*
* Lecture notes by Edward Loper
*
* Course: Ling 591 (Pragmatics II)
* Professor: Ellen Prince
* Institution: University of Pennsylvania
*

[01/16/01 04:38 PM]
Start project soon.
Ideas:
  - pick a construction that isn't the normal way of saying things
    (.e.g, "chocolate I like" or "chocolate, I like it", etc.).  Under 
    what discourse conditions are they used?

> Form/Function
Is there a deep/meaningful/hardwired relationship btwn form &
function?  If so, what is it?

>> Theories
>>> Theory: Function is a reflex of form
Form is the basic notion
e.g., new info comes at end of the sentence.

>>> Theory: Form is a reflex of function
Function is the basic notion

>>> Theory: Relationship bwtn form & function is arbitrary
This has traditionally been Ellen's position
The relationship is learned and language-dependant.  Nothing hardwired 
about it. 

>>> Theory: Form is "motivated" but not "determined" by function
This theory is not (yet) very well articulated.  c.f., construction
grammar (Zwicky, Goldberg, etc.)
e.g., there may be a good processing motivation to put new info at the 
end of the sentence.

>> Consider these theories from perspective of various issues:
  - given/new info
  - topic/comment
  - centering
  - discourse function of (non-cannonical) syntax
  - point of view/discourse structure
  - language contact: what happens when 2 languages interact?

What are the rules for placing different kinds of info structurally?
Can we find statistical correlations?  What are their implications?

[01/24/01 04:32 PM]
> Discourse
discourse -- coherent linguistic output (or interaction).  (not
necessarily spoken, but spoken is primary).

stylistic conventions may make written text have unusual patterns of
language use.. e.g., peoples' use of complementizers in text is
different.  esp. for edited text/speech

Macro-level stuff in discourse:
  - turn taking, etc.
  - stuff that doesn't require parsing
Micro-level discourse:
  - patterns of use
  - ways specific propositions are expressed

We'll be looking at micro-level discourse.  In particular, any
proposition can be expressed in many different ways.  Why?  What do we 
use to choose which way to express something?  What function do these
variations play?

What types of meaning *can* syntax have?  What meanings can lexical
items have?  Is there a meaning that can't be expressed lexically?  

E.g., we take into account what a hearer knows about an
entity.. discourse new/old, hearer new/old, etc..

Information structuring -- giving (hierarchical) structure to
propositions..  e.g., in "john ate the bannana," difference between
stress on john, ate, bannana.. and "it's john that ate the bannana"
and "the one who ate the bannana is john"...  

>> On structure-preserving transformations
Emonds looked at various transformations.  He divided them into
structure-preserving (basically source\to source transformations) and
non-structure-preserving (called them "root").  The
structure-preserving transformations can be input to further
transformations, and can occur in non-matrix clauses.  The root
transformations mainly just occur in the matrix clause, and can't be
further transformed.

It's much easier to figure out the function of root transformations
than structure-preserving transformations.  So we don't have as much
hope of (easily) figuring out the function of structure-preserving
transformations (such as passivization).

>> On derivation vs. surface structure
From what we've seen so far, it seems that what matters in examining
the function of a form is its surface structure, not its derivation.
This is good, because there's not much agreement on how all these
forms are derived. :)

>> Types of Anaphora
coreferrentiality      (it)
set-membership         (the groceries/the milk)
functional-dependancy  (the house/the door)
script-relation        (we were going there/we honked the horn)

> Class Format
same as ling 590: class presentations and final conference with 15+5
minute presentations.

[01/30/01 04:22 PM]
> Clark & Marshall 1981: Definite Reference and Mutual Knowlege
Mutual Knowlege Paradox: definite reference requirees speakers to
check a list of facts or beliefs thtat is infinitely long.  But
definite NPs are selected in a finite amount of time.

Central questions:
  - What type of shared knowledge is needed for language use?
  - How is shared knowledge assessed & secured in practice

Define:
# t = term:     The definite NP
# r = referent: The definite referent

Conditions for felicitous definite reference:
  1. Mary must know that t is R
  2. Mary knows that John knows t is R
  3. Mark knows that John knows that Mary knows that t is R
  4. etc\ldots

[02/06/01 04:36 PM]
> Lambrecht 1988
A man {who} met my brother went to the store.
  vs.
A man {who} I met went to the store.

So we decide we can drop 'who' for non-subjects, but dropping for
subjects causes unacceptable ambiguity, so we eliminate it..

> Ward and Birner 1995

[02/21/01 04:36 PM]
> Gundel, Hedberg, Zacharski 1993

Questions:
 - How do speakers choose a form for a referent?
 - How do hearers correctly identify the referent?

Theory:
  - forms signify different cognitive statuses
  - allows hearer to resrict possible referents

Propose a hierarchy:
# in focus >                     (it)
# activated >                    (that)
# familiar >                     (that N)
# uniquely identifiable >        (the N)
# referential >                  (indef this N)
# type identifiable              (a N)

Inferrable from activated/in focus usually is uniquely identifiable..
# I threw a ball at the SUV, and _the_driver_ scowled at me.

> Horn 1986

[02/28/01 04:33 PM]
> Centering! Whee!
  - How you process discourse
  - discourse complexity
  - topichood

[03/21/01 04:41 PM]
> Ward 1990

What is the function of VP preposing?
1. Affirm belief in a salient proposition
2. Suspend belief in a salient proposition

[03/28/01 04:38 PM]
> Kuno 1976a
Formalism vs. Functionalism
  - Formalists (e.g., chomsky) say that language facility is modular,
    with syntax module operating in isolation from outside
    information, etc.
  - Functionalists say that it's not so modular, and that many
    features affect the syntax module, etc.
  - Moderates\ldots

- Atonomy of syntax?
- We only have access to acceptability, not gramaticality..

>> Gapping

John likes rice, and Mary [likes] apples.
John showed me a bowl, and Harry [showed me] a cup.

>> NAC (No-Ambiguity Condition)
blocks gapping in acses where there could be structural ambiuity as to
the location of the "gap"..

[04/04/01 03:35 PM]
> Relative Clauses
>> Kuno 1976
# *This is the man that Mary knows a girl jealous of t
#  This is the man that I read a statement about t

in terms of elementary trees?

go vs. come -- empathy difference..
bring vs. take

>> Akmajian
# He said that the book was bought by young women.

We can extract from [that..]
# Those are the young women that he said the book was bought by.

But if we change it to:
# He said about the book that it was bought by young women.

Then we can't extract anymore:
# Those are the young women that he said about the book that it was
# bought by. 

>> Antohny Kroch: Resumptive Pronouns

# John said about the car that the engine had to be replaced

# Rest def:      The man that (*he) has 2 heads| \ldots
# Nonrest def:   George|, who (he) has 2 heads, \ldots
# Rest indef:    A man| that (he) has 2 heads \ldots
# Nonrest indef: A man| named George, who (he) has 2 heads, \ldots

We have a unique card constructed/retrieved at the point marked "|".
We can't refer back until we have a single card.. 

[04/11/01 04:27 PM]

> Ward 1998