http://www.blackwellreference.com/subscriber/uid=532/tocnode?id

14. Empathy and Direct Discourse

14. Empathy and Direct Discourse Perspectives

Perspectives

SUSUMU KUNO

1 Introduction

Introduction

This paper discusses two functional perspectives - the

EMPATHY

PERSPECTIVE

and the

DIRECT

DISCOURSE

PERSPECTIVE

- that are indispensable for the study of syntactic phenomena in natural language. These

perspectives, like other discourse-based perspectives, such as

PRESUPPOSITION

and

THEME

RHEME

, that

interact closely with syntactic constructions, help us distinguish what is non-syntactic from what is
syntactic and guard us from mistakenly identifying as syntactic the effects of non-syntactic factors on
the construction under examination.

2 The Empathy Perspective

2.1 The Empathy Principles

Assume that John and Bill, who share a dormitory suite, had an argument, and John ended up hitting
Bill. A speaker, observing this event, can report it to a third party by uttering (1a-c) or (2a-b), but not
(2c):

(1) a. Then John hit Bill.
b. Then John

hit his

roommate.

c. Then Bill

's roommate hit him

(2) a. Then Bill was hit by John.
b. Then Billj was hit by hisj roommate.
c. ??/*Then John

's roommate was hit by him

These sentences are identical in their logical content, but it is generally felt that they are different
with respect to the speaker's attitude toward the event, or toward the participants of the event. It is
intuitively felt that in (1b), the speaker has taken a perspective that places him/her closer to John than
to Bill, whereas in (1c), the speaker is closer to Bill than to John. The notion of

EMPATHY

was proposed

in Kuno (1975), Kuno and Kaburaki (1977), and Kuno (1987) to formalize this intuitive feeling and
thus to account for the unacceptability of (2c) and many other related phenomena.

The following definitions, assumptions, and hypotheses are in order:

(3) a. E

MPATHY

: Empathy is the speaker's identification, which may vary in degree, with a

person/thing that participates in the event or state that he/she describes in a sentence.
b. Degree of Empathy: The degree of the speaker's empathy with x, E(x), ranges from 0 to 1,
with E(x) = 1 signifying his/her total identification with x and E(x) = 0 signifying a total lack of
identification.

Theoretical Linguistics

Pragmatics

discourse

10.1111/b.9780631225485.2005.00016.x

Subject

Key

Key-

-Topics

Topics

DOI:

Page 1 of 21

14. Empathy and Direct Discourse Perspectives : The Handbook of Pragmatics : Blac...

28.12.2007

http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9780631225485...

c. D

ESCRIPTOR

MPATHY

IERARCHY

: Given descriptor x (e.g.

John

) and another descriptor f(x) that

is dependent upon x (e.g.

John's roommate

), the speaker's empathy with x is greater than that

with f(x): E(x) > E(f(x))
d. S

URFACE

TRUCTURE

MPATHY

IERARCHY

: It is easier for the speaker to empathize with the

referent of the subject than with that of any other NP in the sentence: E(subject) > E(other NPs)
e. T

OPIC

MPATHY

IERARCHY

: Given an event or state that involves A and B such that A is

coreferential with the topic of the present discourse and B is not, it is easier for the speaker to
empathize with A than with B: E(topic) [] E(nontopic)
f. S

PEECH

MPATHY

IERARCHY

: The speaker cannot empathize with someone else more than

with himself/herself:
E(speaker) > E(others)
g. H

UMANNESS

MPATHY

IERARCHY

: It is more difficult for the speaker to empathize with a non-

human animate object than with a human, and more difficult to empathize with an inanimate
object than with an animate object:
E(human) > E(non-human animate) > E(inanimate) h. T

RANSITIVITY

MPATHY

ELATIONSHIPS

Empathy relationships are transitive.
i. B

ONFLICTING

MPATHY

OCI

: A single sentence cannot contain logical conflicts in

empathy relationships.
j. M

ARKEDNESS

RINCIPLE

FOR

ISCOURSE

ULE

IOLATIONS

: Sentences that involve marked (or

intentional) violations of discourse principles are unacceptable. On the other hand, sentences
that involve unmarked (or unintentional) violations of discourse principles go unpenalized and
are acceptable.

The Descriptor Empathy Hierarchy (EH) states that, given two descriptors

John

and

John's

his

roommate

in (1b), the speaker's empathy with John is greater than that with his roommate. The

Surface Structure EH says that, given

John

in subject position and

his roommate

in non-subject

position, the speaker's empathy with John is greater than that with his roommate. Since these two
empathy relationships are consistent, the sentence does not violate the Ban on Conflicting Empathy
Foci, hence the acceptability of the sentence. I will schematize the above two relationships and the
resulting conclusion in the following way:

(4) (1b) Then John

hit his

roommate.

Descriptor EH E(John

) > E(his

roommate=Bill)

Surface Structure EH E(subj=Johni) > E(non-subj=hisi roommate=Bill)
E(John) > E(Bill) [no conflict]

The acceptability of (1c), on the other hand, is accounted for in the following fashion. According to
the Descriptor EH, the use of the descriptor

Bill

to refer to Bill, and of

Bill's roommate

to refer to John

shows that the speaker's empathy with Bill is greater than that with John. However, the Surface
Structure EH says that the speaker's empathy with the referent of the subject (i.e.

Bill's roommate

John) should be greater than that with the referent of the non-subject (i.e.

him

= Bill). Therefore,

there is a logical conflict between these two empathy relationships. This conflict has been created
non-intentionally, however, by placing the agent NP

Bill's roommate

in subject position and the theme

him

(

=Bill

) in object position, as dictated by the subcategorization requirement of the transitive

verb

hit

. That is, there is no intentional/marked violation of the Ban on Conflicting Empathy Foci.

Hence, there is no penalty for the violation. The above explanation is schematically summarized in (5):

(5) Then Bill

's roommate hit him

. (=1c)

Descriptor EH: E(Bill) > E(Bill's roommate = John)
Surface Structure EH: E(subject=Bill's roommate=John) > E(him=Bill) Transitivity: *E(Bill) > E
(John) > E(Bill)
Markedness Principle for Discourse Rule Violations: The above violation is unintentional ⇒ no
penalty

Now we can account for the marginality or unacceptability of (2c):

Page 2 of 21

14. Empathy and Direct Discourse Perspectives : The Handbook of Pragmatics : Blac...

28.12.2007

http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9780631225485...

(6) ??/*Then John

's roommate was hit by him

. (=2c)

Descriptor EH: E(John) > E(John's roommate = Bill)
Surface Structure EH: E(subject = John's roommate = Bill) > E(him=John) Transitivity of
Empathy Relationships: *E(John) > E(Bill) > E(John) [a violation of the Ban on Conflicting
Empathy Foci]
Markedness Principle for Discourse Rule Violations: The above violation has been created by
the speaker's intentional use of a marked construction (i.e. the passive sentence construction)
⇒ a penalty

Note that the Markedness Principle for Discourse Rule Violations penalizes (2c) because the violation
of the Ban on Conflicting Empathy Foci that the sentence contains is intentional. That is, the conflict in
empathy foci has been created by the speaker's intentional use of the passive sentence construction,
which didn't have to be used.

The marginality or unacceptability of sentences with a first-person

-agentive, as in (7b), illustrates

the working of the Speech Act EH:

(7) a. Then I hit John.
Speech Act EH E(speaker=I) > E(John)
Surface Structure EH E(subject=I) > E(non-subject=John)
E(speaker) > E(John) [no conflict]
b. ??/*Then John was hit by me.
Speech Act EH E(speaker=I) > E(John)
Surface Structure EH E(John) > E(me=speaker)
Transitivity: *E(speaker) > E(John) > E(speaker)
Markedness Principle: The above conflict has been created by the speaker's intentional use of
the passive sentence construction ⇒ a penalty

The marginality or unacceptability of discourses such as (8b) illustrates the importance of the role
that the Topic EH plays in the Empathy Perspective:

(8) a. Mary had quite an experience at the party she went to last week. √She slapped a drunken
reporter on the face.
Surface Structure EH E(she=Mary) > E(a drunken reporter)
Topic EH E(she=Mary) > E(a drunken reporter)
E(Mary) > E(a drunken reporter) [no conflict]
b. Mary had quite an experience at the party she went to last night. *A drunken reporter was
slapped on the face by her.
Surface Structure EH E(a drunken reporter) > E(her=Mary)
Topic EH E(she=Mary) > E(a drunken reporter)
Transitivity: *E(a drunken reporter) > E(Mary) > E(a drunken reporter) Markedness Principle:
The above conflict has been created by the speaker's intentional use of the passive sentence
construction ⇒ a penalty

Thus, (8b) is unacceptable because Passivization, which is an optional process, has been used

intentionally to create a conflict in empathy relationships.

Observe now the next discourse fragments:

(9) Mary had quite an experience at the party she went to last night.
a. She met a

New York Times

reporter.

b. *A

New York Times

reporter met her.

c. A

New York Times

reporter asked her about her occupation.

Both (9b) and (9c) have a non-topic NP in subject position and a topic NP in non-subject position, but
while there is nothing wrong with (9c), (9b) is totally unacceptable in the given context. The Empathy

Page 3 of 21

14. Empathy and Direct Discourse Perspectives : The Handbook of Pragmatics : Blac...

28.12.2007

http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9780631225485...

Principle accounts for this fact by attributing it to the fact that

meet

is a reciprocal verb. If two people

(say, John and Mary) met, the speaker has the following four alternatives in reporting this event using

meet:

(10) a. John and Mary met.
b. Mary and John met.
c. John met Mary.
d. Mary met John.

The relationship that (10c, d) have with (10a, b) is similar to the one that passive sentences have with
their active counterparts in that they involve the speaker's intentional choice of

John

and

Mary

subject position, and

Mary

and

John

in non-subject position in (10c, d), respectively. That is, (10c, d)

can be characterized as the passive versions of (10a, b) and thus constitute marked constructions. In
contrast, if John asks Mary about her occupation, the subcategorization requirement of the verb

ask

automatically places the agent NP

John

in subject position and the theme NP

Mary

in object position.

This difference between

meet

and

ask

accounts for the contrast between the unacceptable (9b) and

the acceptable (9c). More schematically, the acceptability status of the three sentences in (9) is
accounted for in the following fashion:

(11)

Observe next the following sentences:

(12) a. John told Mary that Jane was seriously sick.
b. Mary heard from John that Jane was seriously sick.

a. She met a

New York Times

reporter. (=9a)

Topic EH

E(she=Mary) > E(a

Times

reporter)

Surface Structure EH

E(she=Mary) > E(a NY
Times reporter)

E(Mary) > E(a

NY Times

reporter) [no conflict]

b. *A

New York Times

reporter met her. (=9b)

Topic EH

E(she=Mary) > E(a

Times

reporter)

Surface Structure EH

E(a NY Times reporter)
> E(her=Mary)

Transitivity: *E(Mary) > E(a

NY Times

reporter) > E(Mary)

Markedness Principle: The above conflict has been created by

the speaker's intentional choice of a non-topic NP as subject of

meet

 a penalty

c. A

New York Times

reporter asked her about her occupation.

(=9c)

Topic EH

E(she=Mary) > E(a

Times

reporter)

Surface Structure EH

E(a NY Times reporter)
> E(her=Mary)

Transitivity: *E(Mary) > E(a

NY Times

reporter) > E(Mary)

Markedness Principle for Discourse Rule Violations: The above

violation is unintentional ⇒ no penalty

Page 4 of 21

14. Empathy and Direct Discourse Perspectives : The Handbook of Pragmatics : Blac...

28.12.2007

http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9780631225485...

(13) a. John sent Mary a Valentine's Day present.
b. Mary received from John a Valentine's Day present.

Hear from

in (12b) and

receive from

in (13b) are marked verbs in the sense that they place non-agent

NPs in subject position and agent NPs in non-subject position. That is, they are like passive verbs in
that they represent the speaker's intentional choice of non-agent NPs in subject position. This fact
accounts for the acceptability status of the following sentences:

(14) a. I told Mary that Jane was seriously sick.
b. ??Mary heard from me that Jane was seriously sick.

(15) a. I sent Mary a Valentine's Day present.
b. ??Mary received from me a Valentine's Day present.

(14b) and (15b) are marginal out of context because they contain a conflict in empathy relationships
just as (7b) does, which conflict has been created by the speaker's use of the marked verbs

hear from

and

receive from

Now, we can account for the acceptability status of the following sentences:

(16) a. John

told Mary

what she

had told him

two days before.

b. John

told Mary

what he

had heard from her

two days before.

c. Mary

heard from John

what she

had told him

two days before.

d. ??Mary

heard from John

what he

had heard from her

two days before.

I give below the empathy relationships represented by the main clause and the embedded clause of
each of (16a-d). I use the notation “E(x) m> E(y)” if the empathy relationship is derived from the
Surface Structure EH due to the use of marked patterns (e.g. the passive construction and special
verbs such as

meet, hear from

, and

receive from

(17) a. John

told Mary

what she

had told him

two days before. (=16a) Surface Structure EH

Main Clause: E(John) > E(Mary)
Embedded Clause: E(she=Mary) > E(him=John)
Transitivity: *E(John) > E(Mary) > E(John)
Markedness Principle: No marked patterns are used  no penalty
b. John

told Mary

what he

had heard from her

two days before. (=16b) Surface Structure EH

Main Clause: E(John) > E(Mary)
Embedded Clause: E(he=John) m> E(her=Mary) No conflict
c. Mary

heard from John

what she

had told him

two days before. (=16c) Surface Structure EH

Main Clause: E(Mary) m> E(John)
Embedded Clause: E(she=Mary) > E(him=John) No conflict
d. ??Mary

heard from John

what he

had heard from her

two days before. (=16d)

Surface Structure EH
Main Clause: E(Mary) m> E(John)
Embedded Clause: E(he=John) m> E(her=Mary)
Transitivity: *E(Mary) m> E(John) m> E(Mary)
Markedness Principle: The above conflict has been created by the intentional use of the marked
verb

hear from

 penalty

The Markedness Principle for Discourse Rule Violations states that conflicts in empathy relationships
attributable to the speaker's intentional use of marked constructions result in unacceptability, but
when there is a good reason for using such a construction, an ensuing conflict in empathy
relationships does not result in a penalty. Observe, for example, the following exchange:

(18) A: John says he hasn't met you before.
B: That's not correct. He met me last year at Mary's party.

Page 5 of 21

14. Empathy and Direct Discourse Perspectives : The Handbook of Pragmatics : Blac...

28.12.2007

http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9780631225485...

Empathy Relationships for “He met me last year …”
Speech Act EH E(speaker=me) > E(non-speaker=he)
Surface Structure EH E(subj=he) m> E(non-subj=me)
Transitivity: *E(I) > E(he) m> E(I)

The sentence

he met me last year at Mary's party

in Speaker B's answer involves the use of the marked

verb

meet

, and therefore, there should be a penalty for the conflict in the empathy relationships that

the sentence contains. In spite of this fact, the sentence is perfectly acceptable in the given context. I
attribute this to the application of the following principle:

(19) T

ORRECTIVE

ENTENCE

ATTERN

EQUIREMENT

: In correcting a portion of a sentence uttered

by someone else maintain the same sentence pattern and change only that portion of the
sentence that needs to be corrected, together with necessary tense and personal pronoun
switches.

Speaker B wants to correct Speaker A's remark that “he [John] hasn't met you before.” In accordance
with the Corrective Sentence Pattern Requirement, Speaker B maintains the same sentence pattern,
keeping

(

John

) in subject position, but switching

you

in object position to

referring to the

speaker himself/herself. Thus, the conflict in empathy relationships contained in Speaker B's answer
in (18) is not by design, but is required by the Corrective Sentence Pattern Requirement. Thus there is
no intentionality in the conflict, and hence there is no penalty for the conflict.

There is another discourse principle that makes sentences involving empathy relationship conflicts
acceptable. Observe first the following sentences:

(20) a. John read

War and Peace

last night.

b. ??

/*War and Peace

was read by John last night.

The unacceptability of (20b) is due to the fact that an inanimate NP is in subject position in a passive
sentence construction:

(21) Empathy Relationships for (20b)
Humanness EH E(John) > E(

War and Peace

)

Surface Structure EH E(War and Peace) m> E(John)
Transitivity: *E(John) > E(

War and Peace

) m> E(John)

Markedness Principle: The above conflict has been created by the speaker's intentional use of
the passive sentence construction  a penalty

The above accounting for the unacceptability of (20b) predicts that the following sentences should be
as marginal or unacceptable as (20b), but these sentences are perfectly acceptable:

(22) a.

War and Peace

was written by Tolstoy.

War and Peace

has been read by millions of people all over the world.

War and Peace

has been read even by Bill.

The difference between the unacceptable (20b) and the acceptable (22a-c) lies in the fact that while
the latter sentences characterize what kind of book

War and Peace

is, the former does not have such a

characterizational property. That is, the fact that Tolstoy wrote

War and Peace

gives a robust

characterization of the novel. The fact that millions and millions of people all over the world have read
the novel says what kind of book it is. Likewise, the fact that even Bill -apparently someone who
doesn't ordinarily read books - has read the book implicates that many other people have read it, and
characterizes what kind of book it is. In contrast, (20b) cannot be interpreted as a characterizational
sentence: a single event of John's reading the novel doesn't characterize what kind of book it is.

Given that passive sentences with inanimate subjects and human by-agentives are acceptable if they
robustly characterize the referents of the subject NPs, we still need to explain why the empathy

Page 6 of 21

14. Empathy and Direct Discourse Perspectives : The Handbook of Pragmatics : Blac...

28.12.2007

http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9780631225485...

relationship conflicts created by the use of the marked sentence pattern (i.e. Passivization) that (22a-
c) contain do not make them unacceptable. It seems that this is due to the fact that sentences that
characterize or define the referent of an NP are the most felicitous when that NP is placed in subject
position. For example, observe the following sentences:

(23) a. Whales are mammals.
b. Mammals include whales.

The above sentences are logically identical, but they are different in respect to what they characterize
or define. That is, (23a) is a sentence that characterizes or defines whales, whereas (23b) is a
sentence that characterizes or defines mammals. This observation leads to the following hypothesis:

(24) S

UBJECT

REFERENCE

FOR

HARACTERIZING

ENTENCES

: Sentences that characterize/define X are

most felicitous if X is placed in subject position. (Kuno 1990: 50)

We can now account for the acceptability of (22a-c) by saying that the empathy relationship conflicts
that these sentences contain have been forced by the Subject Preference for Characterizing Sentences,
and thus have not been created intentionally by the speaker.

There is another discourse phenomenon that seems to be explained in a principled way only by the
Empathy Perspective. When two NPs are conjoined, they must be arranged in a fixed order if the
descriptor for one NP is dependent on the descriptor for the other. Observe, for example, the
following sentences:

(25) a. John

and his

brother went to Paris.

John

's brother and he

/John

went to Paris.

This fact can be accounted for by hypothesizing the following empathy principle:

(26) W

ORD

RDER

MPATHY

IERARCHY

: It is easier for the speaker to empathize with the referent

of a left-hand NP in a coordinate NP structure than with that of a right-hand NP.
E(left-hand NP) > E(right-hand NP)

Considering the fact the left-hand position in a coordinate structure is more prominent than the
right-hand position, and considering the fact that, given the subject and non-subject NPs, the former
is the more prominent position, the Surface Structure EH and the Word Order EH can be considered to
be two different manifestations of the same principle:

(27) S

YNTACTIC

ROMINENCE

MPATHY

IERARCHY

: Give syntactic prominence to a person/object

that you are empathizing with.

The Surface Structure EH deals with the manifestation of syntactic prominence in terms of structural
configuration, while the Word Order EH deals with the manifestation of syntactic prominence in terms
of linear order.

Let us examine some more examples relevant to the Word Order EH:

(28) a. I saw John and a policeman walking together yesterday.
b. ??I saw a policeman and John walking together yesterday.

(29) a. I saw you and a policeman walking together yesterday.
b. ??I saw a policeman and you walking together yesterday.

(30) a. John and someone else will be there.
b. *Someone else and John will be there.

The marginality of (28b) and (29b) arises from the conflict between the Word Order EH and the Topic

Page 7 of 21

14. Empathy and Direct Discourse Perspectives : The Handbook of Pragmatics : Blac...

28.12.2007

http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9780631225485...

EH. The unacceptability of (30b) arises from the conflict between the Word Order EH and
Descriptor/Topic Empathy Hierarchies. Note that

someone else

in the sentence is a descriptor that is

dependent upon

John

Recall now that I mentioned previously that since

meet

is a reciprocal verb, if John and Mary met,

there are four ways to describe this event:

(10) a. John and Mary met.
b. Mary and John met.
c. John met Mary.
d. Mary met John.

I have already explained the difference between (10c) and (10d) by saying that the former involves the
speaker's intentional placement of

John

in subject position, while the latter involves his/her

intentional choice of

Mary

in that position. The Word Order EH can account for the difference between

(10a) and (10b): the speaker's empathy with John is greater than that with Mary in (10a), and the
speaker's empathy with Mary is greater than that with John in (10b).

The Word Order EH interacts in an interesting way with a “modesty” principle taught in prescriptive
grammar. For example, observe the following sentences:

(31) a. ??I and John are good friends.
b. John and I are good friends.

(31b) involves the following empathy hierarchy conflicts:

(32) Empathy Relationships of (31b):
Speech Act EH E(I) > E(John)
Word Order EH E(John) m> E(I)
Transitivity: *E(I) > E(John) m> E(I)
Markedness Principle: The above conflict is intentional because the speaker has intentionally
placed

John

in the left-hand position, and

in the right-hand position  a penalty

That is, from the point of view of the Empathy Perspective, (31b) should be unacceptable, and (31a)
acceptable. However, prescriptive grammar says that the first person nominative pronoun should be
placed at the end of a list. I will refer to this artificial rule as the Modesty Principle:

(33) T

ODESTY

RINCIPLE

: In the coordinate NP structure, give the least prominence to the

first person pronoun.

We can now account for the acceptability of (31b) by stating that the choice of the expression

John

and I

, which is instrumental in creating an empathy relationship conflict, is not intentional, but is

forced on the speaker by the Modesty Principle. Therefore, there is no penalty for the violation.

2.2 More on the Markedness Principle for

2.2 More on the Markedness Principle for Discourse Rule Violations

Discourse Rule Violations

The contrast in the acceptability status of the following two sentences seems to be unexplainable
without resorting to the Markedness Principle for Discourse Rule Violations:

(34) a. *At the gate were John

's brother and John

/he

smiling at me.

b. At the top of the rank list were Johni's brother and Johni in that order.

Both (34a) and (34b) contain a conflict between the Descriptor EH (i.e. E(John) > E(John's brother)) and
the Word Order EH (i.e. E(John's brother) > E(John)). (34a) is unacceptable because the placement of

John's brother

to the left of

John

in the NP coordination is by the speaker's design. In contrast, (34b) is

acceptable because the placement of

John's brother

to the left of

John

has been forced on the speaker

by the relative ranking of the two siblings. That is, the empathy relationship conflict that (34b)

Page 8 of 21

14. Empathy and Direct Discourse Perspectives : The Handbook of Pragmatics : Blac...

28.12.2007

http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9780631225485...

involves was forced upon the speaker and was not intentional, and therefore, there is no penalty for
the violation.

Observe next the following sentences:

(35) a. John gave a book to the girl.
b. John gave the book to a girl.

(36) a. John gave the girl a book.
b. ??John gave a girl the book.

(35a, b) are examples of the periphrastic dative sentence pattern, whereas (36a, b) are examples of
the incorporated dative sentence pattern. What needs to be explained is the marginality of (36b). It is
reasonable to assume that (36b) is marginal because it violates the well-known discourse principle

given below:

(37) F

ROM

-O

-T

-N

RINCIPLE

: In languages in which word order is relatively free, the

unmarked word order of constituents is old, predictable information first and new,
unpredictable information last.

Let us assume that the above principle applies to English in places where there is freedom of word
order. In (36b),

a girl

, which represents new information, appears before

the book

, whose anaphoric

nature marks that it represents old information. Therefore, the marginality or unacceptability of (36b)
can be attributed to its violation of the

From-Old-To-New Principle

. However, once one adopts this

approach to account for the marginality of (36b), the acceptability of (35a), repeated below, becomes
a puzzle:

(35) a. √John gave a book to the girl.
New Old

(36) b. ??John gave a girl the book.
New Old

As shown above, (35a) violates the From-Old-To-New Principle as much as (36b) does.

The above dilemma can be resolved by assuming that the periphrastic dative pattern represents the
underlying pattern for giving verbs, and that the incorporated dative pattern is derived by applying

ATIVE

INCORPORATION

to the underlying periphrastic dative pattern.

According to this hypothesis, the

violation of the From-Old-To-New Principle that (35a) involves is non-intentional because the
speaker simply used the underlying sentence pattern, and placed the theme NP

a book

in verb object

position and the goal NP

the girl

in prepositional object position. Therefore, there is no penalty for the

violation. In contrast, the violation of the Principle that (36b) involves is intentional because the
speaker has chosen to apply Dative Incorporation, an optional transformation. Therefore, the resulting
violation of the From-Old-To-New Principle cannot go unpenalized, and the unacceptability of the

sentence results.

2.3 Empathy and reflexive pronouns

There are languages (e.g. Japanese, Korean, and Chinese) that require that the antecedents of
reflexive pronouns be animate and, most preferably, human. This suggests that the reflexive
pronoun, at least in these languages, requires a high degree of the speaker's empathy with its
referent. The fact that sentences such as (38b) below are acceptable might give a false impression
that English reflexives are free from such a requirement, but the fact that (39b) is unacceptable shows
that they are subject to an empathy requirement, albeit to a lesser degree.

(38) a. John criticized himself.
b. Harvard overextended itself in natural sciences in the sixties.

(39) John wrote to his friends about himself.

b. *Harvard wrote to its alumni about itself.

Page 9 of 21

14. Empathy and Direct Discourse Perspectives : The Handbook of Pragmatics : Blac...

28.12.2007

http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9780631225485...

The unacceptability of (39b) shows that English reflexive pronouns in oblique case position require a
high degree of the speaker's empathy with their referents. (39b) is unacceptable because it is not
possible for the speaker to empathize to a high degree with inanimate objects like Harvard University.

Empathy factors influence the interpretation of reflexive pronouns even when their referents are
human. Observe first the following sentences:

(40) a. John talked to Mary about himself.
b. √/?/??Mary talked to Bill about himself.

While all speakers accept (40a), some speakers consider (40b) awkward or marginal. This fact can be
explained by assuming that a sentence containing a reflexive pronoun in oblique position is most
felicitous when the referent of the reflexive receives the highest degree of empathy in the sentence.
There is no problem with (40a) because the subject NP is the sentence's unmarked empathy focus (cf.
the Surface Structure EH) and the reflexive pronoun has that NP as its antecedent. In contrast, (40b) is
problematic in that the antecedent of the reflexive pronoun is not the highest-ranked candidate in the
empathy hierarchy on the unmarked interpretation of the sentence. Likewise, observe the following
sentences:

(41) a. √/?/??Mary talked to Bill about himself. (=40b)
b. ?/??/*I talked to Bill about himself.

There are many speakers who consider (41b) less acceptable than (41a). This can be attributed to the
fact that the subject NP in (41b), because it is a first person pronoun, is even stronger than

Mary

(41a) in its qualification as the focus of the speaker's empathy, and hence as the antecedent of the
reflexive. This makes the

Bill

of (41b) less qualified to be the antecedent of a reflexive pronoun than

the

Bill

in (41a), and makes (41b) less acceptable than (41a).

Observe next the following sentences:

(42) a. √/?/??John talked to Mary about herself.
b. *John talked about Mary to herself.

The fact that (42a) (and (40b)) is acceptable or nearly so for many speakers has been a problem in the
framework of Chomsky's (1981) theory of grammar because the reflexive pronoun is not c-
commanded by its intended antecedent

Mary

. According to Chomsky's binding theory, a reflexive

must be c-commanded by a co-indexed NP in a local context.

Chomsky (1981) circumvented this

problem by claiming that

talk to

in (42a) is reanalyzed as a single V, with a resulting loss of the PP

node dominating to

Mary

. Thus,

Mary

becomes the direct object of the V, and it c-commands the

reflexive pronoun. Chomsky argued that (42b) is unacceptable because reanalysis of

talk about

does

not take place, apparently because

about Mary

is not base-generated next to

talk

There is, however, a serious problem with the above account of the acceptability of (42a) and the
unacceptability of (42b). Observe the following sentence:

(43) ??/*John discussed Mary with herself.

There is no doubt that the reflexive pronoun in the above sentence is c-commanded by a co-indexed
NP (i.e.

Mary

) in its local domain. In spite of this fact, (43) is marginal or unacceptable. Observe,

however, that (43) and (42b) are more or less synonymous. Therefore, the unacceptability of (42b)
seems to be a non-syntactic phenomenon, rather than a syntactic one. In comparing the acceptable
(42a) with the unacceptable (42b), we note that while the antecedent of the reflexive pronoun in the
former is Mary as a human being, the antecedent of the reflexive in (42b) and (43) is semantically
inanimate; that is, the antecedent is what Mary is or what she has done. Thus, the unacceptability of
(43) is automatically accounted for in the framework of the Empathy Perspective via the requirement
that the referents of the antecedents of the reflexive pronouns in oblique position in English must

Page 10 of 21

14. Empathy and Direct Discourse Perspectives : The Handbook of Pragmatics : Bl...

28.12.2007

http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9780631225485...

receive a high degree of the speaker's empathy. That is, (42b) and (43) are unacceptable because

Mary

, the antecedent of the reflexive pronouns, cannot receive a high degree of the speaker's

empathy because it is semantically inanimate.

Finally, observe the following picture-noun sentences involving reflexive pronouns:

(44) a. Mary cost John a picture of himself in the paper.
Intended Interpretation: “In order to impress Mary, John paid for a picture of himself to be
printed in the newspaper.” or “John was adversely affected by the fact that what Mary had done
caused his picture to appear in the newspaper.”
b. *Mary cost John a picture of herself in the paper.
Intended Interpretation: “In order to impress Mary, John paid for her picture to be printed in the
newspaper.”

Note that in these sentences

John

is semantically human because it represents the experiencer of the

cost or damage, whereas

Mary

is semantically inanimate because it represents notions such as “what

Mary had done,” “(John's) desire to impress Mary,” and so on. These sentences also show that reflexive
pronouns in picture nouns are also empathy expressions, and as such require a high degree of the
speaker's empathy with their referents. There are several other factors that conspire to produce the
acceptability judgments for these sentences; see Kuno 1987: Chap. 4.5) for details.

3 Direct Discourse Perspective

3.1 Logophoric NP constraint

Observe the following sentences:

(45) a. John said, “I am a genius.”
b. Johni said that he

was a genius.

(46) a. John said to Mary, “You are a genius.”
b. John said to Mary

that shej was a genius.

(47) a. John said about Maryj, “Mary

/Shej is a genius.

b. John said about Mary

that she

was a genius.

(45a), (46a), and (47a) contain direct discourse quotations, whereas (45b), (46b), and (47b) contain
indirect discourse clauses. In the indirect discourse clauses in (45b), (46b), and (47b), the pronouns
are coreferential with main-clause NPs. However, there is a significant difference between the
pronouns in (45b) and (46b) and the pronoun in (47b). That is, in the former, the pronouns
correspond to the first and second person pronouns in the corresponding direct quotations: they

cannot correspond to non-pronominal NPs because the following sentences are unacceptable:

(48) a. *John

said, “John

is a genius.”

b. *John said to Mary

, “Mary

is a genius.”

Let us refer to saying and asking verbs as l

OGOPHORIC

VERBS

(abbreviated as LogoV), and to their

complement clauses as

LOGOPHORIC

COMPLEMENTS

(abbreviated as LogoComp). Given a sentence with a

logophoric complement, I will use the term Logo-1 NP (or the abbreviation Logo-1) to refer to the NP
in the main clause that refers to the speaker of the utterance represented by the logophoric
complement. Likewise, I will use the term Logo-2 NP (or the abbreviation Logo-2) to refer to the
hearer of the utterance. These terms are illustrated in (49):

(49) John

said to Mary

that she

was a genius.

Logo-1 LogoV Logo-2 LogoComp

I assume that even sentences with complements that do not have direct discourse counterparts are
logophoric complements if they represent the thoughts, feelings, or realization of the referent of the

Page 11 of 21

14. Empathy and Direct Discourse Perspectives : The Handbook of Pragmatics : Bl...

28.12.2007

http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9780631225485...

main-clause Logo-1 or Logo-2 NP:

(50) a. John thinks that he is a genius.
Logo-1 LogoV LogoComp
b. Hypothetical Structure: [John thinks, “[I am a genius.]”]

(51) a. John heard from Mary that she was sick.
Logo-2 LogoV Logo-1 LogoComp
b. Hypothetical Structure: [John heard from Mary, “[I am sick]”]

I will refer to an analytical framework that makes use of the notions described above as a direct
discourse perspective, or alternately as a

LOGOPHORIC

perspective.

Kuno (1987: Chap. 3) has shown that an NP in a logophoric complement (in an extended sense, as
shown above) that is intended to be coreferential with the main-clause Logo-1 or Logo-2 NP behaves
very differently from those NPs that are not coreferential with either of them. I will illustrate this
difference by using a few examples from (Kuno 1987) and add a new set of data from Kuno (1997)
that further illustrates the importance of the direct discourse perspective. Observe, first, the following
sentences:

(52) a. The remark that Churchill

was vain was often made about him

b. *The remark that Churchill

was vain was often made to him

While (52a) is acceptable on the interpretation in which the non-pronominal full NP

Churchill

in the

embedded clause is coreferential with the pronoun

him

in the main clause, such an interpretation is

ruled out for (52b). This contrast can be explained only by paying attention to who said what. I will
represent what was said using a direct discourse representation:

(53) a. [People often made about Churchill the remark “[Churchill is vain]”]
Logo-1 -Logo-1/2
b. [People often made to Churchill the remark “[You are vain]”]
Logo-1 Logo-2

Observe that

Churchill

in the matrix clause of (53a) is marked as -Logo-1/2 because it is neither the

speaker NP nor the hearer NP of the proposition represented by the direct discourse quotation. The
subject of the direct discourse quotation is

Churchill

, and not

you

, because, again, Churchill was not

the hearer of the remark. In contrast,

Churchill

in the matrix clause of (53b) is marked as Logo-2

because it is the hearer NP of the direct discourse quotation, which has

you

, and not

Churchill

, in

subject position. The fact that (52a) is acceptable but (52b) is not suggests that a full NP (i.e. a non-
reflexive and non-pronominal NP) in the direct discourse representation of a logophoric complement
can remain as a full NP if other conditions are met, as in (52a), but a second person pronoun in the
direct discourse representation of a logophoric complement cannot be realized as a full NP. That is, a
second person pronoun

you

must remain pronominal in indirect discourse formation.

Likewise, observe the following sentences:

(54) a. The allegation that John

was a spy was vehemently denied by himi.

b. *The claim that John

was a genius was made by him

While (54a) is acceptable on the interpretation whereby the full NP

John

in the embedded clause is

coreferential with the pronoun

him

in the main clause, such an interpretation is ruled out for (54b).

This contrast can also be explained by observing the direct discourse representation of what was said
by whom:

(55) a. [John denied the allegation “[John is a spy]”]
b. [John made the claim “[I am a genius]”]

Page 12 of 21

14. Empathy and Direct Discourse Perspectives : The Handbook of Pragmatics : Bl...

28.12.2007

http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9780631225485...

Note that John was neither the “speaker” nor necessarily the hearer of the allegation. This explains
why

John

, and not

you

, appears in subject position of the direct discourse representation in

(55a).

In contrast, John was necessarily the “speaker” of the claim, and therefore, the subject of the

direct discourse representation in (55b) must be a first person pronoun. The fact that (54b) is
unacceptable suggests that a first person pronoun in the direct discourse representation of a
logophoric complement cannot be realized as a full NP in the derived surface sentences. That is, the
first person “I” in a direct discourse representation has to remain pronominal in indirect discourse
formation. I should hasten to add that if Passivization does not apply to (52a) and (54a), there is no
way to pronominalize the main clause NPs

Churchill

and

John

and keep

Churchill

and

John

in the

embedded clause unpronominalized:

(56) a. *People often made about him

the remark that Churchilli was vain.

b. *He

vehemently denied the allegation that John

was vain.

But the unacceptability of (56a, b) can be attributed to violation of Principle C (see section 3.2 of this
paper) of my version of the Binding Theory, which says that an R-expression (a full NP) cannot be c-
commanded by a co-indexed NP (with PP nodes not counting for the purpose of delimiting the c-
command domain of a given node).

Let us depart from the account given above, which is based on the direct discourse representation of
logophoric complement clauses, and move to one which assumes that indirect discourse logophoric
complements are base-generated as such. In that framework, the constraint that we have observed
above can be restated in the following manner:

(57) L

OGOPHORIC

NP C

ONSTRAINT

: Given a sentence with a matrix Logo-1/2 NP and a logophoric

complement attributable to that Logo-1/2 NP, a full NP in the logophoric complement cannot
be coreferential with the Logo-1/2 NP in the main clause. (cf. Kuno 1987: 109)

According to this constraint, the acceptability status of the sentences in (52) and (54) can be
accounted for in the following manner:

(58) a. The remark that [Churchill

was vain] was often made about him

LogoComp -Logo-1/2
b. *The remark that [Churchill

was vain] was often made to him

LogoComp Logo-2

(59) a. The allegation that [John

was a spy] was vehemently denied by him

LogoComp -Logo-1/2
b. *The claim that [John

was a genius] was made by him

LogoComp Logo-1

(58a) and (59a) are acceptable because full NPs in their logophoric complements are co-indexed with
main clause NPs that represent neither the speaker nor the hearer of the propositions that the
logophoric complements represent. In contrast, (58b) and (59b) are unacceptable because full NPs in
their logophoric complement clauses are co-indexed with the main clause hearer/speaker NPs that
the logophoric complements are attributable to. Observe next the following sentences:

(60) a. *The claim [

LogoComp1

that John

said [

LogoComp2

that Bill

was a spy]] was made by him

cf. John made the claim:

“I

said that Bill is a spy.”

b. *The claim [

LogoComp1

that John

said [

LogoComp2

that Billj was a spy]] was made by him

cf. Bill made the claim: “John said that

am a spy.”

The fact that the sentence is unacceptable on the

him = Bill

interpretation shows that the Logophoric

NP Constraint applies between a full NP (e.g.

Bill

in (60b)) in a logophoric complement and a Logo-1/2

NP (e.g.

him

in (60b)) in a higher clause, even if there is an intervening Logo-1/2 NP (e.g.

John

(60b)) between the two.

Page 13 of 21

14. Empathy and Direct Discourse Perspectives : The Handbook of Pragmatics : Bl...

28.12.2007

http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9780631225485...

There are many phenomena that can be accounted for only in the Logophoric Perspective. They are
discussed in detail in Kuno (1987: Chap. 3).

3.2 The Logophoric NP

3.2 The Logophoric NP Constraint and the Binding Theory

Constraint and the Binding Theory

I will now show that the Logophoric NP Constraint can resolve the puzzle given in (61) that has defied
attempts at explanation by scholars working in the framework of Chomsky's

BINDING

THEORY

(see also

Huang, this volume).

(61) a. *Which claim that John

was asleep was he

willing to discuss? (Chomsky 1993)

b. Which claim that John

made did he

later deny? (Lebeaux 1992)

The problem here is at what stage the unacceptability of these sentences can be captured as involving
a violation of Principle C of the Binding Theory:

(62) Principle C: An R-expression cannot be co-indexed with a c-commanding NP.

C-command: A c-commands B iff the branching node α1 most immediately dominating A
either dominates B or is immediately dominated by a node α

that dominates B, and α

is of

the same category type as α

. (Reinhart 1976)

For those readers who are not familiar with the notion of c-command and Chomsky's Binding Theory,
it is sufficient for the purpose of this paper to interpret Principle C in a much more limited sense as
meaning that, given a sentence with NP

in subject position and NP

elsewhere in the same sentence,

cannot be interpreted as coreferential with NP

, as illustrated below:

(63) a. *John

/*He

hates

Johni's

mother.

b. *John

/*He

hated the man that

John

shared an office with.

c. *John

/*He

thinks that

John

is a genius.

The above sentences are all unacceptable because the italicized

John

is intended to be coreferential

with the main clause subject, in violation of Principle C.

Returning to (61a), it has been assumed that Principle C applies to the open sentence portion of the
abstract representation (called the LF representation) of the structure of the sentence informally
shown in (64):

(64) LF representation of (61a):
[Which x [he was willing to discuss [x claim that John was asleep]]]
{______Open sentence______}

That is, it has been assumed, in essence, that the LF representation given in (64) is illicit on the
coreferential interpretation of

and

John

because it violates Principle C in the same way that the

following sentence does:

(65) *He

was willing to discuss which claim that John

was asleep.

The above account of the unacceptability of (61a) immediately runs into difficulty, however, because it
predicts that (61b) should also be unacceptable because it violates Principle C in the same way that
(67) does:

(66) LF representation of (61b):
[Which x [he did later deny [x claim that John made]]]

(67) *He

did later deny the claim that John

made.

Attempts have been made in the framework of the Minimalist Program (Freidin 1986, 1994, 1997;
Lebeaux 1988, 1992, 1995; Chomsky 1993) to resolve the above puzzle and account for the

Page 14 of 21

14. Empathy and Direct Discourse Perspectives : The Handbook of Pragmatics : Bl...

28.12.2007

http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9780631225485...

acceptability of (61b) and the unaccept-ability of (61a) by attributing it to the difference in the ways
that complements and adjuncts are introduced into sentence structures. Noting that the embedded
clause in (61a) is a complement clause of the noun

claim

, whereas the embedded clause in (61b) is an

adjunct relative clause of the noun, minimalist theorists have claimed that the contrast in acceptability
status between these two sentences can be accounted for by assuming the following:

(68) (i) The introduction of complements into sentence structures must be cyclic.
(ii) The introduction of adjuncts into sentence structures can be cyclic or non-cyclic.

Thus, they have assumed that Principle C applies to the LF representations that are informally shown
below:

(69) a. LF representation of (61a):
[Which claim that John was asleep [he was willing to discuss {which claim that John was
asleep}]]
b. LF representation of (61b)
[Which claim that John made [he was willing to discuss {which claim}]]

In the above LF representations, the copy of a fronted wh-expression is shown in curly brackets. Note
that the complement clause

that John was asleep

in (69a) is adjoined to the noun

claim

before the

syntactic fronting of the

-expression, whereas the adjunct clause

that John made

in (69b) is

adjoined to the fronted

which claim

, and not to the expression before

-movement takes place.

Principle C disallows the co-indexing of the full NP

John

with the c-commanding

in the open

sentence part (i.e.

[he was willing to discuss {which claim that John was asleep}]

). In contrast, Principle

C does not apply to the open sentence portion of the LF representation in (69b) because there is no
full NP there. Principle C does not apply to

John

in the fronted wh-expression because it is not in the

open sentence part of the LF representation (and

does not c-command

John

anyway).

The above account of the contrast between (61a) and (61b) appears to be credible when coupled with
the contrast between (70a) and (70b), which also appears to show an

ARGUMENT

ADJUNCT

ASYMMETRY

(70) a. ??/*Which pictures of John

did he

like? (Lebeaux 1992)

b. Which pictures near John

did he

look at? (Lebeaux 1992)

Observing that

of John

in (70a) is a complement of

pictures

, but

near John

in (70b) is an adjunct,

Lebeaux (1992) attempts to account for the marginality/unacceptability of (70a) and the acceptability
of (70b) on the coreferential interpretation of

John

and

in the following way:

(71) a. LF representation of (70a)
[Which pictures of John [he did like {which pictures of John}]]
b. LF representation of (70b):
[Which pictures near John [he did look at {which pictures}]]

The introduction of

of John

in (70a) takes place before the fronting of the

-expression

which

pictures

because it is a complement of

pictures

. Therefore, a copy of the fronted

-expression

which pictures of John

is in the object position of the verb

, as shown in (71a). Thus, Principle C

disallows the co-indexing of

John

with the c-commanding

. This explains the unacceptability of

(70a) on the coreferential interpretation of

John

and

. In contrast, the introduction of

near John

(70b) can take place after the

-movement of

which pictures

because it is an adjunct, not a

complement, of

pictures

. Therefore, the copy of

John

is absent in the open sentence part of the LF

representation of the sentence, as shown in (71b). Principle C is inapplicable to the open sentence
part of (71b), and hence Principle C does not mark (70b) unacceptable on the coreferential
interpretation of

John

and

The above account of the contrast between (61a) and (61b) and between (70a) and (70b), based on
the claimed asymmetry in the ways that arguments and adjuncts are introduced into sentence

Page 15 of 21

14. Empathy and Direct Discourse Perspectives : The Handbook of Pragmatics : Bl...

28.12.2007

http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9780631225485...

structures, does not go far beyond these four sentences, however. Once the database of sentences
with the same patterns is only slightly extended, it becomes clear that the claimed asymmetry is an
illusion. Observe the following sentences:

(72) a. Whose allegation that Johni was less than truthful did he

refute vehemently?

b. Whose opinion that Weld

was unfit for the ambassadorial appointment did he

try to refute

vehemently?
c. Whose claim that the Senator

had violated the campaign finance regulation did he

dismiss

as politically motivated?
d. Which psychiatrist's view that John

was schizophrenic did he

try to get expunged from the

trial records?

The embedded clauses in the above sentences are all complement clauses. Therefore, a Minimalist
analysis based on argument/adjunct asymmetry predicts that they should all be unacceptable. In spite
of this prediction, however, most speakers consider these sentences acceptable, and even those
speakers who judge them as less than acceptable report that they are far better than (61a).

The argument/adjunct asymmetry-based analysis of the contrast between (70a) and (70b) fares as
poorly, as witnessed by the acceptability of sentences such as the following:

(73) a. Which witness's attack on John

did he

try to get expunged from the trial records?

b. Which artist's portrait of Nixon

do you think he

liked best?

c. Whose criticism of John

did he

choose to ignore?

(73) a. Which doctor's evaluation of John

's physical fitness did he

use when he

applied to

NASA for space training?
b. Which psychiatrist's evaluation of John

's mental state did he

try to get expunged from the

trial records?

The PPs in the above sentences are all complements of the nouns (i.e.

attack, portrait, criticism,

evaluation

) and therefore, Freidin (1986), Lebeaux (1988) and Chomsky (1993) all predict that their LF

representations violate Principle C. But these sentences are all perfectly acceptable. The acceptability
of the sentences in (72)–(74) shows not only that an argument/adjunct-asymmetry-based account of
the contrast between (61a) and (61b) and between (70a) and (70b) is untenable, but also that to the
extent that the account is derived from the theoretical framework of the Minimalist Program, there is
something wrong with the theory itself.

3.3 Logophoric

3.3 Logophoric analysis

analysis

Observe now the contrast in acceptability status of the following sentences:

(75) a. *Which claim that John

had helped develop new technologies did he

make at last year's

national convention?
b. Which claim that John

made did he

later deny? (Lebeaux 1992)

The Logophoric NP Constraint can automatically account for the contrast between these two
sentences: (75a) involves a logophoric complement that is attributable to the matrix subject NP

The sentence violates the Logophoric NP Constraint because a full NP (i.e.

John

) in the logophoric

complement is co-indexed with the matrix Logo-1 NP (i.e.

). (75b), in contrast, does not involve a

logophoric complement, and therefore, the Logophoric NP Constraint has nothing to do with this LF
representation, hence the acceptability of this sentence.

The acceptability of the sentences in (72) can be accounted for in the same fashion. For example,
observe the following:

(72) a. Whose allegation that John

was less than truthful did he

refute vehemently?

The above sentence has a logophoric complement (i.e.

that John was less than truthful

), but the Logo-

Page 16 of 21

14. Empathy and Direct Discourse Perspectives : The Handbook of Pragmatics : Bl...

28.12.2007

http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9780631225485...

1 NP of this complement is not

John

), but

whose

. Therefore, the Logophoric NP Constraint does

not disallow the co-indexing of

John

and

Now let us re-examine (61a), the sentence that Chomsky, Lebeaux, and Freidin have all considered
unacceptable:

(61) a. *Which claim that John

was asleep was he

willing to discuss? (Chomsky 1993)

It seems that the sentence is potentially ambiguous with respect to whether the claim that John was
asleep is to be interpreted as John's claim or someone else's claim. As must be clear to the reader by
this time, the Logophoric NP Constraint predicts that the sentence is unacceptable if the claim is
interpreted as John's, but acceptable if it is interpreted as someone else's claim. This prediction is
consistent with the judgments that most, if not all, native speakers make about the sentence.

The above observations show that the account of the contrast between (61a) and (61b) that is based
on argument/adjunct asymmetry is untenable. That is, there is no justification for assuming that
arguments and adjuncts are different with respect to when they must or must not be introduced into
sentence structures. Let us assume that they are both introduced cyclically. According to this
hypothesis, (72a), (61b), and (75a) have the structures shown below:

(76) a. LF representation of (72a)
[Whose allegation that John was less than truthful [he did refute vehemently {whose allegation
that John was less than truthful}]]
b. LF representation of (61b)
[Which claim that John made [he did later deny {which claim that John made}]]
c. LF representation of (75a)
[[Which claim that John had helped develop new technologies] [he did make {which claim that
John had helped develop new technologies} at last year's national convention]]

The way that the Binding Theory (cf. Chomsky 1981) is organized, Principle C applies to the open
sentence part of these LF representations, and marks them as unacceptable. But (72a) and (61b) are
perfectly acceptable. Therefore, the Binding Theory needs to be re-examined to see if it has been
properly organized. The problem with its current organization is that it is based on the assumption
that since there are three types of NPs (i.e. anaphors, pronominals, and R-expressions), there should
be one rule for each NP type. I have proposed a different organization of the binding theory in Kuno
(1987). I show below first the overall difference in organization and then present revised binding
principles:

(77) Chomsky's Organization of the Binding Theory
a. anaphors in a local domain [coreference]
b. pronominals in a local domain [disjoint reference]
c. R-expressions in all domains [disjoint reference]

(78) Kuno (1987)'s Organization of the Binding Theory
a. anaphors in a local domain [coreference]
b. non-anaphors (pronominals and R-expressions) in a local domain [disjoint reference]
c. R-expressions in all domains [disjoint reference]

(79) Kuno (1987)'s Binding Principles
Principle A': An anaphor may receive a coreferential interpretation only with a c-commanding
NP within its local domain. N.B. An LF representation that contains an anaphor which is not
interpreted coreferentially with any NP in it is unacceptable.
Principle B': A non-anaphor (pronominal or R-expression) is obligatorily assigned disjoint
indexing

vis-à-vis

a c-commanding NP within its local domain.

Principle C': An R-expression is barred from receiving coreferential interpretation

vis-à-vis

a c-

commanding non-anaphor NP in either A- or A'-position.

N.B. Principle C does not apply to the reconstructed portion of the LF representation. (That is,
in a theoretical framework in which the binding theory applies to syntactic structures rather

Page 17 of 21

14. Empathy and Direct Discourse Perspectives : The Handbook of Pragmatics : Bl...

28.12.2007

http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9780631225485...

than to LF representations, Principles A' and B' apply cyclically, and Principle C' applies post-
cyclically.)
C-command: A c-commands B iff the non-PP branching node α

most immediately dominating

A either dominates B or is immediately dominated by a node α

that dominates B, and α

is of

the same category type as α

With the above revised binding theory, (76a-c) pose no problem. In these LF representations, since
the non-anaphor

John

is in the embedded clause and the c-commanding NP

in the main clause

(that is, since

is not in

John's

local domain), Principle B' does not apply. Therefore, they are not

assigned disjoint indexing. Furthermore, Principle C' does not apply to an R-expression which is in
the reconstructed portion of the LF representation, and therefore, it is not assigned disjoint indexing

vis-à-vis

the c-commanding

. (The

John

in the fronted portion of the LF representation is not

assigned disjoint indexing with

either, because the latter does not c-command the former.) Thus,

there are no binding principles that block the coreferential interpretation of

and

John

. The

Logophoric NP Constraint does not apply to (76a) because

is not the Logo-1 NP of the complement

clause. It does not apply to (76b), either, because that sentence does not involve a logophoric
complement at all. Hence the acceptability of (72a) and (61b). On the other hand, the Logophoric NP
Constraint applies to (76c) because

is the Logo-1 NP of the complement clause and disallows a

coreferential interpretation of

and

John

, hence the unacceptability of (75a).

The above revised binding principles can also account for the contrast among the following three
sentences:

(70) a. ??/*Which pictures of John did he like? (Lebeaux 1992)

(73) a. Which witness's attack on John did he try to get expunged from the trial records?

(70) b. Which pictures near John did he look at? (Lebeaux 1992)

These sentences have the following LF representations:

(80) a. LF representation of (70a)
[Which pictures of John [

did like {which pictures of

John

}]]

b. LF representation of (73a)
[Which witness's attack on John [

did try to get {which witness's attack on

John

} expunged

from the trial records]]
c. LF representation of (70b)
[Which pictures near John [

did look at {which pictures near

John

}]]

Note in (80a) that the non-anaphor

John

is c-commanded by

in its local domain. Therefore,

Principle B' obligatorily assigns disjoint indexing to

and

John

. Hence the unacceptability of (70a). In

contrast, in (80b),

John

is not c-commanded by

in its local domain because

John's

local domain is

which witness's attack on John

. Hence Principle B' does not assign disjoint indexing to

and

John

Principle C does not apply to

John

in the reconstructed portion of the LF representation. (It applies to

John

in the fronted wh-expression, but it is not c-commanded by

, and thus no disjoint indexing

takes place.) Therefore, there is no binding principle that disallows the co-indexing of

and

John

Furthermore, the Logophoric NP Constraint does not apply to

and

John

either, because no

logophoric complement is involved. Therefore, it does not disallow the co-indexing of the two NPs,
hence the acceptability of (73a). Finally, in (80c), the c-commanding

is not in

John's

local domain

because there is a clause boundary between

and

John

, as witnessed by the fact that

John

is not in a

reflexive context:

(81) a. *He looked at pictures near himself.
b. *[He did look at [pictures [PRO near himself]]]

(82) a. *Mary talked with people angry about herself.
b. *[Mary talked with people [PRO angry about herself]]

Page 18 of 21

14. Empathy and Direct Discourse Perspectives : The Handbook of Pragmatics : Bl...

28.12.2007

http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9780631225485...

Therefore, Principle B' does not apply to (80c), and consequently, there is no obligatory disjoint-
indexing of

and

John

by Principle B'. Principle C does not apply to

John

in the reconstructed portion

of the LF. (It applies to

John

in the fronted

-expression, but since that is not c-commanded by

it does not bar co-indexing of the two NPs.) The Logophoric NP Constraint does not apply because
the sentence does not contain a logophoric complement. Thus, there is no rule that disallows the co-
indexing of the two NPs, hence the acceptability of (80c).

4 Concluding Remarks

In this paper, I have examined various syntactic constructions in English and shown how syntactic and
non-syntactic constraints interact with one another to produce the acceptability status of sentences
that employ those constructions. Given a contrast in acceptability status like the one in (61), the
linguist who is unaware of the existence of various non-syntactic factors that interact with syntax
assumes that the contrast is due to syntactic factors and proposes syntax-based hypotheses to
account for it. In contrast, the linguist who is aware of various non-syntactic factors that closely
interact with syntax begins his/her analysis bearing in mind that the contrast might be due to one or
more such non-syntactic factors. I hope I have amply demonstrated in this paper which approach is
more productive in arriving at the correct generalizations on such constructions.

ACKNOWLEDGMENTS

I am indebted to Karen Courtenay, Nan Decker, Tatsuhiko Toda, and Gregory Ward for their numerous
invaluable comments on earlier versions of this paper. I am also greatly indebted to Bill Lachman for
providing me with the phrase structure representation in note 9.

1 It is possible for a speaker to take a detached view of an event involving the referent of a topic NP. This is
why the Topic EH E(topic) ≥ E(non-topic) has “≥” rather than

“>”

2 Sentences of the same pattern as (14b) are acceptable if they are used as corrective sentences:
(i) Speaker A: Mary heard from Bill that Jane was seriously sick. Speaker B: No, she heard it from ME. I will
discuss empathy principle violations in corrective sentences later in this section. Likewise, sentences of the
same pattern as (15b) become acceptable if placed in contexts in which the time and location of the receipt
of the thing the speaker sent are at issue. For example, observe the following sentence:
(ii) When Bill

received from me

a package containing a maternity dress for his wife, they had already broken

up. Note that (ii) is not synonymous with (iii):
(iii) When

I sent

Bill a package containing a maternity dress for his wife, they had already broken up. It is

clear that while the mailing-out time of the package is at issue in (iii), the receipt time of the package is at
issue in (ii). That is, the speaker's use of the marked expression

receive from

in (ii) has been forced on the

speaker because of the necessity to refer to the receipt time rather than the mailing-out time. Therefore,
there is no intentionality in the speaker's use of the expression

receive from

in (ii), nor is there a penalty for

the conflict in empathy relationships that the sentence contains.

3 (16d) is acceptable if the pronouns are stressed, as shown in (ib) below. But note that (ii), which has the
same relative order of full names and pronouns, is acceptable without stress on the pronouns.
(i)a. ?? Maryi heard from Johnj what hej had heard from heri two days before. (=16d)
b. Mary

heard from John

what HE

had heard from HER

two days before.

c. Mary

told John

what he

had told her

two days before. The above phenomenon is similar to the

pronominalization phenomenon with possessive NPs as antecedents. First note that (ii) is perfectly
acceptable:
(ii) Billi's brother is visiting him

. But when two such sentences are juxtaposed as in (iiia), unacceptability

results:
(iii)a. *Bill

's brother is visiting him

, and John

's uncle is visiting him

b. Bill

's brother is visiting HIM

, and John

's uncle is visiting HIM

c. Bill

is visiting his

brother, and John

is visiting his

uncle. The right-hand pronoun, if unstressed, cannot

be interpreted as coreferential with

John

- it must be interpreted as coreferential with the left-hand pronoun

- but if stressed, it is interpreted as coreferential with

John

. This contrasts with the fact that (iiic), which has

the same relative order of full names and pronouns, is acceptable without requiring the stressing of the
pronouns. In Kuno (1975: 289), the unacceptability of (iiia) with the pronouns unstressed was attributed to

Page 19 of 21

14. Empathy and Direct Discourse Perspectives : The Handbook of Pragmatics : Bl...

28.12.2007

http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9780631225485...

the following constraint: (iv) T

ONSTRAINT

RONOMINALIZATION

WITH

ENITIVE

NTECEDENT

: The

coreference linkage between a noun phrase in the genitive case and a pronoun is weak unless they are
coreferential with the discourse topic. According to the above hypothesis, (ii) is acceptable because it is easy
to assume that Bill is the topic of the preceding discourse, but (iiia) is unacceptable because the assumption
needed to obtain the specified coreferential interpretation for the latter half of the sentence (i.e. the
assumption that John should be the topic of the preceding discourse) is contradicted by the first part of the
sentence. The fact that (iiia) becomes acceptable if the pronouns are stressed suggests that the following
principle applies (cf. G. Lakoff 1971b, Hirschberg and Ward 1992): (v) T

UNCTION

TRESSED

RONOUNS

The stressing of pronouns signals that they are coreferential with NPs that they are not expected to refer to
according to the normal set of rules in operation. The above principle can account for the contrast between
(via) and (vib) below (cf. Lakoff 1971b):
(vi)a. John

hit Billj, and then he

i/*i

hit Tom.

b. John

hit Bill

, and then HE

*i/j

hit Tom. If the pronoun

is unstressed, it is coreferential with

John

. We can

attribute this interpretation to the general principle that says, given coordinated clauses, try to assign
parallel interpretations to the clauses as much as possible (G. Lakoff 1971b, Hirschberg and Ward 1992). In
contrast, if the pronoun is stressed, it must be interpreted as coreferential with

Bill

. Thus, the stress on the

pronoun signals to the hearer not to follow the parallel interpretation principle, but to look for a different
antecedent, hence arises the

HIM = Bill

interpretation of (vib). Returning now to (iiib), the stress on the

pronouns in the sentence signals that the pronouns refer to NPs that the Constraint on Pronouns with
Genitive Antecedents would otherwise ban them from referring to, hence the specified coreferential
interpretation of the sentence. Likewise, the stress on the pronouns in (ib) signals that they refer to NPs that
the Empathy Principles would otherwise ban them from referring to, hence the acceptability of the sentence
with the specified coreferential interpretation.

4 There are many speakers who prefer (i) to (31b):
(i) Me and John are good friends. The above pattern violates both the artificial Modesty Principle and the
prescriptive Nominative Case Marking Rule for subject NPs. There are fewer speakers who say:
(ii) John and me are good friends. What this seems to suggest is that the acquisition of the Modesty Principle
and of the Nominative Case Marking Rule go together.

5 This principle was first proposed by Mathesius (1939) and became the central theme of Prague School
linguistics. See also Firbas (1964), Daneš (1970) and Gundel and Fretheim (this volume).

6 There are several pieces of evidence that support this assumption, but there is no space to present them
here.

7 As far as I know, Olga Yokoyama (personal communication, 1975) was the first to observe the marginality
of sentences of the pattern of (36b) and to attribute it to the violation of the From-Old-To-New Principle
caused by application of optional Dative Incorporation.

8 The non-reflexive counterpart of (39b) is also unacceptable:
(i) *Harvardi wrote to its alumni about it

9 There are several different definitions of c-command, but for the purpose of the present discussion, the
following definition will suffice:
(i) C-command: Node A c-commands node B if the first branching node that dominates A also dominates B,
and A does not dominate B. (Reinhart 1976) Let us assume that (iia) has the structure shown in (iib):
(ii)a. John talked to Mary about himself.

Page 20 of 21

14. Empathy and Direct Discourse Perspectives : The Handbook of Pragmatics : Bl...

28.12.2007

http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9780631225485...

Bibliographic Details

The Handbook of

The Handbook of Pragmatics

Pragmatics

Edited by:

Edited by: Laurence R. Horn And Gregory Ward
eISBN:

eISBN:

eISBN: 9780631225485
Print publication

Print publication

Print publication date:

date:

date: 2005

In (iib), NP

c-commands NP

because S, the first branching node that dominates it, also dominates NP

and NP

does not dominate NP

. Furthermore, NP

and NP

are in the same local domain (i.e. in the same

simplex S). Therefore, the reflexive pronoun

himself

satisfies Principle A of the binding theory:

Principle A: An anaphor (a reflexive or reciprocal pronoun) must be co-indexed with a c-commanding NP in
a local domain. In contrast, NP

does not c-command NP

in (iib) because PP

, the first branching node that

dominates NP

, does not dominate NP

. Therefore, co-indexation of

Mary

with the reflexive pronoun

herself

in (42a) constitutes a violation of Principle A, and the sentence is predicted to be unacceptable, which it is
not for many speakers.

10 It goes without saying that (48a, b) are acceptable if

John/Mary is a genius

is what John actually said. In

this connection, note the following sentence, which is perfectly acceptable because Ali used to say “Ali is the
greatest”: (i) Mohammed Alii used to say, “Alii is the greatest.”

11 If the alleger had made the allegation directly to John, the direct discourse representation of what the
alleger said would have a second person pronoun in subject position: (i) [John denied the allegation “[You
are a spy]”] The above representation would not yield (54a) because

you

must remain pronominal in indirect

discourse formation.

12 An R(eferring)-expression is a non-reflexive/non-reciprocal NP or a trace of a moved NP.

13 Observe the following sentences: (i)a. *Himi, Johni's father dislikes.
b. *Himi, Johni's mother thinks Mary is in love with. The unacceptability of these setences can be captured
only if Principle C is allowed to apply to

John

, with the c-commanding co-indexed

him

in A'-position. See

Kuno (1987: Chap. 2) for details.

Cite this article

KUNO, SUSUMU. "Empathy and Direct Discourse Perspectives."

The Handbook of Pragmatics

. Horn, Laurence

R. and Gregory Ward (eds). Blackwell Publishing, 2005. Blackwell Reference Online. 28 December 2007
<http://www.blackwellreference.com/subscriber/tocnode?
id=g9780631225485_chunk_g978063122548516>

Page 21 of 21

14. Empathy and Direct Discourse Perspectives : The Handbook of Pragmatics : Bl...

28.12.2007

http://www.blackwellreference.com/subscriber/uid=532/tocnode?id=g9780631225485...