*!*Ghomeshi, Jila and Diane Massam. 1994. To Appear, Linguistic Analysis



tải về 302.24 Kb.
trang4/10
Chuyển đổi dữ liệu11.06.2018
Kích302.24 Kb.
#39842
1   2   3   4   5   6   7   8   9   10
causative event.

[Intrans $\rightarrow$ temporal progress is identical and single simplex event]

There are causally related events which are NOT linguistically represented
as being in a causative relation.

*!*Hovav, Malka Rappaport and Beth Levin. 1998. Building


Verb Meanings.  In The Projection of Arguments: Lexical and Compositional
Factors. Miriam Butt and Wilhelm Geuder (eds).  CSLI Publications. 97-134

%Bar Ilan and Northwestern Universities. 1996.


%IATL Workshop on Syntax-Semantics Interface, Workshop on
%%Argument Structure at the Deutsche Gesellschaft fure Sprachwissenschaft, the Wuppertal Conference on Lexical Structures, the Collloque de Syntaxe et Semantique de
%%Paris, the University of Maryland and the University of Pennsylvania.
%(what the Elactisity paper turned into)

Projectionist theories: theories which project lexical properties


onto syntax.

A major challenge: the fact that many verbs can appear in a "bewildering


range" of syntactic contexts (exs: sweep, whistle, run.
(they note that the range of contexts is associated with classes
of verbs)

Constructional approach (Borer 1993; Ertischik-Shir and Rapoport 1995;


Ghomeshi and Massam 1995; Goldberg 1995; Hoekstra 1992; Hoekstra and Mulder 1990): only some aspects of meaning reside in the
verbs themselves.  Certain constructions or
positions are also associated with elements of meaning.

``the major difference between the two [approaches] concerns


whether the association between the
constant and the template is registered in the lexicon or not." (pg 13)

This is a case study of two verbs break and sweep.

Verbs with impressive elasticity in meaning lexicalize manner.

Result verbs show a much narrower range of variation.


A verb of change of state, such as break, dry or widen..lexicalizes a particular
achieved state, and the verb denotes the bringing about of this state." 101
Directed motion verbs (come, go, arrive) are considered to be a type of
result verbs.

footnote: some verbs lexicalize both (e.g cut)

Manner verbs: allow omission of their DO (Leslie swept.
*Kelly broke.); can occur with a variety of unsubcategorized
objects (*Kelly broke the dishes off the table.
to mean, Kelly broke the dishes and as a result they went off the table;
*Kelly broke the dishes into a pile to mean Kelly broke the dishes and made a pile out of them.

[
BUT:


\eenumsentence{ Kelly broke the eggs into a bowl.
Kelly sliced the salami into a pile.
]

directed motion verbs are rigid:


\eenumsentence{ The students went.
The students went to the beach
* THe jetsetters went themselves ragged.
*The runner went his shoes to shreds.
*The pediestrian went clear of hte oncoming car
*The coach went the athlets around the track.

[BUT:
\eenumsentence{ The students climbed the mountain.


The students climbed to the top.
The students climbed themselves ragged.
The students climbed their shoes to shreds.
The cat climbed clear of the approaching bear.]

Manner is not a necessary prerequisite to variability in argument expression.


Get has no manner component, and is not an activity predicate and
yet it is one of the most flexible verbs found in English:
\eenumsentence{ She got cold.
She got a car.
She got him a car.
She got into the car.
She got him into the car.
She got him to drive.
She got at the problem.

Directed motion verbs are also rigid in their behavior (have no


manner).

``we point out that manner verbs and result verbs have different [basic] lexical


aspectual classifications: manner verbs ar activities, whereas result verbs are either achievements (e.g. arrive or accomplishments (e.g., transitive break)."
pg 104.

The possible expansions in XX all involve the `expansion' of an activity to yeild varios kinds


of accomplishments.  ...Accomplishments have fully lexically-specified representations.  They
cannot take on activity readings without eliminating a lexically-specified
component of verb meaning (the resulting state)." pg 105

Canonical Realization Rules pg 109:

\eenumsentence{ manne

Semantic bleaching the news broke do involve non-monotonic changes in meaning.


[BUT: Pat mended/washed/dried: result; *Pat rubbed/smeared/caressed.: manner]

Manner verbs are activities, result verbs are achievements or
accomplishments.

Variations in verb meaning all involve the ``expansion" of


an activity to yield various kinds of accomplishments (pg 10).

The addition of the resulting state is signalled by an element in the


syntax: verb meaning is built up incrementally.  (except
in the case of semantic bleaching (metaphorical uses))

They propose that nothing can be added to accomplishments because


they are the most complex they can be to start with.

Both the projectionist and the constructional approaches


recognize a distinction between the "structural" (grammatically
relevant/event structure)
and the "idiosyncratic" aspects of meaning.  (cite a ms
by Grimshaw 1994).

UG: provides an inventory of basic lexical


semantic templates corresponding to basic
event types.

The set of constants is open-ended.

The ontological type of a constant determines its basic association
with a particular event structure template.  The associations
are specified via ! Canonical Realization Rules: minimal
elemetns of meaning in the constants
must be given syntactic expression.  These rules also insure
monotonicity.

Constants maybe modifiers of predicates or arguments of predicates


in the templates.
The constant determines the basic number of participants (see also
Goldberg 1995).

Canonical Realization Rules: associates a manner constant with an


activity template.  Either one or two arguments can be expressed.
(the agent argument is hte only one that is absolutely required).

Break requires a DO in every context: it has two subevents in its event


structure

Subdivision among verbs of change of state:

\medskip

! Free Composition:  Event structure templates may be freely


augmented up to other possible templates.

\medskip


! Subevent Identification: each subevent in the event structure template
must be identified by a predicate in the syntax.
(cite Grimshaw and Vikner 1993 and Van Hout here)

! Argument Realization:

there msut be at least one argument XP in the syntax per subevent in the event structure template
(objects can be dropped from activity predicates if the object
is prototypical (Brisson 1994)
each arguemnt XP in the sytnax must be associated with an identified
subevent in the event structure template

:elements in teh semantic representation must be syntactically


encoded.

\medskip


[Q: what about topic drop langs?]

\medskip


They claim to predict:
{*Phil swept into the corner
because there is no argumetn XP in the syntax corresponding
to the second subevent.  [But:
{Phil spit/bled into the corner
are fine]

They also predict:


{*kelly broke the dishes off the table
because the template associated with break
cannot be augmentd further, no other achieved state or location
can be added to a sentence with break.
{Kelly broek the dishes to pieces
is cited and it is said that  the resultative phrase further
specifies the change already lexicalized into verb.

[
However:


\eenumsentence{ The butcher sliced the salami onto the wax paper.
Joey clumped his potatoes into the middle of his plate.
Joey grated the cheese on to a serving plate.
Sam shredded the papers into the garbage pail. (Goldberg 1005: 171)
]

*!*Hovav, Malka Rappaport and Beth Levin. Two types of Derived


Accomplishments.  Bar Ilan and Northwestern Universities. 1996.
LFG workshop Grenoble.
FOR FARRELL (From Adele)  8/21/96

\smallskip

There does seem to be a subclass of resultatives involving
verbs of surface contact that has the interesting
generalizations they note, but it's not clear you need to
posit two different ways of deriving the two cases.
Instead, the differences seem to follow from the differences
in verb meaning, as combined with two closely related constructions
(the caused-motion and the resultative).

\medskip

! Type 1 :
\eenumsentence{ The dog barked the neighbor awake.
The phone rang me out of my slumber
Sleep your wrinkles away.
! Claim: these are ! syntactically formed (although, they note, maybe
not, but they are at least different than the ones below pg 9).

They suggest that these result from a


syntactic process of complex V0 formation involving an activity verb
and result phrase.

They claim that bark awake does not involve direct causation


and is not a possible simplex lexical item [I would say that it
is direct causation--you don't imagine a scene in which the dogs
startle the cat, which then jumps into the sleeper's bed.  But barking
awake is not name-worthy enough; if the situation
became conventionalized, we could use a single word for it].

\bigskip


! Type 2:
\eenumsentence{ Tracy washed the soap out of her shirt.
Pat rubbed the oil into the wood.
Terry swept the leaves off the sidewalk.
! Claim: these are ! lexically derived via template augmentation from
activities; two different but related senses a la prototype theory;
pure surface contact is assumed to be basic.

\smallskip

It seems that the differences between the two types of examples
in (1) and (2) are actually attributed directly to differences
in verb meaning, without the need to argue that the two types
of examples are generated in completely different ways.

L and R's initial observations about the expressions of Type 1:

the verbs are basically intransitive or can appear intransitively
the intransitive meaning of the verb is preserved
the range of result phrases is broad

Initial observations about the expressions of Type 2:

the verbs are verbs of surface contact
the surface argument is normally expressed as an object,
but can be expressed  as an oblique with a "non-thematic" theme
as DO
the result phrase codes a directional

Starting with the verbs of the type 2 class, we find


transitive verbs of surface contact which may or may not
appear intransitively.  In the following, obligatorily
expressed arguments appear in bold face:

wash $<$ ! washer, washed-surface $>$

wipe $<$ ! wiper, wiped-surface $>$

rub $<$ ! rubber, rubbed-surface $>$

sweep $<$ ! sweeper, moved-dust, swept-surface $>$
(sweeping seems to imply some dust argument, even if that argument
is not expressed)

The "surface" argument in each of these verbs is compatible


with being treated either as an affected entity (patient) or
as a location, because of its inherent semantics as a surface.
In fact it seems to be generally true that arguments which code
surfaces or containers can appear either as DO patients or
as locational obliques:
{Pat loaded the truck with hay. \\
Pat loaded the hay onto the truck.

Because of the fact that surfaces make good locations,


these verbs are able to naturally combine with the
"caused-motion" construction, an extension of simple caused-motion
expressions such as those in \ex{1.
\eenumsentence{ Pat put the book on the table.
Pat threw the garbage out the window.
In this case, the surface-argument "fuses" with the locational argument,
and the construction supplies a theme argument.
Since the meaning of the construction is "cause-move" and the surface
argument appears as the resulting location, the full expression can
mean "cause to move into/out of/off of surface."

Thus observations 2 and 3 follow from observation 1, that verbs


of surface contact are involved.

\bigskip

Similarly, observations 2 and 3 about type \#1 expressions are
captured by observation 1: the fact that the verb has an
intransitive use:

ring $<$ ! ringer $>$

sleep $<$ ! sleeper $>$

bark $<$ ! barker $>$

This fact captures the observation that the verb is interpreted
with its intransitive meaning (observation 2).

These verbs can either combine with the caused-motion construction


described above or with the resultative construction:
\eenumsentence{ The dog barked the neighbors awake (resultative)
The dog barked the cat into the tree. (caused-motion)
In either
case, both the theme argument and the result argument must be added
by the construction, since neither is supplied by the verb.
Because the result phrase is added by the construction and not
the verb, it is only constrained by the construction: thus it
can be either a directional (if the caused-motion construction
is used) or a general resultative (if the resultative construction
is involved).  This gives us observation 3.

\bigskip


Further observations noted by Levin and Rappaport

the result phrase is more tightly connected semantically in type 2


type 2 expressions more easily admit of paraphrases of
the middle construction
type 1 expressions are more rare cross-linguistically

Because the result argument in the expressions of type 2


is an argument of the verb, the verb and result phrase are
more tightly connected semantically than the verb and result
phrase in expressions of type 1. 
While middle constructions are not unavailable for expressions
of type 1:
{Those tires drive bald easily.
the close semantic relationship of verb and result in type 2
expressions acts to greatly facilitates the formation of
middles (see pgs 183-186 of my book for discussion
of why middles are hard to get for type 1 cases).

Finally, the fact that type 2 resultatives are common cross-linguistically


follows from a combination of facts.  First, the caused-motion
construction, being directly based on simple expressions such
as those in XXX above, is ubiquitous in languages around the world.
Secondly, the verbs of surface contact make exceptional candidates
for appearance in the caused motion construction since they entail
in their basic senses that an agent acts on a surface. Our world
knowledge tells us that it is typical for agents to act on surfaces
such that something is caused to move on or off the surface;

\bigskip


\smallskip

"The existence of two distinct types of accomplishments is not easily captured ...[if] event structure is directly read off of the syntax: there is no


obvious way to distinguish between a result phrase
added lexically and a syntactically-added once since they both
receive the same syntactic encoding." pg 10

But this is misleading. It is necessary


to realize that phrasal patterns can be paired with more than one
semantic structure:  formal constructions can
be ambiguous/polysemous just
like words can.  In order to figure out just which form-meaning
is relevant, you have to pay attention to the verbs (or verb classes).
E.g., She considered him a fool, She made him a fool, and She
gave him a fool may all involve the same ternary structure, but there
is more than one construction involved).

[end concise notes]


\bigskip
[rewording of the same]

The basic difference between this account and my CG account


is that two classes of resultatives are posited which are
accounted for in completely different ways (one lexically
by "augmentation"; the other syntactically).

\smallskip

The main argument for this seems to be:

\smallskip

"The existence of two distinct types of accomplishments is not easily captured ...[if] event structure is directly read off of the syntax: there is no
obvious way to distinguish between a result phrase
added lexically and a syntactically-added once since they both
receive the same syntactic encoding." pg 10
[But here they, and others like Hoekstra who have misled them, need
to realize that syntactic structures can be paired with more than one
semantic structure:  formal constructions can
be ambiguous/polysemous just
like words can.  In order to figure out just which form-meaning
is relevant, you have to pay attention to the verbs (or verb classes).
E.g., She considered him a fool, She made him a fool, and She
gave him a fool may all involve the same ternary structure, but there
is more than one construction involved).

There does seem to be a subclass of resultatives involving


verbs of surface contact that has the interesting
generalizations they note, but it's not clear it's entirely
different from the other cases.  Could be another nice instance with
which
to talk about different but related constructions, related via
a hierarchy.]

\medskip


! INTRO:

Two aspects of verb meaning: structural (syntactically relevant);


idiosyncratic = core = constant = "name" determinate.

Two types of building blocks of verb meaning: a basic stock of


lexical semantic templates ("Lexical event structure templates")
a la Dowty/Vendler; secondly an open-ended set of constants encoding "core" verb meanings.

The constant determines the minimal number of args (cites me and


Van Hout here).

\bigskip


! Type 1:
Verbs are basically intransitive, or independently allow
objects to be omitted. Meaning of the verb in intransitive use is
preserved (contrast with Tracy washed which has reflexive
interpretation missing in \ex{2 below); they do not occur in Hebrew,
French or Italian; do not appear with two telicizing phrases
EXAMPLES:
\eenumsentence{ The dog barked the neighbor awake.
The phone rang me out of my slumber
Sleep your wrinkles away.
! Claim: these are ! syntactically formed (although, they note, maybe
not, but they are at least different than the ones below pg 9).

They suggest that these result from a


syntactic process of complex V0 formation involving an activity verb
and result phrase.

They claim that bark awake does not involve direct causation


and is not a possible simplex lexical item [I would say that it
is direct causation--you don't imagine a scene in which the dogs
startle the cat, which then jumps into the sleeper's bed.  But barking
awake is not name-worthy enough; if the situation
became conventionalized, we could use a single word for it].

\smallskip

[It seems that the differences between the two types of examples
in (1) and (2) are actually attributed directly to differnces
in verb meaning, without the need to argue that the two types
of examples are generated in completely different ways.

L and R's initial observations about the expressions of Type 1:

the verbs are basically intransitive or can appear intransitively
the intransitive meaning of the verb is preserved
the range of result phrases is braod

Initial observations about the expressions of Type 2:

the verbs are verbs of surface contact
the surface argument is normally expressed as an object,
but can be expressed  as an oblique with a "non-thematic" theme
as DO
the result phrase codes a directional

Starting with the verbs of the type 2 class, we find


transitive verbs of surface contact which may or may not
appear intransitively.  In the following, obligatorily
expressed arguments appear in bold face:

wash $<$ ! washer, washed-surface $>$

wipe $<$ ! wiper, wiped-surface $>$

rub $<$ ! rubber, rubbed-surface $>$

sweep $<$ ! sweeper, moved-dust, swept-surface $>$
(sweeping seems to imply some dust argument, even if that argument
is not expressed)

The "surface" argument in each of these verbs is compatible


with being treated either as an affected entity (patient) or
as a location, because of its inherent semantics as a surface.
In fact it seems to be generally true that arguments which code
surfaces or containers can appear either as DO patients or
as locational obliques:
{Pat loaded the truck with hay. \\
Pat loaded the hay onto the truck.

Because of the fact that surfaces make good locations,


these verbs are able to naturally combine with the
"caused-motion" construction, an extension of simple caused-motion
expressions such as those in \ex{1.
\eenumsentence{ Pat put the book on the table.
Pat threw the garbage out the window.
In this case, the surface-argument "fuses" with the locational argument,
and the construction supplies a theme argument.
Since the meaning of the construction is "cause-move" and the surface
argument appears as the resulting location, the full expression can
mean "cause to move into/out of/off of surface."

Thus observations 2 and 3 follow from observation 1, that verbs


of surface contact are involved.

\bigskip

Similarly, observations 2 and 3 about type \#1 expressions are
captured by observation 1: the fact that the verb has an
intransitive use:

ring $<$ ! ringer $>$

sleep $<$ ! sleeper $>$

bark $<$ ! barker $>$

This fact captures the observation that the verb is interpreted
with its intransitive meaning (observation 2).

These verbs can either combine with the caused-motion construction


described above or with the resultative construction:
\eenumsentence{ The dog barked the neighbors awake (resultative)
The dog barked the cat into the tree. (caused-motion)
In either
case, both the theme argument and the result argument must be added
by the construction, since neither is supplied by the verb.
Because the result phrase is added by the construction and not
the verb, it is only constrained by the construction: thus it
can be either a directional (if the caused-motion construction
is used) or a general resultative (if the resultative cosntruction
is involved).

Further observations noted by Levin and Rappaport

the result phrase is more tightly connected semantically in type 2
type 2 expressions more easily admit of paraphrases of
the middle construction
type 1 expressions are more rare cross-linguistically

Because the result argument in the expressions of type 2


is an argument of the verb, the verb and result phrase are
more tightly connected semantically than the verb and result
phrase in expressions of type 1. 
While middle constructions are not unavailable for expressions
of type 1:
{Those tires drive bald easily.
the close semantic relationship of verb and result in type 2
expressions acts to greatly facilitates the formation of
middles (see pgs 183-186 of my book for discussion

tải về 302.24 Kb.

Chia sẻ với bạn bè của bạn:
1   2   3   4   5   6   7   8   9   10




Cơ sở dữ liệu được bảo vệ bởi bản quyền ©hocday.com 2024
được sử dụng cho việc quản lý

    Quê hương