Linguistics 522

Assignment 2 (partial discussion)

Syntactic Arguments


When you make an argument you must show your data, using example sentences.

[p. 61] State whether this is a real constituent and what criteria you applied to determine that result:

    (6a) Susanne gave [the minivan to Petunia.]

Zero credit answer number 1: the minivan to Petunia is not a constituent. It fails to function as a unit.

Zero credit answer number 2: the minivan to Petunia is not a constituent. It fails the replacement, standalone, movement, and coordination tests.

Zero credit answer number 3: the minivan to Petunia is not a constituent. It fails to standalone as a sentence fragment. You cannot do passive, clefting, or preposing with it.

Now why is this last one bad? Two reasons. First, I still dont know whether you know what you're talking about because you have not presented the evidence you used to reach your conclusion. namely a senetnce that you judged ungrammatical. There are passive, cleftings, and preposings that DO work in this case:

  1. The minivan was given to Petunia.
  2. It was the minivan that Susanne gave to Petunia.
  3. The minivan was what Susanne gave to Petunia.
Of course all of THESE are irrelevant because they show the constituency of the minivan which is not the question at hand. But if you dont tell me WHICH sentences you've considered, I wont know whether you've actually found the RELEVANT evidence.

The second reason this answer is bad is methodological: We are doing generative grammar, that is, engaging in cognitive science, which means among other things that we have decided to go beyond corpora and use the grammaticality judgments of native speakers as primary data. This means negative judgments are primary data. The fact that a native speaker finds a particular sentence ungrammatical is going to be crucial evidence in all our arguments. We need to know what sentence it is and sometimes even what structure that "candidate" sentence is being assigned because we are going to use evidence like that to make claims about structural constraints that are part of universal grammar.

The moral? If you say something "can't" be done or doesnt happen, construct sentences that are bad to show that it can't be done. Equally important, mark these sentences with asterisks (*) if YOU judge them bad, to help guide me through your argument.

In this case, here's what we need, the relevant movement evidence:

  1. * It was the minivan to Petunia that Susanne gave.
  2. * The minivan to Petunia was what Susanne gave.
  3. * The minivan to Petunia was given (by Susanne). [by-phrases not needed in passive examples]
What's critical in producing such negative data is that you have a particular issue in mind to settle. These are all sentences we would expect to find grammatical if the minivan to Petunia was a constituent. The fact that they are not is evidence against that idea.

Notice it's important for the argument to work that there not be some independent reason why these sentences are bad. So the art of constructing a convincing negative example is to try to make the examples as plausible as possible, and to remove as many interefering factors as possible. So the examples should be short; you should have an idea of what one would be trying to express with this sentence, and what would be expressed should make sense.

To this end syntacticians often use minimal syntactic pairs in making arguments. Minimal syntactic pairs are sentences which usually mean the same thing and differ only in some syntactic property. One meber of the pair is ghood, one member is bad. The sentences are so close that the only explanation for the badness of the bad example is its minimally differing syntactic property. In this case there are natural candidates to provide us with minimal pairs.

  1. Clefting
    1. It was the minivan that Susanne gave to Petunia.
    2. * It was the minivan to Petunia that Susanne gave.
  2. Preposing
    1. The minivan was what Susanne gave to Petunia.
    2. * The minivan to Petunia was what Susanne gave.
    3. Passive
      1. The minivan was given to Petunia.
      2. * The minivan to Petunia was given
      The good sentences differ from the bad ones only in the syntactic property of how much has been moved. Thus it cant be the meanings that are responsible for the anomaly. That anomaly would be explained if we assume that a minivan to Petunia is NOT a constituent. because we are assuming non constituents can't be moved by clefting, preposing, and passive.

      Notice the structure of the argument in this case. The primary data excluding constituency is negative (ungrammatical sentences). The positive sentence are there to help us exclude alternatiuve explanations for why the sentences might sound funny.

What is

This is not an easy question.

Many of you DID use negative data but you had some surprising judgements. I suspect you were just conflating awkwardnmess with ungrammaticality.


  1. It was a passionate love letter from Stacy that Clyde got.
This is definitely awkward, not prize-winning prose.

Unquestionably in most contexts it would be avoided. But that's not the question here. The question is: Is it inappropriate in all contexts?

That is, if you can think of a context in which the candidate sentence would be perfectly appropriate, then it is not ungrammatical. How about this?

  • Clyde looks a little preoccupied reading his mail over there. Did Clyde just get another bill?
  • No, it was a passionate love letter from Stacy that Clyde got.

So the general rule is: When judging a sentence, try to give it its best shot. Find a good context. Having a minimal pair is a good tool for assuring there are appropriate contexts, but when minimal pairs are unavailable just try to find a good discourse context.

for irrelevant

The following judgement feels right:

    * A passionate love letter from Stacy was gotten by Clyde.
If this IS grammatical it seems to mean some thing more like "Clyde obtained/retrieved a passionate love letter from Stacy" than "Clyde received a passionate love letter from Stacy".

This anomaly was used as an argument for the non-constituency of a passionate love letter from Clyde.

But could there be another explanation?

    ? A letter was gotten by Clyde
get at least in this "receive" meaning (as opposed to its "obtain" meaning), doesnt seem to passivize well. That fact aloing could explain the anomalousness of our example.

The moral: Test the quality of your arguments by looking for alternative explanations of the badness of a sentence. If the ungrammaticality of a sentence has an alternative explanation, modify the example in some way to block the alternative explanation and see if the sentence is still ungrmmatical.

In this case we are interested in whether a passionate love letter from Stacy is a constituent. There are two directions we could go.

  1. We can use movement tests other than passive.
  2. We can change the verb, since it is not central to our argument.

Way one:

  1. It was a love letter from Stacu that Clyde got.
  2. A love letter from Stacy is what Clyde got.

Way two:

  1. [They] found a love letter from Stacy ion the defendant's sock drawer,
  2. A love letter from Stacy was found in the defendant's sock drawer.

We conclude with a short eclectic list of things that can go wrong in constructing an example.

  1. Overgeneralizing a syntactic process in trying to apply a movement test. As we just saw, passive has lexical exceptions:
    1. Mary had a fever.
    2. * A fever was had by Mary.
    Yet a fever is a perfectly good Noun Phrase. Always test to make sure the verb passivizes when using a passivization test to argue against constituency.

    Also: Only apply passive to NPs. The only thing that ever gets moved by passivization is an NP. Some people used this example in the particle problem:

      * Out the light was blown by John.
    The problem is this example would be ungrammatical on EITHER of the proposed constituent structures in that problem
    1. [V blow out ] [NP the light ]]
    2. [V blow [PP out [NP the light ]]]
    You can never passivize a PP:
      [They] flew over the field.
      *Over the field was flown.
    Also: The NP you passivize must directly follow the verb (before movement). NP's following prepositions don't generally passivize:
      The bottle fell under the table.
      * The table was fallen under by the bottle.
  2. Information packaging: Moving pronouns to the right is hard.
    1. They gave the book to John/ They gave John the book.
    2. They gave it to John./ * They gave John it.
    3. They looked the address up./ They looked up the address.
    4. They looked it up/ * They looked up it.
    5. He shot a large 3-point buck./? A large 3-point buck was shot by him
  3. Semantics/pragmatics
      * Know the answer.

When doing the standalone test always provide a context that would make a fragment make sense.

For example, this is right:

  1. What did John blow out?
  2. * Out the candle
  3. The candle.

This is wrong:

    * out the candle.

    With fragment and ellipsis tests context is always critical. For example, "Nixon thinks with a fork" sounds pretty weird but in the right context:

    1. How did Gerald Ford eat his salad?
    2. Nixon thinks with a fork.

The correct structure is [[V P] NP].

The key weak point of the other proposal is that it makes [P NP] a constituent. But there is strong evidence that this is wrong:

  1. Standalone:
    1. What did John blow out?
    2. * out the candle.
    3. Movement:
      1. It was out the candle John blew.
      2. Out the candle is what John blew.
    4. Replacement: THere is no clear replacement data.
Bogosity of

Here is a bogus use of the replacement test:

Is the minivan to Petunia a constituent in Susanne gave the minivan to Petunia?

I argue yes on the basis of the following grammatical replacement sentences:

  1. Susanne gave up.
  2. Susanne gave in.
  3. Susanne gave generously.
What went wrong? Aren't all these sentences grammatical? Aren't they all the result of replacing the minivan to Petunia in Susanne gave the minivan to Petunia with a single word?

What went wrong is the sentences you got by replacement bore no systematic connection to the sentence you started with. Arbitrary replacements have --- well -- arbitrary results:

    John believes that Mary stole the beans.
Let's replace the boldface part with one word:
    John hates beans.
Surely this does not show that
    believes that Mary stole the
is a constituent.

We hereby revise the replacement test as follows.

    The sentence resulting from replacement must have the same meaning as the sentence you started with in context.

This can be done two ways.

  1. Anaphoric replacement. The new element is an anaphoric (pronoun-like element) he,she, it, them, one, do(so). The meanings of these elements depend on the words around them (linguistic context).
    1. John bought a bagel and Mary stole it.
    2. John bought a large bagel after Mary stole a small one.
    3. John bought a bagel and Mary did so too.
  2. replacement by synonym.
    BEFORE John blew out the candle.
    John extinguished the candle.
    AFTER WRONG John hid the candle.
    Replacement by synonym will generally be hard to do, since it will be hard to words that mean exactly the same as the replaced material (Is there a word that means exactly the same as the minivan to Petunia). You will almost always find it easier to do anaphoric replacement.

Both work with NPs only.

Case in point, this assignment. Proposed argument against [PP out the candle] analysis of He blew out the candle.

  1. * Out the candle was blown.
  2. * Out the candle is what he blew
So this isnt a constituent, the argument claims. The problem is: What these arguments show is that out the candle isn't an NP, but the prosal on the table (the one you shd be argueing against) is that it's a PP. So you would not on that proposal EXPECT these sentences to be grammatical.

Evidence of MY claim:

  1. The ball fell [PP under the table.]
  2. * Under the table was fallen by the ball.
  3. * Under the table is what the ball fell.