hand, the statements about object-to-object-raising are highly dubious… CapnPrep 19:24, 24 May 2007 (UTC)

ECM verbs[edit]

If we assume that there is a subject-to-object raising (NP "him" raises from Spec-IP in the embedded clause to the complement of VP in the matrix clause) it would make NO sense to postulate the existence of ECM verbs. NP "him" would simply get the accusative case from the verb "expect" and there won't be anything "exceptional" about case marking. If we adopt the subject-to-object raising in such constructions we don't need to have ECM verbs at all. Russky1802 19:53, 24 May 2007 (UTC)

That was exactly my point. CapnPrep 21:04, 24 May 2007 (UTC)

ECM verbs_2[edit]

However, the biggest problem for your analysis would be the following: where exactly does it raise? The embedded clause is in the complement position of VP. There would simply be no empty node for the raised subject.

/    \
V    XP
    /  \
   NP   I'

So NP moves where? Russky1802 20:01, 24 May 2007 (UTC)

Keep in mind that although everyone uses the term "raising", not everyone actually assumes a raising transformation, and the idea of "empty node" doesn't mean much in a lot of frameworks. Similarly, hypotheses/assumptions/principles like the EPP, theta criterion, case filter, UTAH, etc. etc. may not exist in the same form, or at all, in other frameworks, so it is not appropriate to make use of them in the introductory part of this article (i.e., the entire article, at the moment). Eventually it might make sense to have a section about the transformational analysis of raising, with a discussion of ECM and why object raising doesn't exist in this particular model, with its particular assumptions. CapnPrep 21:04, 24 May 2007 (UTC)

Raising of non-subjects[edit]

I moved this paragraph out of the article:

Further, it is possible for the raised argument to be an object even in the more-embedded clause; for example, consider "He is easy to please," where he is syntactically the subject of is (easy) but semantically the object of to please. To allow for such verbs, a more general terminology is used, one that specifies both the syntactic argument type (relative to the raising verb) and the semantic argument type (relative to the verb argument). In this terminology, seem is subject-subject-raising, expect is object-subject-raising (at least in sentences like "She expected him to do that"), and is (easy) is subject-object-raising. Object-object-raising verbs also exist, as in a sentence like "I consider him easy to please," where him is syntactically the object of consider (easy), but semantically the object of to please, and the sentence could be rewritten as, "I consider it easy to please him."

English only allows raising (and control) of embedded subjects. The easy example given here as an example of object-to-subject raising is problematic for two reasons: (i) It cannot involve raising, since easy assigns a semantic role to its subject, and (ii) tough-movement (in English) is a non-local dependency, unlike raising (He is easy to want to try to […] please). Also, easy is not a verb, but that's a problem with the current name of the article.

There are (apparently) languages that have raising and control of embedded non-subjects. If these are mentioned in the article (if someone has references), I think we should use the less confusing terms "object-to-subject" and "object-to-object" raising. CapnPrep 10:52, 25 May 2007 (UTC)

object-subject raising[edit]

Though I didn't write this article and the passage that you removed, I think there is some point in it. Compare:

Doll was hard [PRO to see t]

What problems does that analysis have? Russky1802 22:05, 25 May 2007 (UTC)

I don't understand what you mean by "compare": this sentence is structurally identical to the original example, so it has the same problems I mentioned above. First, the subject of hard apparently must be referential (unlike the subject of seem or the object of expect). E.g. if one accepts that make it to the church on time contains an expletive object, then the ungrammaticality of *Iti was hard to make –i to the church on time suggests that the object is not simply raised to subject position. Second, the supposed raised object is not necessarily the object of the infinitival complement of the adjective; it can be arbitrarily deeply embedded inside this complement (Dolli was hard to persuade John to tell Mary to get Bill to see –i). CapnPrep 23:52, 25 May 2007 (UTC)

First, the subject of hard apparently must be referential.

Not really. What about such sentences like:

  • It was hard to do it.
  • It is hard to say.

I see nothing referential in expletive subject "it".

Your examples do not involve tough-movement/"object raising"! (This is clear in the first example, since do is followed by its DO. The second example is ambiguous, but there is no analysis where the initial DO of say is non-referential it.) So they don't tell us anything about the properties of hard in the other construction. CapnPrep 09:34, 26 May 2007 (UTC)

In your example we have the idiomatic "make it" which is different from "make something": I couldn't make it on time. (non-referential it) BUT: I made a copy of the report. I made it. (referential it)Russky1802 04:46, 26 May 2007 (UTC)

Of course make can take a referential DO in other uses. But again, this is irrelevant to the analysis of my example, where the DO is non-referential, and cannot be "raised". CapnPrep 09:34, 26 May 2007 (UTC)

Second, the supposed raised object is not necessarily the object of the infinitival complement of the adjective; it can be arbitrarily deeply embedded inside this complement (Dolli was hard to persuade John to tell Mary to get Bill to see –i)

I see lots of things violated in this sentence. The most obvious and simple is that the NP-trace is not bound in its GC, which would violate Principle A. Then the ECP is violated because government is blocked by a kazillion barriers.Russky1802 04:54, 26 May 2007 (UTC)

Sorry, I should have said: my coindexing only indicates coreference (and my "–" is not a trace, it only indicates the canonical position of the unrealized DO of see). I am not suggesting that Doll moves from the end of the sentence to the beginning; I am not suggesting that anything moves at all. I don't know what the transformational analysis of this example would be, but the point is that it must involve something more than raising. CapnPrep 09:34, 26 May 2007 (UTC)

1. Your examples do not involve tough-movement/"object raising"!

Of course they are not. I tried to question your claim that s-selectional features of the adjective "hard" require a [+HUMAN] subject (external argument).

2. do is followed by its DO.

Which framework is it in? Who proposed that? What do you mean by DO? I don't recall such an analysis. it doesn't look like GB or Minimalism.

3. The whole thing about this tough-movement is that it is so un-Chomskian, I mean language-specific (at least, so far). Whta about other languages? Examples from French, German, Latin, Russian, Chuvash, Ancient Greek are welcome.

Russky1802 18:14, 26 May 2007 (UTC)

make it[edit]

1. Sorry, I just realized what you mean by DO (direct object). As for the idiomatic "make it" it can't be extracted because the verb and "it" are one constituent (you're welcome to prove the opposite). How can you move the immovable and extract the unextractable? So your example is totally irrelevant for this discussion.

2. I think you are mixing things from traditional grammar with the nothion of object and TG-based grammars where we have complements. In traditional grammar that long sentence would have the Complex Object. There were not any complements in that framework. However, some people stick to some mishmash of classical grammar and GB. It's like wearing a tuxedo at McDonald's. Russky1802 17:47, 27 May 2007 (UTC)

Tough movement[edit]

Removed from article (sorry): Tough movement (GB interpretation)

There has been much debate about such sentences as:
John is easy to please.
It is easy to please John.
While in the second example the analysis is rather obvious we cannot say the same about the first one. At first it seems the structure is as follows:
Johni is easy [CP PRO to please ti]
However, such an analysis would pose many problems. The chain <Johni, ti> would be doubly case-marked, NP John would get [NOM] and the trace would get [ACC]. This would violate the condition on chains (Chomsky 1986), according to which there must be a unique case-marked position and unique theta-position in a well-formed chain. To avoid further problems with case assignment we have to suppose that the trace is not case-marked. It would mean that the trace is [-anaphor, -pronominal]; to put in simply, it is a wh-trace, which is subject to the Bijection Principle (it must be A'-bound by an operator). Thus we have to postulate the existence of an non-overt (null) operator in Spec-CP in the embedded clause. The structure of the sentence would be as follows:
Johni is easy [CP Opi PRO to please ti]
(for detailed discussion see Haegeman 1994, Chapter 8.4 Non-overt antecedents of wh-movement, pp. 463-473)

As far as I know, there has never been a serious proposal for treating tough-movement as an instance of raising. So it doesn't deserve a whole section in this article, and especially not a section that will be incomprehensible to a large majority of readers. I have added a couple of sentences about this kind of structure, including your reference, but without displaying all the theory-specific technical details. CapnPrep 12:43, 30 May 2007 (UTC)

ok. i don't mind. :) Russky1802 17:38, 30 May 2007 (UTC)

Raising: a form of argument control?[edit]

I have a problem with the first sentence: In linguistics, raising is a form of argument control... For me, raising and control are clearly separate, if similar. What do you think? -- UKoch (talk) 15:10, 4 April 2012 (UTC)