Author Topic: Crossover effects as evidence for movement  (Read 554 times)

Offline clavela

  • New Linguist
  • *
  • Posts: 2
Crossover effects as evidence for movement
« on: July 25, 2017, 07:54:27 AM »
Hi, I am wondering about the validity of crossover effects as evidence for movement.

Strong crossover as in (1) and weak crossover as in (2) are said to be due to movement of the wh-operator across a coreferential pronoun, i.e. here, movement of the relative pronoun (who and that) across he.

(1) *The man whoi hei thinks you saw ti
(bad in the intended meaning the man such that he thinks that you saw him, but good in a non-coreferential reading, i.e. he thinks that you saw somebody else)

(2) *The man thati hisi mother loves ti
(bad in the intended meaning the man such that his mother loves him, but good in a non-coreferential reading, i.e. somebody else's mother loves him (the man))

These movements cause that in the original position, now a trace occurs - which is an R-expression and, hence, has to be free everywhere according to Principle C of the Binding Theory. I.e.: t cannot be bound by the relative pronoun, therefore coreference is impossible.

So far, so good. I see the point definitely for the strong crossover, given that the pre-movement structure is possible in a coreferential reading:

(3) Hei thinks you saw himi.

I.e., you take (3), you move him, and then the sentence is not possible any longer: evidence for movement. Fine.

However, in the case of weak crossover, I believe that also the pre-movement structure is impossible or at least impaired:

(4) ?Hisi mother loves himi.

If this is the case, how can you take the phenomenon as depicted in (2) as evidence for movement? Given that (4) is already bad, you cannot attribute the impossibility of (2) to there being movement, right? Do you see what my confusion is? I don't even want to say that in (2), there is no movement - I think there is - but I am wondering about the logic of the argument.

I hope you can help me out! Thanks!

Offline Daniel

  • Administrator
  • Experienced Linguist
  • *****
  • Posts: 1581
  • Country: us
    • English
Re: Crossover effects as evidence for movement
« Reply #1 on: July 25, 2017, 08:08:57 AM »
Quote
However, in the case of weak crossover, I believe that also the pre-movement structure is impossible or at least impaired:

(4) ?Hisi mother loves himi.
What's wrong with that? It couldn't be "his mother loves himself".

On the other hand, while I do find (2) potentially confusing, I don't think it's that bad. "John is the man that only his mother loves." It works better with a modifier like that. And certainly if you add "own" as in "the man that his own mother loves". I think one hypothetical possibility for this would be to say that these are instances of relevance coincidental coreference rather than coindexation, if you really want to save the other analysis. But that reading seems possible to me.
Welcome to Linguist Forum! If you have any questions, please ask.

Offline clavela

  • New Linguist
  • *
  • Posts: 2
Re: Crossover effects as evidence for movement
« Reply #2 on: July 25, 2017, 08:30:36 AM »
Hi Daniel, thanks for the quick answer!

Quote
However, in the case of weak crossover, I believe that also the pre-movement structure is impossible or at least impaired:

(4) ?Hisi mother loves himi.
What's wrong with that? It couldn't be "his mother loves himself".

Indeed not. But is the referent of his and him really the same? It's probably clearer with a proper name:

(5) His mother loves John

Is it possible that it's John's mother that loves John? Or only somebody else's mother?

On the other hand, while I do find (2) potentially confusing, I don't think it's that bad. "John is the man that only his mother loves." It works better with a modifier like that. And certainly if you add "own" as in "the man that his own mother loves". I think one hypothetical possibility for this would be to say that these are instances of relevance coincidental coreference rather than coindexation, if you really want to save the other analysis. But that reading seems possible to me.

You are right, I probably shouldn't have used an asterisk for (2). Let's say that the intended coreferential reading is not unproblematic and certainly not the default one, but not as impossible as it is in (1).
What do you think of the following?

(6) The man that his mother loves is John.

I would like to not put either "own" nor "only", as it probably changes the meaning of the relative to some kind-defining structure, i.e. "John is the type of guy that only his mother loves" - where "his" does not refer any longer to "John" but to the concept of person that the speaker is thinking about. What do you think?

Offline Daniel

  • Administrator
  • Experienced Linguist
  • *****
  • Posts: 1581
  • Country: us
    • English
Re: Crossover effects as evidence for movement
« Reply #3 on: July 25, 2017, 10:03:13 AM »
I don't necessarily see (5) as parallel. But--
Quote
(5) His mother loves John

Is it possible that it's John's mother that loves John? Or only somebody else's mother?
It can be coreferential, but not due to structural binding, just coincidental/accidental coreference. "No one loves John." -- "That's not true! His mother loves John!" (Actually, "him" would be more appropriate there because he is contextually introduced. Making a more complex scenario like "No one loves John or Bill" as the first sentence would make the disambiguating proper name work in the second sentence.)

Quote
What do you think of the following?

(6) The man that his mother loves is John.
Very awkward at best!

Quote
I would like to not put either "own" nor "only", as it probably changes the meaning of the relative to some kind-defining structure, i.e. "John is the type of guy that only his mother loves" - where "his" does not refer any longer to "John" but to the concept of person that the speaker is thinking about. What do you think?/quote]I'm not sure we can use coreferentiality to diagnose structure anymore, then. If your intuition is correct about "own" changing things, then I would assert that they can also be 'changed' without "own", in the sense that the structure is ambiguous. This relates to what I said about accidental/coincidental coreference. It's emphasized pragmatically by "own". As I said, that's one way to attempt to save the derivation from a generative perspective. It's a little odd in some ways, but I suppose it can work here. Personally I wonder if the difficulty is in part just due to being pragmatically awkward. It's hard to think of natural contexts for these things. Usually I try to think of a natural context, then saying this sentence in it (or what the alternative would be). But the judgments here are a little tricky.
Welcome to Linguist Forum! If you have any questions, please ask.