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

part (ASK-QUESTION is directed toward chin; GIVE is directed

tải về 302.24 Kb.
Chuyển đổi dữ liệu11.06.2018
Kích302.24 Kb.
1   2   3   4   5   6   7   8   9   10
part (ASK-QUESTION is directed toward chin; GIVE is directed
toward chest).  Location of the hand is not fixed: depends on
where referent is.

Morphemic solutions:

unlimited number of lexical units

But there is no way to list them because their number is indeterminate.

(if there were 10 different referents in discourse, PRO could
refer to any of the 10 by pointing to 10 distinct locations).

one lexically fixed element with indepterminate form

But, Liddell claims this is inconsistent with our conception of
what morphemes are.  Comparison with replicative morpheme of
Thai pg 25.

Location, unlike reduplication, is not dependent on any linguistic

features or category.

Liddell's soln: handshapes, certain aspects of the hand's orientation

and movement are lexically specified, but there are no linguistic
features specifying the location the hands are directed toward.
"the hands are directed toward the speicif part of the referent's
body in Real Space by non-discrete gestural means.  An indicating
verb that moves betwen its subject
nd object is produced from a verb root combined with two
nondiscrete locations to
account for the direction of the movement." pg 26

Note that aspects of the verb's use are clearly grammatically

specified: see chin vs chest above; Padden notes that only subject may
be left unspecified in subj-obj "agreement" verbs; also certain
lexical idiosyncracies:
BAWL-OUT and FLIRT can be directed twoard addressee, but only
BAWL-OUT can be directed toward signer to
indicate signer is the object of the verb.

Therefore PRO directed at addressee is not grammatical "you," but

is just PRO without specifying the location in Real Space to
which the sign is directed.

So, indicating verbs  do not demonstrate "agreement" as Padden 1988 had proposed., according to L.

L. follows Klima and Bellugi (1982) who argue that when a person
is phyically present, grammatical reference to that person is
deictic not anaphoric.
(deixis: those features of language which refer directly to
the personal, temporal or locational characteristics of the
situation within which an utterance takes palce, whose meaning
is thus relative to that situation." pg 27)

\subsection{Surrogate Space

surrogate: full-sized invisible entity used during "role-shifting"

Surrogate Space: a mental space in which aspects of  events are grounded in

the physical space that includes the signer.


Ex: PRO-1, WHat?
I went, "what?"


Ex: Sue, What?
Sue went, "what?"


During PRO-1, signer's head position and eyegaze are straight ahead.

During question, signer's head position and eyegaze directed upward and

to the right toward an imaginary addressee.

Two surragotes:  the invisible one that the signer directs eyegaze

at; but ALSO, the signer herself imagining either herself at a
different time and place or Sue at a different time and place.

The pure memory of what happened is a non-grounded mental space:

a memory or recollection.

But the surrogate space itself, which is directly responsible for

the signing above is grounded: the surrogate is
treated as physicallly present.

"It is the location of the surrogate itself, notsome aspect of

linguistic elements within the clause containing either PRO or
an indicating verb that serves as the basis of the
way the sign is directed." pg 31.

\subsection{Token Space

establishing an index (creating a 3-D token):

BOY LOC-ATx (there is a boy at this place)

GIRL LOC-ATx (there is a girl at that palce)

Ty-Kick-Tx (She kicked him)

(LOC-ATx: short downward movement of 1 hadnshape twoard and held at
location x)

Liddell: it is a token, not an abstract spatial locus that

is pointed to.


Token space is also grounded.  Differs from Surrogate space in
that: token is not normally sized, involves a featureless region,
appears within the smaller domain of signing space and plays only
a 3rd person role in discourse. (pg 34)


Real space is our conception of our currrent, perceivalbe phsyical
environment.  NO one would accept a propsal that the physical entities
around us are part of English, ASL or any other language." pg 34

Surrogate Space and Token Space are similar.

Liddell concludes:

neither Token Space, Surrogate Space nor Real Space contains
any linguistic elements.

\subsection{Abstract Space

C classifer, expaiining that it shows her language and culture;
holds it near chest.

Describes two people approaching one another, with 1-cl handshapes,

one near the chest and one to the right; handshapes are then replaced
with C classifier handshapes representing two
containers of cultural
behaviors facing one another.

Abstract concepts are therefore treated as if they occupy grounded


Signers can look at different sides of an abstract concept, pick it

up, etc.

"the signer is directly referring to elements within Surrogate Space or

Token Space in a way which parallels reference to elements
within Real Space." pg 39.

Signers are not only apble to point to the conceptual

entities in space but are able to manipulate them.
*!*Liddell, Scott K. 1995. Spatial Reprsentations in Discourse:
comparing spoken and signed languages. Ms. Gallaudet U.

Liddell argues agains the view that spatial reprsentations are

part of sytnax; instead, he argues that spatial representations
are part of a topographically organized space.  Spatial representations
contain no linguistic elements and "thus are not part of what
linguists typically mean by the grammar of a language."

Analogy is made to the use of props in spoken language:

Ok, the coffee cup is my house.  The sugar boal is the apt building
next door and the knife is Fla street.

There is a bicycle right about here.


"This side of my house gets direct sunlight in the morning"

(can be said while pointing to coffee cup)

Notice that although pointing to the coffee cup identifies

the house, mentioning the cup does not:

*This spring we plan to add a new bedroom to the coffee cup.

Signers do not use props like that, but they almost always
use a spatial representation to describe spatial relationships.


1-cl ato-b F-L-A   S-T
Right along here (a to b)  is Fla St.


My home is located here.

Not far from and to the side of the house is a (much taller) apt building

located here(y). pg 6

Bicycle there(z)


\subsection{Mental Spaces

mental spaces are distinct from linguistic structures: they are
structured sets with elements and relations holding
between those elements.

What is remarkable about the spatial representations in signed

languages is that they are 3-D and can be directly pointed at
and manipulated.

trigger (linguistic entity), target(entity in mental space)

\subsection{Analyzing the use of props in spoken lng
Does the sugar boal itself represent the apt building?
Liddell: no the sugar boal is a landmark showing us where the
conceptual apt building is, but the physical sugar boal
is not the same as the conceptual apt building in the
spatial representation.  (It just picks out a 3-D location). pg 16.

Placing props on the table creates a distinct conceptual representation.

The speaker can point to a location high above the top ofthe
sugar bowl and say "from up here on top of the apt building"

Three mental spaces: 1) R: speaker's conception of the neighborhood

being described (non-grounded), 2) M spatial representation of
conceptual houses and apt building, resulting from
playing the coffee cup in position while uttering "this is my
house" (grounded), 3) RS the conception of the
physical table and objects on it.

The elements of M and RS both occupy parts of the same physical


The ID principle does not work straightforwardly (in either direction):

(missing "not" page 17)
\eenumsentence{ *The coffee cup was just remodeled last year.
*Bill lives on the 3rd floor of the sugar bowl
*My house has a lipstick stain on its rim.
*Please put the spoon back in the apt building.

"There is no reason to believe that the 3 mental spaces in FIgure 5

contain any elemnts of the English language." pg 22.

\subsection{Analyzing ASL discourse

Classifier predicates place conceptual entities called tokens
in the space in front of the signer.

It is clear that the tokens are conceptual entities, since they

exist even when the classifiers are not overtly identifying
(they can be referred to again in subsequent discourse).

! Conclusion:"This paper argues that the 'use of space' in

a sign language such as ASL is not conceptually different from the 'use
of space" in a spoken language such as English....[They differ in that]
SIgn languages hav extensiev parts of their grammars devoted to
the construction and use of spatial represntations, while
spoken langauages typcailly do not." pg 26

*!*MacWhinney, Brian. 1982.   Basic Syntactic Processes.

In Language Development, Syntax and SEmantics Vol 1.
Stan A Kuczaj II editor. New Jersey: Lawrence Erlbaum Associates.

! BASIC Strategies:

! Rote : memorized string (word or phrase): semantic analysis
serves as a stimuls to morphological analaysis

Brown, cazden and Bellugi (1968): I'm, that-a, drop-it,

put-it, get-it, want-to, have-to, another-one, what-that
learned by rote: single intonational contour, acquired
as wholes.

other evidence suggestsed (some better than others):

incorrect imitation ('init" to refer to "elephant" after
hearing that's an elephant, isn't it?: R. Clark;

rituals: some formula restricted to contexts

Pronoun errors: I carry you as request to be carried.

Production of whole before its parts: cant',

won't, don't before can, will, do

Production of contraction before expansion

Precocious strings: open the door at 14 months with
only 6-10 word vocab: Peters (1977); Macwhinney:
No, Mommy, I don't want to go to bed; I like it; I love it
before other 2 word utterances.

PHonlogical structure: Peters (1977): some children learn the tune beofre

the words.

Superfluity in Context

Presence of an Unusual Element

Redundancy: have it egg; have it milk, miss it garage

\subsection{ Analogy

Rote cannot explain neologisms, errors, nonce forms; too many

sentence swould have to be memorized, these sentences would be
too long.

this is my wug produced by analogy with this is my


my book analogized to my toy

\subsection{ Combination

The pattern develops a life of its own apart form the lexical items
from which it developed: no sharp line bewteen analogy and combination

my book and my toy abstracted to my X

\subsection{Evidence for Item Based Patterns

! Ordering in short strings

Ingram (1979): most of the sentences in his two corpora seemed
to use some fom of item based pattern:

here, my , hi, a, that, want, its, what before object words.

this, it used after action words

I before action words

! Children's failure toogeneralize: Kuczaj and Brannick (1979): aux aplies to wat before how long (what are you doing?

how long you (are) staying?

Bowerman (1976): want +X; more +X, but didn't generalize to

open +X, close+X, bite+X, etc.

Control of disontinuous morphemes

Item-phrase orderings

Competitive orderings: A+B and C+B: what's a child to do?  In fact,

children learn order of adjectives relatively late.

Strings not atributable tonalogy

The nonoccurence of certain types of errors


Simple analogy is not enough: novel operators  acn be extended

Evidence for general actor action schema

Across the board changes in gruops of lexcal items:
Bellugi 1971: movement of aux to sentence initial position
in Y/N Qs occurs around teh same time for all auxs

*!*MacWHinney, Brian. 1997 April. Phone conversation

item based patterns = MIke's verb islands

believes strongly in verb by verb learning for argument structures, but

he doesn't assume Wh- constructions are verb based

confirmed core of word: things that are known to be true of a word

Competition Model was basically designed to account for subject
facts as a test case for Grammatical Relations (but assumed
all verbs act the same which he now thinks is an oversimplification).
He shares Bill M's view of how GRs are acquired, sounds like.

*!*Mandelblit, Nili. 1997.  Creativity and Schematicity in

Grammar and Translation: The Cognitive
Operation of Blending.  UCSD Diss.

A single blending schema is associated with each binyan to

characterize a large percentage of teh verbs that occur in the binyan.

from Berman (1975):

nif'al: passive, middle, basic (and a few reciprocals)
pu'al: pssive
huf'al: passive
hitpa'el: middle reflexive (and a fe reciprocal, inchoative,
iterative and basic

Innovative forms from Berman and Sagi (1981)

\eenumsentence{ hisxi: hif'il form of saxa ('to swim'): the child
refers to her father who is helping or teaching her to swim.
heshin an hif'il form of yashan "to sleep": teh child
refers to the event of putting someone to sleep.
Other novel examples come from slang.

Hebrew double object construction: NP N et NP et NP:

rec is volitional, transfer is successful.

Rubinstein (1976): discusses examples of 2 arg tansitive verbs in Biblilical

Heb whos semantics "shifts" when they appear in the Acc-Dative
argument structure: a component of "causation of possession"
is added to the core meaning of the verb:
\eenumsentence{   leharim "to lift"
means contribute or offer when used in teh Acc-Dative syntactic pattern (same form of the verb).

To Nili:

I wanted to send you a follow up message about my questions
during your defense.  I'd like to stress that I really like
the idea of treating the Hebrew binyanim as blends.

It also makes a lot of sense that each binyanim offers

a different perspective on the scene.  Your slide with the
blue and orange lines is really nice (although I would interpret it
a little differently-see below)

My question really concerns the nature of the two input spaces.

I would have thought that the
Hebrew verbs were a blend of the root with the binyan, each with
its own conceptual semantics.  Or, given that the phrasal construction
plays a role, a blend of the root (with its arguments)
with the binyan+phrasal construction.

It's still not entirely clear to me what the nature of the two

input spaces you propose is.  On the one hand, you say
that Input 2 is the very abstract: Event1 causes Event2,
but at the same time, you'd like Input2 to have the specific lexical
items in it.

Are there lexical items associated with act1, act2, CAUSE and all of the

participants even if they don't end up getting expressed?
I wouldn't think so, but if not, isn't there a sense in which
what you are blending is really the conceptual structure associated with
the binyan+construction with the conceptual structure of the verb that
is expressed and its arguements?
If so, then it's really exactly parallel to the way I analyzed
the English caused-motion and I guess F&T's analysis is really the same
as mine.

I think the diagram you had of the various binyanim with blue

and orange lines to various partipants and events is an elegant
way of capturing the *binyanim*'s semantics (or binyan+ phrasal
construction's semantics).  Each binyanim
offers a different perspective on a causative scene.
The binyanim gives instructions as to what relation the verb root can
bear, just like the English cases (the ditransitive allows instance,
means, but *not* result; the transitive allows result, instance,
instrument, means, etc etc).

"Event1 CAUSES Event2" allows for a nice way of specifying the

binyanim's perspective, since it shows what is common across binyanim,
and therefore allows you to highlight was is different among the
binyanim. But is it clear that speakers have absracted out this
generalization and in any real sense conceptualize this most abstract
causation event without perspective (the perspective is attached to
particular binyanim).

Then, I would say, the binyanim gets integrated (blended) with

a verb root yielding the richer semantics.

--"NP V et NP" looks like V, et, and NP are all sisters. 

You want to make it clear
that "et" is part of the NP.  "NPet"  (like "PPto") may be better

--the inventories of constructions *are* definitely different

cross-linguistically i.  E.g. there 's no ditrans in Fr and no impersonal
passive in English.

--I also was after the basic means of expression.  The sneeze examples

were useful to make the point, but I was arguing that more
ordinary cases like put and throw work the same way.

The reason I didn't use the transitive construction was that

it is MUCh harder to make the point: since the construction itself
has such a huge range of meanings, the verb really does seem to play
the determinate role in choosing a meaning. 
When you give the example of S. kissed D, all the spaces
are isomorphic: given that, it's not clear that you *need*
a blend.  I understand you want to say a blend is involved,
and I agree, but it's not a good case to make the point with.

Also if you want to claim NP V NP is associated with "agent

acts-on patient" you have to address how it is that
"It rained cats and dogs" "It is a stone" and
"She became a movie star" fit. (if you instead use Subj V Obj
those particular problems go away--you are still left with "A saw
B" but that is arguably based onthe active semantics).

--I think John's question was in part why it was that lines

should go to act1 for "kill" since "kill" (unlike "strangle") doesn't
specify what the agent's activity is.

More generally, I was a little unclear about what the meaning of

pa'al is.  At one point you seem to  say that it codes lexical
causatives (like kill), but isn't it also the form used to
code simple, non-causative wash (requiring the hifil for causative

*!*Matsumoto, Yo. 1996.  On the Nature of Compound Verbal

Nouns in Japanese.  Presentation at Stanford U July 17, 1996.

Considers cases like:

Lexical compounds: functionally monoclausal
{ Taro wa Eigo o KENKYUU-KAISHI suru \\
Taro top English acc research-beginning do \\
Taro will begin research on English.

Syntactic compounds: functionally bi-clausal:

{ Taro ga Seebu e  RYOKOO-KAISHI-go\\
Taro nom West goal trip-beginning-after \\
After Taro began his trip to the West

*!*Murphy, Gregory L.  1991 Meaning and Concepts In The Psychology

of Word Meanings ed by Paula J Schwanenflugel. Hillsdale: NJ: Erlbaum

Meaning: semantic components of words

Concepts: mental representations of coherent classes of entities.

Putnam (1973, 1975, 1988): meanings are not necessarily in the head:

certain ``lemons" could in actuality turn out to be oranges.

``glass snakes" are really a kind of lizard.

Twin-Earth ``gold"

Linguistic division of labor: we delegate some responsibility to experts

Are there meanings to words in dead languages?

Murhpy: let's look at what is in the head.

Possibility 1: meanings are strictly internal to the linguistic system, they

arenot part of general language of thought used to think about the world.

Feature theory of meaning (e.g. Katz and Fodor 1963)

But in order to relate words to real objects, we need to interpret each of
the features.  But if we need an interpretation for each feature, why not
allow the features to refer directly to the interpretations or concepts?

Concepts with no corresponding meaning?  ``things to do at teh beach when it's

raining"; children's prelinguistic concepts;

meanings that involve more

than one concept (democracy).

Meanings are built out of concepts: word's meaning is constructed by mapping conepts

onto the semantic component of the lexicon.

How to account for the stability of meanings?  Language is a social convention.

Barselou (1987, 1989): meanings (concepts) are not so stable.  Various according to

person's knowledge, recent experience and current context.

*!*Murphy, Gregory L and Jane M ANdrew. 1993.  The Conceptual

Basis of Antonymy and Synonmy in Adjectives.  Journal of Memory and
Language 32, 301-319.

comparison with Miller et al. WordNet: anyonymy is relation of associations, not

conceptualizalion (whereas synonymy is a conceptual relation)

Big and large differ in register

ascend/descend imply a certain elegance and control not in rise/fall.

Adjectival interpretation differs depending on what N is modified:

red hair, pen, book, orange

``The changing significance of the adj. is a produtive part of the intepretation

processes." pg 306.

Subjects were queried for antonym of adj in various contexts: found that subjects

would provide different antonyms.

Does the polysemy exist independently of the combination?  Are the various senses


Polsyemy view: interaction is a matter of selecting a previously existing sense, to

whihc an antonym is already associated.

Conceptual combination view: interpretation of hte adjective is modified when it is

integrated with the N meaning.  Antonym is computed by following rule:


Anyonyms are gradable adjectives that differ solely in one conceptual dimension,
such as that the values of the two adjectives on that dimension are equal
distances in opposite directions from a neutral point.

*!*Ninio, Anat. 1997. MS. Pathbreaking verbs in syntactic

development and the question of prototypical transitivity.
Hewbrew University, Jeruselem Israel. to appear in Linguistics?
presented at the 7th Int'l Congress for the Study of CHild
Language, Istanbul, Turkey July 1996

Longitudinal languge of 15 Hebrew speaking children and

1 English speaking child were considered.  Also corss-sectional
sample of 84 18 month old Hebrew speakers.

Children learn SVO and VO patterns initally in an item by item

way, with a long time lag between first and 2nd verbs.
``They soon begin transferring some more general and abstract

1   2   3   4   5   6   7   8   9   10

Cơ sở dữ liệu được bảo vệ bởi bản quyền ©hocday.com 2019
được sử dụng cho việc quản lý

    Quê hương