117
254
THS

 

ll fljllllll will“ in (Ill! ll! [11 will l l l

THEIts

This is to certify that the

thesis entitled

CONTRASTIVE STRESS
AND SENTENCE COMPREHENSION

presented by

Joanne Regina Mazzella

has been accepted towards fulfillment
of the requirements for

M.A . degreein Psychology

Mme-63‘

Major professor

Date August 3, 1979

0-7639

 

 

OVERDUE FINES ARE 25¢ PER DAY
PER ITEM

Return to book drop to remove
this checkout from your record.

 

 

 

CONTRASTIVE STRESS AND SENTENCE COMPREHENSION

by

Joanne Regina Mazzella

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF ARTS

Department of PsycholOgy

1 979

ABSTRACT
CONTRASTIV'E STRESS AND SENTENCE COMPREHENSION
by

Joanne Regina Mazzella

The effects of sentence stress on comprehension were investigated
in two experiments. The measure of comprehension used was subject's
reaction time to prerecorded sentences which varied with respect to
locus of heaviest stress.

Experiment I considered the idea that placing contrastive stress
on appropriate sentence items can facilitate a listener‘s comprehen-
sion. Results indicated that comprehension can be facilitated by using
contrastive stress-to point out the correct candidate for a role given
in a preceding sentence. Subjects comprehended sentences carrying ap-
propriate contrastive stress significantly faster than sentences with
inappropriate or no stress.

Experiment II compared effects of stressing the new information
versus the given information in sentences. Stressing the given infor—
mation of a target sentence slowed down comprehension when the stressed
item was the subject of the sentence but not when it was the object,
indicating that the effect of the position of sentence items should be
investigated further.

In both experiments, stress was used to indicate new information,

but in different ways. In Experiment I, new information presented in

Joanne Regina Mazzella

a target sentence functioned as a replacement for information from a
previous sentence. In Experiment II this new information added to,
rather than replaced, previous information. It appears that stress
may function differently in different types of sentences.

These experiments demonstrate the importance of taking sentence

stress factors into account in models of comprehension.

ACKN OWIEDGI’IEINTS

I would like to express my sincere appreciation to the members of
my thesis committee, Dr. Gordon Wood (Chairman), Dr. J. Kathryn Book
and Dr. Rose Zacks, whose guidance and encouragement made this work
possible. I am deeply indebted to Mr. David Ehresman for his valuable
instruction on the use of the IDP—il/LLO, and for his personal assistance
in all phases of this endeavor. I would also like to thank m. Steven
Gitterman for his help with the data analysis. Special thanks to my
parents, Mr. and Mrs. Guido R. Mazzella, and to my husband, Mr. Paul G.

Coady, for their never-ending faith and understanding support.

ii

TABLE OF CONTENTS

LIST OF TABIES I I I I I I I I I I I I I I I I
LIST OF FIGURES I I I I I I I I I I I . I I I I
mm @11an I I I I I I I I I I I I I I I I I
RENT I I I I I I I I I I I I I I I I I
Nethm I I I I I I I I I I I I I I I I I
Subjects . . . . . . . . . . . . . .
materials I I I I I I I I I I I I I
Procedure . . . . . . . . . . . .
Results . . . . . . . . . . . . . .
DiSCUSSiOn I I I I I I I I I I I I I I I
mmmq‘ II I I I I I I I I I I I I
Methm I I I I I I I I I I I I I I I
Subjects . . . . . . . .

materials I I I I I I I I I I I
Procedure . . . . . . . . . . .

Results . . . . . . . . . . . . . . . . .

Discussion . . . . . . . . . . . . . . .
GENERAL DISCUSSION . . . . . . . . . . . .
FOOTNOTES . . . . . . . . . . . . . . . . .

AITENDIX A: Experiment I Sentence Materials .

APEENDIX B: ReCOgnition Test for Experiment I .

AITENDIX C: Experiment II Sentence Materials
AITENDIX D: Lists Presented in Experiment II
APPENDIX E: ReCOgnition Test for Experiment II

LIS T OF “was I I I I I I I I I I I I I I

iii

iv

11

15
15
15

. 17

20

47
1+9
52

-55

56

TABLE

LIST OF TABLES

PAGE
Mean reaction time to target sentences in Experiment I. 20
Mean reaction time to target sentences in Experiment II. 30
Relative reaction times for active and passive target 34
sentences.

iv

LI ST OF FIGURES

FIGURE PAGE

1 Mean reaction time to target sentences in Experiment I. 30

INTRODUCTION

A basic function of language is the communication of new ideas.
Some of the information a speaker communicates to a listener is infor-
mation which is already shared by them, while other information may be
previously unknown to the listener. A speaker assumes that there is
certain information that is being introduced to the listener for the
first time, and the speaker uses the information that they already have
in common as a starting point for presenting this new information.
This is done so that the listener will be able to assimilate the new
information properly, i.e. so that the new information can be related
to the listener's knowledge of the world in the way which the speaker
intended. Information may be shared by virtue of already being in the
knowledge structure of both people, by being part of the immediate
physical environment that both are aware of, or by having been intro-
duced linguistically in previous sentences (Chafe, 1970). For example,
if (1) were uttered, it would typically be the case that the speaker

(1) The President of the United States is in Paris today.
assumes that the listener shares the knowledge that there exists a
President of the United States, and the speaker is using that as a
starting point for communicating the new information that that person
is in Paris today. Information may also be shared from the physical
environment. Sentence (2) could be uttered if both the speaker and

(2) The box is empty.

2

hearer are aware of a box before them, and one communicates to the
other the information that that particular box is empty. Likewise, (3)

(3) The box was empty.
could be uttered if the speaker had already introduced the concept of a
certain box in a prior sentence and now wishes to communicate the fact
(previously unknown to the listener) that it was empty.

The distinction between shared and new information has been dis-
cussed by a number of authors, and a.diverse set of terms has been used
to designate the two types of information. The set includes the terms
mpg-pg (Halliday. 1967; Haviland and Clark, 1974; Chafe, 1974, 1976),
gldfngg (Chafe, 1970), topic-comment (Hornby, 1971), presupposed-
focused (Hornby, 1974), and theme-rheme (Halliday, 1967). The conven-
tion of using the designations ggggg and Egg will be adopted in this
paper unless otherwise noted.

When a speaker presents the given information of a sentence, in
effect the context or reference is being set up within which the new
information of that sentence is to be interpreted. Chafe (1974) sug-
gests that what a speaker says is selected from what is in his or her
consciousness at the time, and the speaker intends any utterances to
enter the listener's consciousness. Since at any particular moment a
listener may have things in consciousness unrelated to the information
a speaker wishes to convey, the speaker must construct utterances to
assure that the listener does indeed bring into consciousness that part
of the knowledge store to which new information is to be related.

Chafe (1970) suggests that a speaker constructs sentences so that the
surface structure subject carries given information, thereby using Eng-

lish surface structure to signal what information the listener is to

3
bring out of the knowledge store and into consciousness.

Haviland and Clark (1974) pr0pose that a tacit agreement exists
between a speaker and listener as to how information is to be presented
in sentences. The speaker agrees to present what is believed the lis-
tener already knows about as given information, and to present as new
information the information the speaker believes the listener does not
yet have knowledge'of. The listener assumes that the speaker is fol-
lowing this convention and therefore, in order to understand what is
said, uses a "Given-New Strategy" to interpret a sentence. This three-
step procedure involves (1) breaking the sentence into given and new
information, (2) searching memory for a unique antecedent matching the
given information, and (3) attaching the new information to this ante—
cedent. Since the authors assume that knowledge is represented in
long-term memory as a set of interrelated propositions, the antecedent
searched for in Step 2 corresponds to a proposition or set of proposi—
tions representing information known to the listener. By attaching new
information to the appropriate antecedent in Step 3, the listener has
integrated information into memory which was previously unknown.1

Haviland and Clark contend that in complying with the tacit agree-
ment between speakers and listeners, or the "Given'New Contract", a
speaker must try to construct sentences so that the listener has only
one direct antecedent for any given information. That unique antece-
dent should be the one intended by the speaker. If (4) were uttered,

(4) Agnes saw somebody. It was Agnes who saw Maxine.
the speaker would be in violation of the Given-New Contract; since the
existence of Agnes is established in the first sentence, the speaker

should present Agnes as given information in the second sentence,2 in

A

violation of the Given-New Contract requirement that a speaker present
information that the listener is believed to be familiar with as given.
Haviland and Clark contend that in order to comprehend this utterance,
the listener uses a procedure known as "Restructuring"; that is, the
listener must restructure the given and new information in the second
sentence so that "Agnes saw someone" is given information rather than
"someone saw Maxine", and "that someone is Maxine" is new information
rather than "that someone is Agnes". Sentences which need to be re-
structured typically sound very awkward to the listener.

Haviland and Clark (1974) have presented psychOIOgical evidence
which.indicates that a listener may in fact be using the Given-New
Strategy in comprehending sentences. The basic assumption of their
experiments is that if a listener is able to locate a (unique) direct
antecedent in memory for the given information of a sentence, it should
take less time to comprehend that sentence than if no direct antecedent
is readily available. Using reading time as a measure of comprehension,
Haviland and Clark found that target sentences were comprehended sig-
nificantly faster when preceded by a context sentence containing a
direct antecedent for the given information of the target sentence,
than when the context sentence did not contain a direct antecedent for
this information. Holyoak and Clark.(unpublished; see Clark, 1973) also
found evidence that sentences which need to be restructured take sig-
nificantly longer to comprehend (with reading time as the measure of
comprehension) than sentences which do not require restructuring.

Many authors have noted (e.g. Halliday, 1967; Chafe, 1970; Chomsky,
1971: Clark, 1973) that items carrying new information in a sentence are

in general given heavier stress than items reflecting given information.

5
Chafe (1970) contends that new information is spoken with higher pitch
and greater amplitude since it is the primary information the speaker
wishes to convey, and this pitch difference would therefore increase
the effectiveness of communication. Hornby (1972) has found evidence
which tends to support the notion that stressing an item may play an
important role in indicating it as new information. He asked subjects
to decide which of two pictures went with a simple sentence. Subjects
had to choose which picture the sentence was about, in other words,
which picture depicted the psychological subject (or given information)
of the sentence. Neither picture was an accurate representation of the
sentence since either the agent or the object differed from the sen-
tence agent or object. When subjects were presented with an active
sentence (e.g. The Indian is building the igloo), the agent was most
often selected as the psycholOgical subject of the sentence. However,
when the same sentence was presented with stress on the agent (i.e. The
Indian is building the igloo),3 the agent was not selected as the psy-
chological subject of the sentence as often as was the sentence object.
In other words, the sentence constituent that was stressed was less
likely to be considered as given information in the sentence. Stressing
an item may thus serve to indicate it as new information.

The information in a sentence that a speaker assumes is not shared
by the listener has been called the "focus" of a sentence (Jackendoff,
1972). Halliday also considers that the focus of a sentence is its new
information, or information that is not recoverable from preceding dis-
course. In sentences (5) and (6), stress is used to indicate which

(5) Felicity eats caviar for breakfast.

(6) Felicity eats caviar for breakfast.

6
items carry new information, and the focus of these two sentences is
different. A speaker would probably utter (5) to convey information
about who it is that eats caviar for breakfast, while (6) would be
uttered to convey information about what it is that Felicity eats for
breakfast. Cutler (1976) contends that if a listener understands (5)
and (6) as identical, these sentences have not been understood since
they differ on the dimension of focus. Likewise, Cutler states that in
order to successfully comprehend (7), a listener must recognize that

(7) Felicity eats gazigr for breakfast?

(8) Yes, she likes to indulge herself.

(9) She's on a fish-only diet.

(10) Did you think it was Samantha who did?
(11) No, for dinner.
(8) and (9) are acceptable responses, but (10) and (11) are not.

Cutler (1975, 1976) argues that focus information constitutes an
integral part of the semantic representation constructed when a sentence
is understood. She points out that while most sentence comprehension
models in some way incorporate the three basic stages of (1) identifi-
cation of word boundaries, (2) lexical look-up, and (3) perception of
syntactic structure, these stages are necessary but not sufficient to
describe the sentence comprehension process. For example, Cutler (1976)
notes that a listener could interpret (12) as being an expression of

(12) Cassandra is a real genius.
praise; however if (12) were spoken with an ironic intonation contour
(i.e. nasalized, with heavy stress on certain words), it could be inter-
preted to have the reverse meaning. An ironically spoken sentence has

a conveyed meaning which is the converse of its literal meaning (Cutler,

7

1974). In order to understand the information a speaker intended to
convey when speaking (12) with.an ironic intonation contour, comprehen-
ding the literal meaning of the sentence would not be sufficient.

Cutler (1976) notes further that if (13) were spoken with heavy
stress on the word "above", this would imply that some other sentence

(13) The above sentence was ironically intended.
was not ironically intended. Stress used in this way is said to be
contrastive. Bolinger (1961) has described contrastive stress as a
phenomenon

by which two or more items are counterbalanced and a pref-

erence indicated for some member or members of the group.

It is the most conspicuous of all the occurrences of pho-

netic highlighting by reason of its frequency amd the extra

oomph that we put into it, and because our attention is

focused in a way that makes us aware of our speech and not

just of our meaning. (p.83)

According to Chafe (1976), three factors are involved in the use of
contrastive stress. When contrastive stress is used, as in (14), the

(14) Rgggld made the hamburgers.
speaker is assuming (1) that the listener knows that someone made the
hamburgers and therefore treats this information as given. The speaker
is also assuming (2) that the listener has a limited number of candi-
dates in mind for the role of the person who made the hamburgers. Fi-
nally, the most important aspect of a contrastive sentence is (3) that
the correct candidate for this role is pointed out to the listener.
That the particular role exists is given information, and the speaker is
presenting new information by pointing out the correct candidate for
that role.

Cutler (1975) contends that the expectation of contrastive stress

can facilitate sentence processing. A phoneme-monitoring task was given

8

to subjects in which sentences such as (15a) and (15b) were presented.

(15) a. She managed to remove the dirt from the rug,
but not the berry stains.

b. She managed to remove the dirt from the rug,
but not from their clothes.

0. She managed to remove the dirt from the rug.
It should be noted that in (15a) the word "dirt" would receive heavy
stress since it is contrasted with "berry stains", and in (15b) the
word "rug" would receive heavy stress since it is contrasted with "their
clothes". The target phoneme in these sentences was /d/ and therefore
was the initial sound of a high stress item ('dirt") in (15a), but in
(15b) the item "dirt" would receive low stress. Sentence (15c) was
also recorded, in a neutral tone of voice. In (15c) the target—bearing
item received an intermediate level of stress, and this item was spliced
into (15a) and (15b) in place of the target-bearing item in those sen-
tences. As a result, the target-bearing items in (15a) and (15b) were
acoustically identical, but the intonation contours preceding them were
different. In (15a) the intonation contour was consistent with the oc-
currence of a high stress item in the position of the target-bearing
word, and in (15b) the contour was consistent with the occurrence of a
low stress item in that position. Phonemes were detected significantly
faster in sentences such as (15a) than in sentences such as (15b), even
though the target-bearing items were acoustically identical. Since re—
action time to target phonemes in phoneme-monitoring tasks is assumed
to be directly related to processing difficulty (i.e. shorter reaction
times are expected when processing is less difficult), Cutler's results
suggest that subjects' sentence comprehension was facilitated by the

information contained in the intonation contour preceding the

9
target-bearing item. However, Newman and Dell (1978) have presented
evidence indicating that phoneme detection latencies are strongly affec-
ted by the length of the word preceding a target-bearing item, with
phonemes being detected faster as the length of the preceding word in-
creases. Since the intonation contour preceding a high stress word is
longer than the contour preceding a low stress item (Lehiste, 1970),
Cutler's results may reflect the extra processing time subjects had to
detect target phonemes. It should be noted however that giving a lie-
tener extra processing (comprehension) time may in fact be an implicit
function of sentence stress.

Shields, uchugh and Martin (1974) and Cutler and Foss (1977) found
that phonemes are detected faster in target-bearing items which are
stressed than in items which are unstressed. Cutler (1975) contends
that phoneme detection is faster when the target-bearing item is
stressed because the stress indicates sentence focus, and rapid identi-
fication of focal (i.e. new) information facilitates comprehension.
Cutler and Fodor (1979) reasoned that if this is true, phoneme detection
should also be faster for target-bearing items whose focus in indicated
in another way (i.e. without the use of stress). Cutler and.Fodor map
nipulated sentence focus by preceding the target sentence with a ques-
tion, the answer to which was the focus of the target sentence. For
example, instead of presenting sentences such as (16) and (17) in which

(16) The man on the.gggggr was wearing the blue hat. ‘

(17) The man on the corner was wearing the plug hat.
focus is indicated by stress, the authors presented an acoustically
identical version of (18) preceded by either (19) or (20). If (18)

(18) The man on the corner was wearing the blue hat.

10

(19) Which man was wearing the hat?

(20) What hat was the man wearing?
were preceded by the question in (19), the focus of (18) would be the
same as the focus of (16). Likewise, the focus of (18) would be the
same as that of (17) if (18) were preceded by the question in (20).
Therefore, if the phoneme /k/ were to be detected, the target-bearing
item in (18) would be focussed information if (18) were preceded by
(19), but would be non-focussed information if (18) were preceded by
(20).“ By using these preceding questions to indicate focus, Cutler
and Fodor eliminated the confounding factor of the length of stressed
versus unstressed sentences, since the sentences containing target items
were acoustically identical. Results indicated that phonemes were de-
tected faster when the target-bearing item was the focus of the sen-
tence. That this finding occurred in the absence of stress factors
suggests that stress may facilitate phoneme detection because it indi—
cates focal (i.e. new) information, and not simply because subjects are
afforded more time for detection when the target-bearing item is
stressed.

Cutler (1975, 1976) and Cutler and Fodor (1979) have argued that
the reason sentence stress facilitates phoneme detection is because the
stress allows a listener to identify the new informatiOn of a sentence
more rapidly. Since reaction time to target phonemes is assumed to be
directly related to the difficulty of sentence processing when the tar-
get phoneme occurs, faster detection of phonemes contained in stressed
items indicates that stress facilitates sentence processing.

Speakers usually place heaviest stress on a sentence's new infor-

mation, and it may be that in doing so the speaker increases the

11
effectiveness of communication by making new information explicit so
that a listener can readily identify it as new information when trying
to comprehend a sentence. It can.be argued that appropriate use of
stress will facilitate sentence comprehension, while inappropriate use
will slow it down. Since contrastive stress can be used to designate a
particular role as given information and a particular element as the
correct candidate (i.e. new information) for the given role, the effects
of stressing both appropriate and inappropriate items in contrastive
sentences can be investigated. Appropriate use of contrastive stress
(i.e. designating an appropriate candidate for a given role) should
facilitate comprehension, while inappropriate use (i.e. contrastively
stressing an inappropriate candidate for a given role) should slow down
comprehension. These predictions were investigated in the following

experiments.

EXIERIMENT I

Sentence (21) has at least three scopes of negation. This sentence
(21) George didn't kill Bill.
could be taken to mean (1) that it wasn't George, but rather someone
else, who killed Bill, (2) that George did something to Bill other than
kill him, or (3) that George killed someone but not Bill. The negation

introduces a contrast, but doesn't explicitly point out where the

12

contrast is. A speaker could use stress to indicate the locus of the
contrast, and thereby communicate to a listener which of these meanings
were intended. If the first meaning above were intended, (22) would be

(22) _G_e_o_r_gp_ didn't kill Bill.

(23) George didn't 3.1.1}. Bill.

(2L1) George didn't kill §_i_l;.
uttered. Sentence (23) would be uttered if the second meaning were in-
tended, and likewise (24) for the third meaning.

In the case of (22), the listener is (explicitly) led to believe
that someone other than George was the killer. If this sentence were
followed by a statement of who the killer actually was, (i.e. someone
other than George), contrastive stress would be placed on the candidate
being pointed out as the correct one for the role of Bill's killer, as
in (25). Likewise,.if (23) were followed by a statement of what George

(25) §§g£g§_didn't kill Bill. ‘Alfggd killed Bill.

did to Bill rather than killing him, stress would be placed on the item
reflecting George's alternate action, as in (26). And finally, if sen-

(26) George didn't g1; Bill. George wounded Bill.
tence (24) were followed by a statement of who George killed (in con-
trast to Bill), stress would be placed on the candidate being pointed
out as the correct one for the role of George's victim, as in (27).5

(27) George didn't kill _B_:g._l_. George killed 51:593.

In (25)-(27), the item stressed in the second sentence of each
pair is the one which is being pointed out by the speaker as the-correct
alternative for the role which is put in question in the first sentence
of the pair. The stress pattern of the first sentence of the pair in-

forms the listener as to which item is being negated. After hearing the

13
first sentence in (28a) a listener knows that it wasn’t George, but

(28) a. George didn't kill Bill. Alfred killed Bill.
b. George didn't kill Bill. Alfred killed Bill.
c. George didn't kill Bill. Alfred killed Bill.
d. George didn't kill Bill. Alfred killed Bill.

 

rather someone else, who killed Bill. The listener then has as given
information that a role exists (i.e. the role of Bill's killer). In

the second sentence in (28a) the new information "Alfred" is presented,
pointing out the correct candidate for the role. In processing this
sentence, the listener may perform the operations of locating the given
information of the previous sentence and filling in this new information
about the correct candidate for the role in question. In other words,
the new information replaces previous information; for example, in (28a)
Alfred replaces George in the role of Bill's killer.

The first sentence in (28c) provides the given information that the
role of George's victim exists, and that the correct candidate for the
role is someone other than Bill. The second sentence in (28c) presents
the new information that Alfred is the correct candidate for the role of
Bill's killer. However, when the listener tries to locate the given in-
formation of the first sentence, the role of "George's victim" is found,
and the new information doesn't replace someone in that role. In order
to make sense of the second sentence and make the new information fit,
the listener may have to locate different given information in the first
sentence; that is, that someone other than George was Bill's killer.
This "Restructuring" operation should take time; therefore it was pre-
dicted that the second sentence in (28c) would take longer to comprehend
than the same sentence in (28a), since in (28a) the new information
"Alfred" is an appropriate replacement for the role (given information)

of the preceding sentence. In other words, in (28a) the context

14
sentence provides a direct antecedent for the given information that
someone other than Alfred killed Bill, but the context sentence in
(28c) does not.

In the first sentence of (28b), no item is stressed and therefore
what is given information may not be as clearly established as in (28a)
and (280). The second sentence in (28b) presents the new information
that Alfred is the correct candidate for the role of Bill's killer.

. The presentation of this new information may resolve the ambiguity of
the first sentence and suggest that the role of Bill's killer was the
role in question. Since the second sentence in (28b) is not actually
inappropriate in the context of its preceding sentence, it was predicted
that it would be comprehended faster than the second sentence in (280),
which is inappropriate. Since the information that the role of Bill's
killer is in question is explicitly given through the use of contrastive
stress in the first sentence of (28a), it was also predicted that the
second sentence in (28a) would be comprehended faster than the second
sentence in (28b) which is preceded by an ambiguous context sentence.

In (28d) no contrastive stress is present in either sentence. It
was predicted that the second sentence in (28b) would be comprehended
faster than the second sentence in (28d). These sentences are preceded
by identical (unstressed) context sentences; however, contrastively
stressing the new information of the target sentence in (28b) may facil-
itate comprehension by more clearly identifying that sentence's new in-
formation.

To summarize, it was predicted that comprehension time for (a) tar-
get sentences would be shorter than for (c) target sentences, comprehen-

sion time for (b) target sentences would be shorter than for (0) target

15
sentences but longer than.for (a) target sentences, and comprehension
time for (b) target sentences would be shorter than for (d) target sen-

tences.

m

Subjects. Subjects were 40 adult native speakers of English, par-
ticipating for extra credit in an introductory psycholOgy course at
Michigan State University. All subjects were tested individually and
participated in both Experiments I and II within a single session.
Twenty subjects, selected at random, served in Experiment I first and
the remaining 20 served in Experiment II first. Within each set of 20
subjects, individuals were assigned at random to each of the 4 lists
described below, 5 subjects to a list. Therefore, each of the 4 lists
was presented to a total of 10 subjects.

Materials. The verbal materials consisted of the 20 sets of 4
sentence pairs listed in Appendix A, plus sentence pairs i-iii listed
in Appendix A which were used in practice trials for all subjects.

Each set consisted of 4 sentence pairs which varied with respect to (A)
whether any item was stressed in the second (target) sentence (A1;
stress; A2; no stress), and (B) whether the stress in the first (con-
text) sentence was placed on the same item as in the target sentence,
placed on no item, or placed on a different item from the one stressed
in the target sentence (B1: same stress; B2; no stress; BB; different
stress). The sentence pairs from (28) above corresponding to each pos-
sible condition are presented below;

AlBi; ggppgg didn't kill Bill. 'Alfrgg killed Bill. (same, stress)

A B

1 2: George didn't kill Bill. Alfred killed Bill. (no stress,

stress)

16

A B : George didn't kill Bill. Alfred killed Bill. (different,
1 3 stress)

AZBZ; George didn't kill Bill. Alfred killed Bill. (no stress,
no stress)

Over the 20 sets of sentences, the item stressed in the target
sentence was either the subject, verb, object, adjective, or object of
the preposition of the sentence an equal number of times. Therefore,
in 4 sentence sets the stressed item was the subject of the sentence,
in 4 it was the verb, and so on.

Four lists of 20 sentence pairs each were constructed so that each
list contained 5 randomly selected sentence pairs from each of the 4
possible conditions, with no list containing more than one sentence
pair from any sentence set. Each of the (20x4) or 80 possible sentence
pairs was present in the lists, so that 20 pairs from each condition
were presented in all. Each list of 20 sentence pairs was preceded by
the same 3 practice sentence pairs (i.e. pairs i-iii in Appendix A).

All sentences were recorded by the experimenter on Scotch brand
magnetic tape using a Revox A-700 tape recorder. Using the above exams
ple, only the following sentences were actually recorded for this par-
ticular (representative) set:

A B :‘Qggrgg didn't kill Bill. Alfggd killed Bill.

1 1

A232: George didn't kill Bill. Alfred killed Bill.

BB: George didn't kill Bill.
The 2 types of target sentences (A1 and A2) and the 3 types of context
sentences (31' B2 and B3) were then used to construct the 4 different
context sentence/target sentence combinations in the following way.
All sentences which had been recorded were digitized using a 16-bit

audio to digital converter, constructed by Three Rivers Computer

17

Corporation. The digital form of each sentence was then stored on mag-
netic disc. Using a EDP-11/40 computer, 4 lists were constructed by
playing the sentences back in the proper order for each list through a
digital to audio converter, and recording them on Scotch brand magnetic
tape using a Revox A-700 tape recorder. Therefore the final recording
consisted of 4 lists of the original 20 sentence pairs (and the 3 prac-
tice sentence pairs) in the proper order for each list.

Procedure. Four groups of 10 subjects each were asked to listen
to a list of 23 prerecorded pairs of sentences, the first 3 pairs of
which were practice trials. Subjects were assigned to groups at random
and each group received one of the 4 lists described above. Lists were
played on a Teac A—334OS tape recorder, and were presented to subjects
binaurally via headphones. Subjects were seated in front of a panel of
2 buttons, in a separate room from the tape recorder. When subjects
were ready to begin the experiment, they pressed the button on the left
side of the panel, designated as the "Ready" button, which initiated
presentation of the first sentence. This button was pressed only to
begin the experiment. Subjects were instructed to press the button on
the right side of the panel as soon as they understood each sentence.
Any subject inquiries as to what was meant by "understanding" the sen-
tence were answered by telling subjects to press the button just as soon
as they felt they understood what the sentence meant. The inquiring
subjects' primary concern was that highly complicated or technical sen-
tences would be presented. When assured they would hear "simple" sen-
tences (the example "John loves Mary" was given verbally to concerned
subjects), all subjects appeared satisfied that they understood the in-

structions before the experiment began.

18

A one-second silent interval elapsed between the subject's re-
. sponse to the first sentence of each pair and the beginning of the sec-
ond sentence of that pair. A five-second silent interval elapsed be-
tween the subject's response to the second sentence of a pair and the
beginning of the first sentence of the next pair. However, each time
a response was made before a sentence ended, these Silent intervals
were measured from the end of that sentence rather than from.the sub-
ject's response. In other words, if a subject responded before a sen-
tence ended, the next sentence was presented after a silent interval
which was measured from the end of the preceding sentence. If a subject
responded after a sentence ended, the next sentence was presented fol-
lowing a silent interval which was measured from the subject's response.

The button switches which the subjects pressed to begin the experi-
ment and to indicate their responses were interfaced with a EDPLBi come
puter. The computer was programmed to monitor a subject's button
presses, and to start and stop the tape recorder at the appropriate
times. The computer recorded the length of time measured from the be-
ginning of each sentence to the subject's button press. The exact
length of each sentence had been determined earlier when the lists were
being constructed. The length of each sentence was subtracted from the
time measured from.the beginning of each sentence to the subject's but-
ton press, and the resulting values were considered as the subject's
reaction times. These values were subsequently used in all data analy-
ses. Reaction times to sentences in the 3 practice trial pairs were
disregarded.

After all sentences had been presented, subjects were given a paper

and pencil recognition memory test (see Appendix B). Test items

19
consisted of a random.listing of all the target sentences which had just
been presented, plus an equal number of distractor sentences interspered
in a random fashion. The distractor sentences were similar to the sen-
tences which had been presented. Subjects were asked to indicate
whether or not each sentence was one that was just heard by writing
"Yes" or "No" after each sentence.

The purpose of the memory test was to provide a check on whether
subjects were actually comprehending the sentences presented to them.

It would have been possible for subjects to merely press the button af-
ter hearing each sentence without actually trying to comprehend them.
However, if a subject were able to recognize the sentences in a later
memory test with a high degree of accuracy, it could be concluded that
the sentences were actually comprehended. Memory test results were used
for no other purpose than as an overall comprehension check.

Subjects were not informed of the memory test at the beginning of
the experiment. It was suspected that if informed, subjects might not
respond as soon as each sentence was comprehended, but rather would re-
hearse the sentences before responding in hopes of improving memory test
performance. Therefore the memory test at the end of the first experi-
ment was a surprise. However, subjects very likely suspected that an-
other memory test would be presented at the end of the second experiment
(since the procedures of both experiments were identical), but by that
time would not be likely to initiate a rehearsal strategy. The ordering
of the two memory tests was counterbalanced since half the subjects re-
ceived Experiment I first and Experiment II second, while the other half

received the two experiments in the reverse order.

20
Results
Mean reaction times (RTS) for target sentences in each of the 4
conditions were computed, collapsing across subjects within each condi-

tion. These mean RTS are presented in Table 1.

Table 1. Mean reaction time to target sentences in Experiment I.

Condition Description of Condition Mean RT (in msec)

 

Stress in context sentence; same
A131 item stressed in target sentence -14
(e.g. George didn't kill Bill.
Alfred killed Bill.)

No stress in context sentence;
A132 stress in target sentence 75
(e.g. George didn't kill Bill.
Alfred killed Bill. )

Stress in context sentence; different
AlB3 item stressed in target sentence 137
(e.g. George didn't kill Bill.
Alfred killed Bill.)

No stress in context sentence;
AZB2 no stress in target sentence 143
(e.g. George didn't kill Bill.
Alfred killed Bill.)

Several planned comparisons were carried out with the following

results: Mean RT for condition A131 (same, stress) was less than mean

RT for condition A133 (different, stress), t(39)= -4.68, p<.0001; mean

RT for condition AlB1 (same, stress) was less than mean RT for condition
A132 (no stress, stress), t(39)= -3.00, p<.005; mean RT for condition

A132 (no stress, stress) was less than mean RT for condition A1B (dif-

3
132
stress, stress) was less than mean RT for condition AZB2 (no stress, no

ferent, stress), t(39)= -1.68, p<.1; and mean RT for condition A’ (no

stress), t(39)= -2.23, p<.05.

Thirty-four of the 40 subjects showed the effect of a reaction

21
time difference in the predicted direction between conditions A1B1
(same, stress) and A1B3 (different, stress); 27 of the 40 subjects
showed this directional effect between conditions A1B1 (same, stress)
and A1B2 (no stress, stress); 24 of the 40 subjects showed the effect
between conditions A1B2 (no stress, stress) and A133 (different,
stress); and 28 of the 40 subjects showed the effect between conditions

A1B2 (no stress, stress) and A (no stress, no stress).

232

Mean RT for stressed context sentences (261 msec) was not signifi-
cantly different from mean RT for unstressed context sentences (281
msec). F(1.39)=-.76.

The physical lengths of all sentences presented in this experiment
were measured. Stressed sentences (i=2125 msec) were significantly
longer than unstressed sentences (I=1861 msec), F(1,98)=12.8, p<.001.

Memory test performance was .90, indicating that subjects did com-
prehend the sentences presented to them. Performance of subjects who
received the experiments in the I—II order was .895 for Experiment I,
while performance of subjects receiving the experiments in the II—I
order was .905 for Experiment I. Since memory test performance in Ex-
periment I of subjects receiving Experiment I first was virtually iden-
tical to the performance of subjects who received Experiment I after
Experiment II, it appears that anticipation of a memory test made no

difference in reCOgnition performance.

Discussion
This experiment was designed to consider whether the use of con-
trastive stress influences sentence comprehension. Target sentences

containing items stressed appropriately in the context of the intended

22
meaning of the preceding sentence were comprehended faster than inappro—
priately stressed target sentences and target sentences preceded by an
unstressed context sentence. Inappropriately stressed target sentences
were also comprehended more slowly than target sentences preceded by
unstressed context sentences. The fact that the target sentences pre—
deded by these unstressed sentences were not inappropriate lends further
support to the notion that contrastively stressing information that is
explicitly presented as given in a preceding sentence does in fact fa—
cilitate comprehension.

Stressed target sentences preceded by unstressed context sentences
were comprehended faster than unstressed target sentences preceded by
this type of context sentence. It appears that comprehension may be
facilitated when new information is explicitly pointed out to a listener
via stress, perhaps because the presence of stress lengthens a sentence
and allows more time for comprehension. However, evidence can also be
presented from this experiment indicating that sentences are not made
easier to comprehend simply by placing heavy stress on an item in the
sentence. Stressed context sentences were comprehended no more rapidly
than unstressed context sentences. This indicates that in the absence
of other factors, the higher pitch and amplitude of stressed items
and/or the extra processing time they afford cannot in itself account
for appropriately stressed target sentences being comprehended faster
than unstressed (but appropriate) target sentences. However, it may be
that after a listener hears a context sentence and encounters ambiguity
in establishing what information is new and what is given, the presence
of a stressed item in the target sentence helps the listener resolve

this ambiguity and facilitates comprehension. It is unclear at this

23
time what operations a listener may perform in order to disambiguate
unstressed sentences, and it is also unclear what effect the extra
processing time afforded by stressed sentences actually has on compre-
hension.

Results of this experiment Show that inappr0priately stressed tar—
get sentences were comprehended about as rapidly as unstressed target
sentences preceded by unstressed context sentences. Examples of these
types of sentences are given in (28c) and (28d), respectively, which are
reprinted below with mean RT for the target sentences in each. Although

(28c) George didn't kill fl. 533391 killed Bill. (137 msec)

(28d) George didn't kill Bill. Alfred killed Bill. (1113 msec)
the stress in the target sentence of (280) is inappropriate and the
sentence may have to be restructured in some way to be understood, it
was still not significantly more difficult to understand than the un—
stressed target sentence in (28d), which does not appear to be inappro-
priate. Perhaps the greater length of the target sentence in (280)
allowed adequate time for the extra operations needed for comprehension.
It is also possible that the lack of stress in the target sentence of
(28d) made the identification of given and new information more diffi-
cult and slowed down comprehension. It is impossible to argue for the
merit of one of these explanations over the other given the conditions
of this experiment; thus, reasons for the relatively slow comprehension
time for the target sentence of (28d) remain unresolved. However, if
an experiment could be performed in which the length of the context
sentences in (28c) and (28d) were identical, the amount of processing
time afforded subjects by stressed and unstressed sentences could be

controlled, and more light would be shed on the results obtained here.

24
The feasibility of such an experiment would depend on whether the length
of a stressed context sentence such as the one in (28c) could be com-
pressed to equal the length of the unstressed context sentence in (28d)
without the sentence sounding unnatural.

The results of the present experiment lend support to the notion
that a listener processes intonational features in the course of come
prehending a sentence. Results indicate that placement of contrastive
stress can greatly influence the speed of sentence comprehension depen-
ding on whether an appropriate or an inappropriate item is indicated
(through the use of stress) as the correct candidate (new information)
to replace an incorrect candidate in a given role. In Experiment I,
given roles were established in context sentences through the use of
contrastive stress. Experiment II examined unstressed context sen-
tences with.target sentences which require as an antecedent for the
contrastive role component the same role (as in Experiment I) and a
different role element (also as in Experiment I); however, in the types
of sentences used in Experiment II, the new role element does not re-

place the old but is instead added to it.

EXPERIMENT II

As discussed above, Chafe (1976) contends that a speaker who utters
the contrastive sentence Eggn§l9_made the hamburgers" is assuming that
the listener knows that someone made the hamburgers and therefore treats
this information as given. The correct candidate (Ronald) for the role
of hamburger maker is presented as new information and receives heaviest
stress.

A simple active sentence spoken under normal intonation contains no
specific markers of presupposition and new information is generally con-
sidered to be found at the end of the sentence (Halliday, 1967; Chafe,
1970). Therefore, upon hearing (29), it would in general be assumed

(29) Annie slew a gnu too.
that the presupposed information (which would be earlier in the sen-
tence) is that Annie slew something else. However, if (29) were uttered
as in (30), the higher pitch on "Annie" would identify that item as new

(30) Angig slew a gnu too.
information, leaving as presupposed that someone else slew a gnu, or in
other words that someone else is also a correct candidate for the given
role of gnu slayer.

If stressing an item in fact indicates that it contains new.infor-
mation, then (30) should be comprehended faster when found in the con-
text of a sentence like (31), which contains a direct antecedent for the

(31) Debbie slew a gnu.

25

26
presupposed information that someone else slew a gnu, than when found in
the context of (32), which does not contain a direct antecedent for the

(32) Annie slew a yak.
given information of (30). In other words, the target sentence in (33)
should be comprehended faster than the target sentence in (34) in which

(33) Debbie slew a gnu. Annig slew a gnu too.

(34) Annie slew a yak. ‘Aggig slew a gnu too.
the stressed item carries information that is given in the context
sentence.

In (33), "Annie" is presented in the target sentence as new infora
mation, indicating that She too is a gnu slayer, a role which is consid—
ered given information. That role was set up in the context sentence,
where Debbie was pointed out as a correct candidate for the role. The
new information of the target sentence indicates that Annie is also an
appropriate candidate for the role. The new information of the target
sentence indicates that Annie is also an appropriate candidate for the
role, and therefore it appears that in these sentences the new informa-
tion (Annie) is added to previously given information (Debbie) in a role
established in the context sentence.

Likewise, in the target sentence of (34) Annie is presented as an
additional correct candidate for the (supposedly given) role of gnu
slayer. However, such a role was not established in the context sen-
tence of (34). Rather, the role of yak slayer was set up, with Annie
as the correct candidate for it. In the target sentence of (34) "Annie"
is given information by virtue of the information presented in the sen—
tence preceding it. Therefore, it is the fact that a gnu was slain in

addition to a yak that is new information in the target sentence, and

27

Annie is given information. If stressing a sentence item does indicate
that it is new information, then stressing "Annie" in the target sen-
tence of (34) is inappropriate. The inappropriately stressed target
sentence in (34) should be comprehended more slowly than the same sen-
tence in (33), which is stressed appropriately given its context.

Experiment II tested this prediction by presenting subjects with
pairs of prerecorded sentences in which the target sentence'of each pair
was a simple active sentence containing one of the adverbs "still",
"either", "again", or "too", and by manipulating the locus of heaviest
stress in these target sentences. The context sentence of each pair
contained either an appropriate antecedent for the stressed information
of the target sentence, as in (35), or contained an inappropriate ante-
cedent for that information, as in (36). It was predicted that it would

(35) Jane had a cold. M111 had a cold too.

(36) Mary had the measles. 'M§£y_had a cold too.
take listeners more time to comprehend the target sentence in (36) than
in (35). Likewise, the target sentence in (37) was expected to be com-
prehended faster than the same sentence in (38).

(37) Mary had the measles. Mary had a.ggld too.

(38) Jane had a cold. Mary had a cold too.

Bear-d

Subjects. Subjects were the 40 participants described in Experi-
ment I.

Materials. The verbal materials consisted of 16 sets of sentences
of the same form as sentence pairs (39)-(42), which are representative

of one sentence set (using the adverb "either"). Each set consisted of

28

(39 Fred can't play the violin. Bob can't play the violin either.

40 Bob can't play the piano. Bob canft play the violin either.

41 Bob can't play the piano. Bob can't play the violin either.

42 Fred can't play the violin. Bob can't play the violin either.
4 sentences pairs which varied with respect to locus of heaviest stress
and whether or not the context sentence contained an appropriate ante-
cedent for the stressed information of the target sentence. Type of
context sentence and locus of heaviest stress were varied factorially
to produce 64 defferent sentence pair/intonation combinations. There-
fore each set contained on sentence pair from each of the following
conditions:

a) Appropriate Antecedent, Subject (AS): Context sentence contain-
ing an appropriate antecedent for stressed information of
the target sentence; heaviest stress on subject

b) Appropriate Antecedent, Object (A0): Context sentence contain-
ing an appropriate antecedent for stressed information of
the target sentence; heaviest stress on object

o) Inappropriate Antecedent, Subject (IS): Context sentence con—
taining an inappropriate antecedent for stressed informa-
tion of the target sentence; heaviest stress on subject

d) Inappropriate Antecedent, Object (IO): Context sentence con-
taining an inappropriate antecedent for stressed informa-
tion of the target sentence; heaviest stress on object

The 16 sentence sets, plus 3 additional sentence pairs used in practice
trials at the beginning of the experiment, are presented in Appendix C.
Four lists of 16 sentences each were constructed in the following
way. A random.order of the 16 sentence sets was chosen and held con-
stant over lists. The 16 sentence pairs of List 1 were made up of 4
sentence pairs from each of the 4 (AS, A0, IS and IO) conditions, selec-
ted at random. Lists 2-4 contained the same ordering of the 16 sen-
tences as List 1, and were constructed so that over the 4 lists each
sentence was represented as an AS, A0, IS, and IO sentence exactly once.

Therefore each list contained 4 sentence pairs from each of the 4 con-

ditions. Each list of 16 sentence pairs was preceded by the same 3

29
practice sentence pairs (i.e. pairs i-iii in Appendix C). The 4 lists
are presented in Appendix D.

All sentences were recorded by the experimenter on Scotch brand
magnetic tape using a Revox A-700 tape recorder. Only AS and A0 sen-
tences were actually recorded. These sentences were then used, in a
different order, for the sentences in the IS and IO conditions. In
this way the recorded target sentences in the AS and IS conditions were
identical, as were the target sentences in the A0 and IO conditions.

As in Experiment I, all sentences which had been recorded were digitized
and the digital form of each sentence was stored on magnetic disc.

Using the PDP=11/40 domputer, 4 lists were constructed by playing the
sentences back in the proper order for each list through the digital to
audio converter, and recording them on Scotch brand magnetic tape using
the Revox A-700 tape recorder. The final recording consisted of 4 lists
of the original 16 sentence pairs (and the 3 practice sentence pairs) in
the proper order for each list.

The rec0gnition test used in Experiment II is found in Appendix E.

Procedure. The procedure followed in conducting this experiment

was the same as in Experiment I.

Results
Mean reaction time (RT) for each type of target sentence (AS, A0,
IS, and I0) was computed, collapsing across lists and subjects. These

mean RTs are presented in Table 2.

A 2 x 2 x 40 x 4 repeated measures analysis of variance was carried
out on the data. The major comparison of interest was the comparison of

mean RT for the Appropriate Antecedent condition with the Inappropriate

30

Table 2. Mean reaction time to target sentences in Experiment II.

 

Condition RT (in msec)
AS 137
A0 206
IS 270
IO 198

 

 

 

Subject
----- Object

300-
1?

3 250”
s
c
0H

" 200‘
Q)
s
w-l
.p

o 150‘
O
.H
+3
8

m 100.1
s:

503

Appropriate Inappropriate

Antecedent Antecedent

Figure 1. Mean reaction time to target sentences in Experiment II.

31
Antecedent condition. It was predicted that mean RT in the Appropriate
Antecedent condition would be significantly faster than in the Inappro-
priate Antecedent condition. Eran RT for the Appropriate Antecedent
condition (171 msec) was in fact faster than mean RT for the Inappropri-
ate Antecedent condition (234 msec), F(1,39)=- 4.83, p<.05. Thirty of
the 40 subjects in the experiment showed this effect. No main effect of
Subject versus Object was found; however an interaction between the Ap—
propriate Antecedent/Inappropriate Antecedent and Subject/Object factors
occurred, F(1,39)= 13.62, p<.001 (see Figure 1). This interaction re-
veals that the main effect of the Appropriate Antecedent/Inappropriate
Antecedent factor is a result of a difference between the AS and IS con-
ditions; mean RT in the A0 condition was not significantly different
from mean RT in the IO condition.

Memory test performance was .90, indicating that subjects did come
prehend the sentences presented to them. Performance of subjects who
received the experiments in the II-I order was .88 for Experiment II,
while performance of subjects receiving the experiments in the I-II
order was .92 for Experiment II. Memory test performance was slightly,
but insignificantly, better in Experiment II when this experiment was
the second one presented. Therefore, it appears that anticipation of a

memory test made little or no difference in rec0gnition performance.

Discussion

 

In this experiment, subjects were presented with sentence pairs in
which the context sentence contained either an appr0priate or an inap-
propriate antecedent for the stressed information of the target sen-

tence. It took subjects longer to comprehend target sentences in the

32
Inappropriate Antecedent condition than in the Appropriate Antecedent
condition when the stressed item was the subject of the sentence. How-
ever, when the stressed item was the object of the sentence, there was
no difference between comprehension time of Appropriate Antecedent and
Inappropriate Antecedent condition target sentences. In other words, a
target sentence such as the one in (39) was comprehended faster than the
same sentence in (41), but a target sentence such as the one in (40) was
not comprehended faster than the same sentence in (42). These results
do not fully support the prediction that sentence comprehension will be
facilitated when an item carrying new information (in the context of a
preceding sentence) receives heavy stress, and that comprehension will
be slowed down by stressing an item (i.e. designating an item as new
information) which carries given information in the context of a previ-
ous sentence.

Since all sentences used in this experiment were active, the under-
lying subject of each sentence always occurred at the beginning and the
underlying object was found at the end of the sentence. It appears that
the position of the stressed item in the target sentence is an important
variable influencing comprehension time in this experiment, but the con-
founding of underlying subject and object with surface position requires
further investigation.

These results suggest that a model describing the way listeners
comprehend the types of sentences presented should incorporate both of
the following factors: (1) the type of information (given or new)
which receives heaviest stress in the sentence, and (2) the position of
the given and new information in the sentence. It appears that compre-

hension is facilitated when new information is given heaviest stress

33
but slowed down when it is the given information which is stressed.
This effect was predicted in Experiment II. However, it also appears
that comprehension is facilitated when new information occurs at the
beginning of a sentence, but slowed down when given information is pres-
ent in that position. This finding is contradictory to the traditional
production hypothesis (Halliday. 1967; Chafe, 1970) that new information
is usually presented at the end of simple active sentences.

Mean comprehension time was fastest in the AS condition (137 msec),
slowest in the IS condition (270 msec), and intermediate in the AC (206
msec) and I0 (198 msec) conditions. In the fastest (AS) condition, the
new information is stressed and is found at the beginning of the sen-
tence; both of these could be considered facilitating (+) factors. In
the slowest (IS) condition, the given information is stressed and the
new information is found at the end of the sentence; both could be con-
sidered as inhibiting (-) factors. In the two conditions with inter-
mediate comprehension times, one facilitating and one inhibiting factor
are present: in the A0 condition the new information is stressed (+)
but is found at the end of the sentence (-), and in the I0 condition
the new information is at the beginning of the sentence (+) but the
given information is stressed (-). Examples of sentences in each of the
4 conditions are given in (43)-(46):

43 AS EVelyn kissed Jeremy. Rhonda kissed Jeremy too. (+ +)
£44? A0 Rhonda kissed Jason. thfidg—kissed ggggmy too. (+ -;
45 IS Rhonda.kissed Jason. ‘Rhggdg kissed Jeremy too. (— -

(46) IO Evelyn kissed Jeremy. Rhonda kissed Jgggmy too. (- +)

To determinewhether the two factors described above (i.e. the type
of information (given or new) which is given heaviest stress, and the

position of given and new information in the sentence) have the predic-

ted effects on comprehension time, it would be necessary to repeat

34

Experiment II with several added conditions. In an active sentence the
underlying subject is found at the beginning and the underlying object
at the end of the sentence. In a passive sentence the underlying sub-
ject is found at the end of the sentence while the underlying object
occurs at the beginning. Since only active sentences were used in this
experiment, there was no control for position of underlying subject and
object in the sentences. If passive sentences such as (47)-(50) were

E47) Evelyn kissed Jeremy. Jeremy was kissed by Rhggdg too. (+ -)

483 Rhonda kissed Jason. Jgggmy was kissed by Rhonda too. E+ +g
(49 Rhonda kissed Jason. Jeremy was kissed by Rhggdg too. - +

(50) Evelyn kissed Jeremy; £23321 was kissed by Rhonda too. (- -)
to be used, the original predictions made in Experiment II should be
supported when the stressed item is the object of the sentence but not
when it is the subject. This result would be the opposite of that found
in Experiment II, and would support the notion of incorporating a posi-
tion factor in a description of the comprehension process. The relative

comprehension times listed in Table 3 are predicted for target sentences

such as those in (43)-(46) and (47)-(50).

Table 3. Relative reaction times for active and passive target

 

sentences.
Active Sentences Passive Sentences
Relative RT . Relative RT
Condition (Obtained-Experiment II) Condition (Predicted)
AS (+ +) Fastest (137 msec) AS (+ -) Intermediate
A0 (+ -) Intermediate (206 msec) A0 (+ +) Fastest
IS (- -) Slowest (270 msec) IS (- +) Intermediate
IO (- +) Intermediate (198 msec) IO (- -) Slowest

In Experiment II, sentences having their new information at the

'beginning were comprehended faster than sentences with their new

35

information at the end. Since all sentences were of the same form as
those in (51)-(54), it is possible that after several trials a subject

(51) AS Jane had a cold. 'Mggy had a cold too.

E52 A0 Mary had the measles. Mary had a.ggld too.
53) IS Mary had the measles. ‘Mgry had a cold too.
54 IO Jane had a cold. Mary had a 3235; too.
could "comprehend" a target sentence once its new information was heard,
since the given information was always information which had been given
in the context sentence. In other words, if new information was located
at the beginning of a target sentence, as in (51) and (54), a subject
might be able to accurately predict the rest of the sentence and could
press the button indicating comprehension before the target sentence had
been completed. If new information were found at the end of the sen—
tence, as in (52) and (53), subjects might be less likely to respond
prematurely because the end of the sentence could not be accurately pre-
dicted. Although there is the possibility that this type of subject
strategy was responsible for the faster RT results for target sentences
with new information at the beginning, this does not appear to have been
the case in this experiment. If subjects were actually responding more
quickly to target sentences with their new information at the beginning
because the given information of the sentence wasn't necessary for com-
prehension, more premature responses would be expected to those target
sentences with their new information at the beginning than to target
sentences with their new information at the end.

The data analyzed in this experiment were the responses of 40 sub-
jects to 16 target sentences each, for a total of 640 responses.- Of the
640 responses obtained, 143 of these were made before the end of the

target sentence. Of these 143 premature responses, 82 of them, or 57%

were responses to sentences with the new information at the beginning

36

of the sentence. In other words, subjects were almost as likely to re-
spond prematurely to target sentences with new information at the begin-
ning as they were to sentences with the new information at the end (i.e.
differential responding was close to chance level).

The set of 82 premature responses to target sentences with their
new information at the beginning, and the set of 61 premature responses
to target sentences with their new information at the end were analyzed
to determine if subjects' responses were more premature in the former
set than the latter. Mean RT for the former set was -390 msec, while
mean.RT for the latter set was -319 msec (RTs measured from the physi-
cal end of the sentence). A MannAWhitney‘U analysis was carried out on
these data, and no significant difference was found between the two sets
of responses (z=.14, n.s.). The Mhnnrwhitney test was considered appro-
priate because the distribution of premature responses was nonnormal.

There was no significant difference between the sets of premature
responses in the AS condition (i= -389 msec, n=42) and the I0 condition
(2- -390 msec, n=40), z- -.45, n.s.; there was also no significant dif-
ference between the sets of premature responses in the A0 condition (i=
-335 msec, n=33) and the IS condition (5::- -3oo msec, n-28), z- -.31,
n.s. The former conditions (AS and IO) had their new information (i.e.
information not recoverable from the context sentence) at the beginning
of the target sentence and the latter conditions (A0 and IS) had their
new information at the end; the stressed item in the AS and IS condi-
tions occurred at the beginning of the target sentence, while inthe A0
and I0 conditions the stressed item was at the end of the target sen-
tence. The above analysis by condition yields no evidence that either

the location of stress or the location of new information influenced

37
the occurrence of premature responses.

Little evidence can be derived from this experiment for the possi-
bility that subjects responded more rapidly to target sentences with
new information at the beginning because they realized the given infor-
mation which followed was unnecessary for comprehension. However, in
future experiments the precautionary measure could be taken of including
target sentences which contain all new information, as in (55)-(58).

(55) Jane had a cold. ‘Mggy had the measles too.

(56) Mary had the measles. Jane had a cold too.

(57) Mary had the measles. Jane had. a ITO—Id too.

(58) Jane had a.cold. Mary had the measles too.

Including such sentences presumably would prevent subjects from using
the response strategy discussed above, since they would be unable to

predict whether given or new information was going to follow initial

new information.

GENERAL DISCUSSION

The two experiments in this study were designed to consider the
role of stress in sentence comprehension. It was postulated that a par-
ticular communicative function is served by the appropriate use of
stress in that such use can facilitate a listener's comprehension of
certain types of utterances.

The phenomenon of contrastive stress is a conspicuous form of

38
phonetic highlighting which appears to have a significant effect on
sentence comprehension. Contrastive stress plays an important communi-
cative function by focusing a listener's attention on a particular sen-
tence item.and by indicating how the listener should interpret that
sentence. In a sentence such as "George didn't kill Bill", contrastive
stress is used to indicate the sentence's scope of negation and that a
particular role is given information. In a sentence such as "George
killed Alfrgdf, contrastive stress is used to mark a particular item as
new information carrying the correct candidate for the role in question,
thereby excluding other candidates the listener may have been consider-
ing. It appears that contrastive stress may have this dual function of
indicating both given and new information in these types of sentences.
Results indicated that proper use of contrastive stress can facilitate
comprehension, and that inappropriate use makes comprehension more dif—
ficult for listeners. Contrastive stress appears to be an important
factor in determining the speed at which sentences are understood.

In Experiment II, context sentences were unstressed and the new in-
formation of target sentences was designated by placing heaviest stress
on a particular sentence item. The item stressed was either information
which had an appropriate antecedent in the preceding sentence, or infor-
mation which had an inappropropriate antecedent. An appropriate antece-
dent was one to which the new (stressed) information of the target sen-
tence could be added; i.e. an appropriate antecedent would carry the
same role as the new information but a different (although not incorrect)
role element. It was predicted that stressing a target sentence item
with an inappropriate antecedent would slow down comprehension. This

occurred when the stressed item was the subject of the sentence but did

39

not occur when it was the object. Position of the item in the sentence
could be an important factor contributing to such a result. A future
experiment using both active and passive sentences could control for the
position factor and shed more light on the results obtained here.

Haviland and Clark (1974) contend that when a direct antecedent is
presented for the given information of a sentence, that sentence will be
easier to comprehend than if a direct antecedent were not presented.
Haviland and Clark's analysis is different from the one presented here
in that the present study assumes that new information must have an ap-
propriate antecedent (i.e. a given role) if comprehension is to proceed
smoothly. It is not possible to determine at this time whether listen—
ers' comprehension is slowed down (in the conditions considered here as
being inappropriate) because the antecedent for the new (stressed) in-
formation is inappropriate, or alternatively because the antecedent for
the given (unstressed) information is inappropriate.

Haviland and Clark (1974) presented to subjects for comprehension
sentences such as (59) in which presupposition is indicated by the

(59) Annie slew a gnu too.
adverb "too". The given information of this sentence (on one reading)
is that someone else slew a gnu. In Experiment II of the present study,
target sentences such as (60) were presented in which both the adverb

(60) 59313 slew a gnu too.
and the presence of stress served to indicate presupposition, or what
information was given. It is unclear how the Haviland and Clark model
would handle information which is given by virtue of the existence of a
role, when that existence is implied by contrastive stress. Therefore,

the present description of sentence comprehension may not be inconsistent

40
with the Haviland and Clark model.

Haviland and Clark (1974) found that a target sentence whose given
information had a direct antecedent in a preceding sentence took less
time to comprehend than if the preceding sentence contained no direct
antecedent for the given information. These authors used reading time
as the measure of comprehension in their experiments, whereas the meas-
ure used in the present experiments was reaction time to aurally presen-
ted sentences. Perfetti and Lesgold (1977) have argued that comprehen-
sion by listening and comprehension by reading are highly similar at
some level of analysis. Jackson and McClelland (1979) found that faster
readers performed more accurately in a listening comprehension test than
did poorer readers. These authors suggest that the ability to compre-
hend spoken material is an important correlate of reading ability. That
the results of the present experiment (in which an auditory task.was ems
ployed) were similar to the results of Haviland and Clark's experiments
(in which a visual task.was used) lends support to such a claim.

Sequences such as (61) were presented to listeners in Experiment I.

(61) ge_orgp_ didn't kill Bill. m killed Bill.

That the role of Bill's killer exists is established in the context sen-
tence, and in the target sentence Alfred is indicated as the correct
candidate for the role, replacing George (an incorrect candidate).

Sequences such as (62) and (63) were presented in Experiment II.

(62) Jane had a cold. _Ma_ry had a cold too.

(63) Fred can't play the violin. gap can't play the violin either.
In the target Sentence of (62), Mary is new information which is pointed
out as an additional correct candidate for the role of "person who had

a col ". In the target sentence of (63), Bob is new information which

41
is pointed out as a correct candidate for the role of "someone who can't
play the violin", in addition to Fred.

In both experiments, stress was used to indicate new information.
However, in the types of sentences used in Experiment I, the new infor-
mation of a target sentence replaced information from a previous sen-
tence, whereas in the types of sentences used in Experiment II, this new
information added to, rather than replaced, previous information. It
appears that sentence stress may function differently in different types
of sentences.

As these experiments have demonstrated, the phenomenon of stress
plays a major role in the communication of ideas between speakers and
listeners. It appears that stress is an important factor which should
be more fully considered in future psycholinguistic research, and should

be incorporated into models of sentence comprehension.

FOOTNOTES

FOOTNOTES

1Chafe (1976) objects to Haviland and Clark's (1974) notion that
given information is information that the listener already knows about
and new information is information not yet known to the listener. Chafe
argues that given information should be thought of as "that knowledge
which the speaker assumes to be in the consciousness of the addressee at
the time of the utterance" and new information as "what the speaker as-
sumes he is introducing into the addressee' s consciousness by what he
says" (p. 30). Chafe points out that a speaker may assume that a listen-
er is not thinking about certain information at the time, even though it
is actually "known" by the listener (for example, the existence of a
particular person), and that presenting it as new information serves to
activate it in the listener's knowledge store. Haviland and Clark refer
to given and new information in terms of what the listener does and does
not "know", but it is often clearly the case that the listener does have
some prior knowledge of some aspect of the information presented as new.
Chafe argues that the terms "already activated" and "newly activated"
are actually more appropriate, but yields to the less awkward terms
"given" and "new", pointing out that the distinction should be kept in
mind.

2The sentence "It was Agnes who saw Maxine" is an example of the
cleft sentence structure. This type of structure was termed "cleft" by
Jesperson (1961): "Cleft sentences. A cleaving of a sentence by means
of it is (often followed by a relative pronoun or connective) serves to
single out one particular element of the sentence and very often, by di-
recting attention to it and bringing it, as it were, into focus, to mark
a contrast" (pp. 147-148). The so-called "pseudo-cleft" sentence struc-
ture is reflected in a sentence such as "The one who saw Maxine was
Agnes", which again directs attention to the new information (Agnes) by
making it the focus of the sentence.

3Sentence items which should be read with heaviest stress will be
underlined in the examples used in this paper.

“Cutler and Fodor (1979) also controlled for the position of the
target-bearing item in the sentence by having half the subjects detect a
phoneme in the first part of the sentence and the other half detect a
phoneme occurring in the latter part of the sentence. For example, in
(18) half the subjects would be iven the phoneme /k/ as a target and
half would be given the phoneme 7b/

(18) The man on the corner was wearing the blue hat.

42

43

5The highest stress level in the first and second sentences in
each of these sequences may not necessarily be the same.

6A target sentence containing all new information would presumably
be very difficult for subjects to comprehend since the context sentence
would contain no antecedent for any information in the target sentence.

APENDIX A

ii. Ralph doesn't go to college.

APPENDIX A

Experiment I Sentence Materials

i. The cat isn't on the roof.

iii.

1.

a.
b.
c.
d.

a.
b.
c.
d.

.a.

b.
c.
d.

a.
b.
c.
d.

.8.-

b.
c.
d.

a.
b.
c.
d.

The house isn't made of wood.

George didn't kill Bill.
George didn't kill Bill.
George didn't kill Bill.
George didn't kill Bill.

Lygg didn't go to
Lynn didn't go to

Lynn didn't go to

Lynn didn't go to

Arnold didn't
Arnold didn't
Arnold didn't
Arnold didn't

Dennis wasn't
Dennis wasn't
Dennis wasn't
Dennis wasn't

Henry doesn't
Henry doesnit
Henry doesn't
Henry'doesn't

Frank isn't growing a
Frank.isn't growing a
Frank.isn't ggowing a
Frank isn't growing a

Ray didn't play football in college.

college.

Ray didn't play football in college.

college.

Ray didn't play football in college.

college.

Ray didn't play football in college.

college.

fix
fix
fix

drinking warm milk.
drinking warm milk.
drinking warm milk.
drinking warm milk.

have
have
have
have

The cat is in the tree.

Ralph goes
The house

killed
killed
killed

arses.
sass
Alfred

 

Brazil.
Brazil.
Brazil.

Brazil.

the radio.
the radio.
.___ the radio.
fix the radio.

Alfred killed
went to
went to
went to
went to

Joan
Joan
Joan
Joan

 

Doris
Doris
Doris
Doris

 

Res.
Pee
Pee
Dan

 

to high school
is made of brick.

Bill.
Bill.
Bill.
Bill.

Brazil.
Brazil.
Brazil.
Brazil.

an older Sister. Henry has an
an older sister. Henry has an
an older Sister. Henry has an
an older sister. Henry has an
beard. Frank is growing a
beard. Frank is growing a
beard. Frank.is growing a
beard. Frank is growing a
Ray played

Ray played

Ray played

Ray played

an

fixed the radio.
fixed the radio.
fixed the radio.
fixed the radio.

was drinking warm milk.
was drinking warm milk.
was drinking warm milk.
was drinking warm milk.

older brother.
older brother.
older brother.
older brother.

moustache.
moustache.
moustache.
moustache.
baseball in
baseball in
baseball in

baseball in

10.

11.

12.

13.

14.

15.

16.

17.

a.
b.
c.
d.

a.
b.
c.
d.

a.
b.
c.
d.

a.
b.
c.
d.

a.
b.
c.
d.

a.
b.
c.
d.

a.
b.
c.
d.

a.
b.
c.
d.

a.
b.
c.
d.

45

Jack isn't reading a newspapgr.
Jack isn't reading a newspaper.
Jack isn't reading a newspaper.
Jack isn't reading a newspaper.

Jill doesn't
Jill doesn't
Jill doesn't

take dancing lessons.
take dancing lessons.
take dancing lessons.

Jack is
Jack is
Jack is
Jack is

Jill
Jill
Jill

 

 

 

reading a gggazin .
reading a Eggazine.
reading a Eggazin .
reading a magazine.

takes singing lessons.
takes singing lessons.
takes singing lessons.

Jill doesn't take dancing lessons. Jill takes singing lessons.
Marcia isn't wearing a Egg coat. Marcia is wearing a blue coat.
Marcia isn't wearing a red coat. Marcia is wearing a blue coat.
Marcia isn't wearing a red coat. Marcia is wearing a blue coat.
Marcia isn't wearing a red coat. Marcia is wearing a blue coat.
Betty didn't buy a 225 car from.Mike. Betty bought a used car
from Mike.

Betty didn't buy a new car from Mike. Betty bought a used car
from.Mike.

Betty didn't buy a new car from Mike. Betty bought a used car
from Mike.

Betty didn't buy a new car from Mike. Betty bought a used car
from.Mike.

The plastic bottle didn't break. The glass bottle broke.

The plastic bottle didn't break. The glass bottle broke.

The plastic bottle didn't break. The glass bottle broke.

The plastic bottle didn't break. The glass bottle broke.

Harry doesn't live near Chicgg . Harry lives near Detroit.
ibury’doesn't live near Chicago. Harry lives near Detroit.
Harry doesn't live near Chicago. Harry lives near Detroit.
Harry doesn't live near Chicago. Harry lives near Detroit.

The hockey game isn't on Tuesd y. The hockey game is on Frid y.
The hockey game isn't on Tuesday. The hockey game is on Friday.
The hockey game isn't on Tuesday. The hockey game is on Friday.
The hockey game isn't on Tuesday. The hockey game is on Friday.
The officer wasn't in the Army. The officer was in the Navy.
The officer wasn't in the Army. The officer was in the Navy.
The officer wasn't in the Army. The officer was in the Nayy.
The officer wasn't in the Army. The officer was in the Navy.
John didn't get a job in Michigan. John got a job in Wisconsin.
John didn't get a job in Michigan. John got a job in Wisconsin.
John didn't get a 122 in Michigan. John got a job in Wisconsin.
John didn't get a job in Michigan. John got a job in Wisconsin.
Marvin doesn't 233 the building. Marvin rents the building.
Marvin doesn't own the building. Marvin rents the building.
Marvin doesn't own the building. Marvin rents the building.
Marvin doesn't own the building. Marvin rents the building.

18.

19-

20I

a.
b.
c.
d.

a.
b.
c.
d.

a.
b.
c.
d.

Alice didn't bgy the banana bread.
buy the banana bread.
buy the banana.bread.
buy the banana bread.

Alice didn't
Alice didn't
Alice didn't

direct the
direct the
direct the
direct the

Ralph didn't
Ralph didn't
Ralph didn't
Ralph didn't

Jane didn't get married
Jane didn't get married
Jane didn't get married
Jane didn't get married

46

movie.
movie.

movie.

movie.

in June.
in June.
in June.
in June.

Alice
Alice
Alice
Alice

baked
baked
baked
baked

 

Ralph pgoduced the
Ralph pgoduced the
Ralph pgoduced the
Ralph produced the

the banana bread.
the banana bread.
the banana bread.
the banana bread.

movie.
movie.
movie.
movie.

Jane
Jane
Jane
Jane

got divorced in June.
got divorced in June.
got divorced in June.
got divorced in June.

APPENDIX B

APTENDIX B

Rec0gnition Test for Experiment I

Read each of the-sentences below and decide whether or not it is

one that you just heard. In the space after each sentence, write "Yes"
if it is a sentence you heard and "No" if you did not hear the sentence.

\Om'flmkn-{Z‘WNH

10.

. Ray played baseball in college.

The hockey game is on Friday.
Lee was eating french fries.
Ray bought a new television.
Frank is growing a moustache.
The glass bottle broke.
Alice baked the banana bread.
Kay is growing tomato plants.
The accountant gave flowers to Michelle.
Ralph produced the movie.

Henry has an older brother.

 

 

 

 

 

 

 

 

 

 

. Jean works for a dentist.

 

Dan was drinking warm milk.
The policeman arrested the thief.
The chief led the Indians.

The student is going to Hawaii.
John got a job in Wisconsin.
Dale went to church on Sunday.
Jill takes singing lessons.
Jane got divorced in June.

The old woman sold the painting.
Alfred killed Bill.

The book was in the library.
Sam wrote the term paper.
Jack is reading a magazine.

The officer was in the Navy.
Marshall has two daughters.

Doris fixed the radio.

Nick was in the Marines.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

. Tammy is married to a carpenter.

 

Harry lives near Detroit.
Joan went to Brazil.

The manager fired the clerk.
Beth was driving a school bus.
Marvin rents the building.
Gina tore her new skirt.

 

 

 

 

 

 

 

47

37. Marcia is wearing a blue coat.

 

38. The tennis palyer broke his ankle.
39. Betty bought a used car from Mike.
40. Rose went to the play.

 

 

 

APPENDIX C

i

ii.

iii

1.

APTENDIX C

Experiment II Sentence Materials

. Linda doesn't have an Oldsmobile. Linda doesn't have a Cadillac
either.

Brenda ordered a hamburger. Jennifer ordered a hamburger too.

. The student doesn't take biology anymore. The student still takes
chemistpy.

(A5;

(A0
(I3
IO
(AS)
(A0)
(IS)
(IO)

AS)

Jane had a cold. May had a cold too.
Mary had the measles. Mary had a cold too.
Mary had the measles. ‘Mg£y_had a cold too.
Jane had a cold. Mary had a cold too.

The butcher joined the Credit Union. The tailor joined the
Credit Union too.

The tailor joined the Men's Club. The tailor joined the Credit
Union too. ”"_"

The tailor joined the Men's Club. The tailor joined the Credit

Union too.
The butcher joined the Credit Union. The tailor joined the

Credit Union too.

The Joneses went to Cleveland. The Smiths went to Cleveland too.

A0) The Smiths went to Pittsburgh. The Smiths went to Cleveland too.

(15)
(IO)

B
AO
IS)
(IO)
1“”
A0
IS
a)
(As)
(A0)
(IS)

(10}

The Smiths went to Pittsburgh. The Smiths went to Cleveland too.
The Joneses went to Cleveland. The Smiths went to Cleveland too.

 

Carl was drinking beer. Paul was drinking beer too.
Paul was drinking whiskey. Paul was drinking beer too.
Paul was drinking whiskey. Paul was drinking beer too.
Carl was drinking beer. Paul was drinking beer too.

Fred can't play the violin. Bob can't play the violin either.
Bob can't play the piano. Bob can't play the violin either.
Bob can't play the piano. Bob can't play the violin either.
Fred can't play the violin. Bob can't play the violin either.

The lawyer doesn't like apples. The doctor doesn't like apples

either.

The doctor doesn't like bananas. The doctor doesn't like apples
either.

The doctor doesn't like bananas. The doctor doesn't like apples
either.

The lawyer doesn't like apples. The doctor doesn't like apples

either. 49

10.

11.

12.

13.

50

(AS) Dave doesn't know how to ski. John doesn't know how to ski
either.
(AD) John doesn't know how to ice skate. John doesn't know how to

ski either.

(IS)
(Io)

ski either.
either.

(AS)
either.

(A0) Rachel didn't give Mark a card.

Dave doesn't know how to ski.

Diane didn't give Mark a gift.

John doesn't know how to ice skate.

John doesn't know how to

John doesn't know how t°.§§l

Rachel didn't give Mark a gift

Rachel didn't give Mark a gift

 

either.

(IS) Rachel didn't give Mark a card. Rachel didn't give Mark a gift
either.

(IO) Dian didn't give Mark a gift. Rachel didn't give Mark a gift
either.

(AS Pat doesn't have a collie anymore. Ann still has a collie.

(A0 Ann doesn't have a beagle anymore. Ann still has a collie.

(IS Ann doesn't have a beagle anymore. Ann still has a collie.

(IO) Pat doesn't have a collie anymore. Ann still has a collie.

(AS; Barry can't speaszussian anymore. Vivian still speaks Russian.

A0 Vivian can't speak German anymore. Vivian still speaks Russian.

(18% Vivian can't speak German anymore. Vivian still speaks Russian.

IO Barry can't speak Russion anymore. Vivian still speaks Russian.

(AS) Marilyn doesn't like to knit anymore. Kathy still likes to
knit.

(A0) Kathy doesn't like to crochet anymore. Kathy still likes to
knit.

(IS) Kathy doesn't like to crochet anymore. Kathy still likes to
knit.

(IO) Marilyn doesn't like to knit anymore. Kathy still likes to
knit. .

AS; The floor isn't dirty anymore. The window is still dirty.

A0 The window isn't broken anymore. The window is still dirty.

18% The window isn't broken anymore. The window is still dirty.

(ID The floor isn't dirty anymore. The window is still dirty.

(AS) Bill never returned to the drug store. Carol went to the drug
store again.

(A0) Carol never returned to the shoe store. Carol went to the
drug store again.

(IS) Carol never returned to the shoe store. Carol went to the
drug store again.

(IO) Bill never returned to the drug store. Carol went to the drug

store again.

14.

15.

16.

(AS)
(AS)
(IS)
(IS)

(AS)
(A0)
(IS)
(Io)
(AS)
(A0)
(IS)

(IO)

51

The prince won't hunt tigers anymore. The king will hunt
tigers again.

The king won't hunt bears anymore. The king will hunt tiggrs
again.

The king won't hunt bears anymore. The king will hunt tigers
again.

The prince won't hunt tigers anymore. The king will hunt

tigers again.

Henry will never go back to Canada. Max will visit Canada
again.

Max will never go back to Mexico. Max will visit Canada.again.
Max will never go back to Mexico. Max will visit Canada again.
Henry will never go back to Canada. Max will visit Canada
again.

The barber won't date Ellen anymore. The mailman will date
Ellen again.
The mailman won't date Judy anymore. The mailman will date
Ellen again.
The mailman won‘t date Judy anymore. The mailman will date
Ellen again.
The barber won't date Ellen anymore. The mailman will date
Ellen again.

APPENDIX D

ATTENDIX D

Lists Presented in Experiment II

List 1.

Vivian can‘t speak German anymore. Vivian still speaks Russian.
Jane had a cold. Mggy had a cold too.

The floor isn't dirty anymore. The window is still dirty.

Max will never go back to Mexico. Max will visit Canada again.
Bob can't play the piano. .§2§ can't play the violin either.

The butcher joined the Credit Union. The tailor joined the
Credit Union too.

 

. The mailman won't date Judy anymore.

The lawyer doesn‘t like apples.
either.

Rachel didn't give Mark a card.
either.

Carl was drinking beer. Paul was drinking beer too.

The prince won't hunt tigers anymore. The king will hunt tigers
again.

Bill never returned to the drug store.
store again.

John doesn't know how to ice skate.
either.

Kathy doesn't like to crochet anymore.

The doctor doesn't like apples

Rachel didn't give Mark a gift

Carol went to the drug
John doesn't know how to ski

Kathy still likes to knit.
The mailman will date Ellen
again.

The Smiths went to Pittsburgh. The Smiths went to Cleveland too.
Pet doesn't have a collie anymore. App still has a collie.

List 2.

10.c.
1.b.
12.a.
15.0.
5.d.
2.a.

6.a.
8.d.

4.a.

Vivian can't speak German anymore. Vivian still speaks Russian.
Mbry had the measles. Mary had a cold too.

The floor isn't dirty anymore. The window is still dirty.

Max will never go back to Mexico. Mgg will visit Canada again.
Fred can't play the violin. Bob can't play the violin either.
The butcher joined the Credit Union. The tailor joined the
Credit Union too.

The lawyer doesn't like apples.
either.

Diane didn't give Mark a gift.
either.

Carl was drinking beer.

 

 

 

The doctor doesn't like apples
Rachel didn't give Mark a gift

Paul was drinking beer too.

52

1h.b.
13Ib.
7.d.

11.0.
16.0.

3.d.
9.b.

53

The king won't hunt bears anymore.
again.

Carol never returned to the shoe store.
store again.

Dave doesn't know how to ski.
either.

Kathy doesn't like to crochet anymore.
The mailman won't date Judy anymore.
again.

The Joneses went to Cleveland. The smiths went to Cleveland too.
Ann doesn't have a beagle anymore. Ann still has a collie.

The king will hunt tigers
Carol went to the drug
John doesn't know how to E5;

Kathy still likes to knit.
The mailman will date Ellen

List 3.

10.d.
1.0.
12.b.
15.d.
5.a.

2.b.
6.b.
8.a.

40b.
14.0.

13.0.
7.a.

110d.
16.d.

3.a.
9.0.

Barry can't speak Russian anymore. Vivian still speaks Russian.
Mary had the measles. Mary had a cold too.

The window isn't broken anymore. The window is still dirty.
Henry will never go back to Canada. Max will visit Canada again.
Fred can't play the violin. .329 can't play the violin either.
The tailor joined the Men's Club. The tailor joined the Credit
Union too.

The doctor doesn't like bananas.
either.

Diane didn't give Mark a gift.
either.

Paul was drinking whiskey.
The king won't hunt bears anymore.
again.

Carol never returned to the shoe store.
store again.

Dave doesn't know how to ski.
Marilyn doesn't like to knit anymore.
The barber won't date Ellen anymore.
again.

The Joneses went to Cleveland.
Ann doesn't have a beagle anymore.

The doctor doesn't like apples
Rachel didn't give Mark a gift

Paul was drinking beer too.
The king will hunt tigers

 

Carol went to the drug

John doesn't know how to ski either.
Kathy still likes to knit.
The mailman will date Ellen

The Smiths went to Clevelami too.
Ann still has a collie.

List 4.

10.a.
10d.
12.0.
15.a.
5.b.
2.0.

6.0.

8.b.

Barry can't speak Russian anymore. Vivian still speaks Russian.
Jane had a cold. Mary had a cold too.

The window isn't broken anymore. The window is still dirty.
Henry will never go back to Canada. ‘Mgg will visit Canada.again.
Bob can't play the piano. Bob can't play the violin either.

The tailor joined the Men's Club. The tailor joined the Credit
Union too.

The doctor doesn't like bananas.
either.

Rachel didn't give Mark a card.
either.

The doctor doesn't like apples

Rachel didn't give Mark a gift

4.0.
143d.

13.d.
7.b.

11.a.
16.a.

3.b.
9.d.

54

Paul was drinking whiskey. Paul was drinking beer too.

The prince won't hunt tigers anymore. The king will hunt tiggrs
again.

Bill never returned to the drug store. Carol went to the drug
store again.

John doesn't know how to ice skate. John doesn't know how to
'gki either.

Marilyn doesn't like to knit anymore. Kathy still likes to knit.
The barber won't date Ellen anymore. The mailman will date Ellen
again.

The Smiths went to Pittsburgh. The Smiths went to Cleveland too.
Pat doesn't have a collie anymore. Ann still has a collie.

APEND IX E

APTENDIX E

Re00gnition Test for Experiment II

Read each of the sentences below and decide whether or not it is

one that you just heard. In the space after each sentence, write "Yes"
if it is a sentence you heard and "No" if you did not hear the sentence.

\OmflmU'l-F'UNr-t

H
O

11.

Ann still has a collie.
Janice is moving again.
Ernie doesn't smoke cigars either.
The king will hunt tigers again.
Carla still works as a waitress.
The tailor joined the Credit Union.
The doctor doesn't like apples either.
Anita is wearing lipstick too.
Barbara is still a student.
Richard doesn't know how to swim either.
The Smiths went to Cleveland too.
Vivian still speaks Russian.

The Johnsons went on vacation again.
Paul was drinking beer too.
Julia visited the zoo again.
Rachel didn't give Mark a gift either.
Max will visit Canada again.
Jim paid the electric bill too.
The door isn't locked either.
John doesn't know how to ski either.
The tourist is still taking pictures.
The window is still dirty.
The mailman will date Ellen again.
Bob can't play the violin either.
The governor is still in office.
Mary had a cold too.
Nancy spoke to the gardener too.
Sandra doesn't like cats either.
Kathy still likes to knit.
The necklace was stolen too.
Arthur has the flu again.
Carol went to the drug store again.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

55

LIST OF REFERENCES

LIST OF REFERENCES

Bolinger, D.L. Contrastive accent and contrastive stress. Langpgge,
1961 v 22: 83’96-

Chafe, N.L. Meaning and the structure pf langpggg. Chicago: Univer-
sity of Chicago Press, 1970.

Chafe, N.L. Language and consciousness. Langpggg, 1974, 29, 111-133.

Chafe, N.L. Givenness, contrastiveness, definiteness, subjects, topics,
and point of view. In C.N. Li (Ed.), Subject and topic, New York:
Academic Press, 1976.

 

Chomsky, N. Deep structure, surface structure, and semantic interpretar
tion. In D. D. Steinberg and L. A. Jakobovits (Eds. ), Semantics- An
interdisciplinary reader in philosophy, lingpistics, and psychology.
London: Cambridge University Press, 1971

Clark, H.H. Comprehension and the given-new contract. Paper presented
at conference of "The role of grammar in interdisciplinary linguis-
tic research", University of Bielefeld, Bielefeld, Germany} Decemr
ber 1973.

Cutler, A. On saying what you mean without meaning what you say.
Papers from the Tenth Regional Meeting of the Chicago Linguistic
Society, Chicago: Chicago Linguistic Society, 1974.

Cutler, A. Sentence stress and sentence comprehension. Unpublished
PhaD. dissertation, University of Texas, 1975).

Cutler, A. Beyond parsing and lexical lood-up: An enriched description
of auditory sentence comprehension. In R. J. Wales and E. Walker
(Eds. ), New approaches to language mechanisms: A collection of
psycholinguistic studies. Amsterdam. North-Holland Publishing
Company , 1976.

 

Cutler, A. and Fodor, J.A. Semantic focus and sentence comprehension.

COggition, 1979. Z. 49-59-

Cutler, A. and Foss, D.J. On the role of sentence stress in sentence
processing. Language and Speech, 1977..§Q. 1-10.

Halliday, M.A.K. Notes on transitivity and theme in English (Part 2).
Journal pf Lingpistics, 1967, 3, 177-244.

56

57

Haviland, S.E. and Clark, H.H. What's new? Acquiring new information
as a process in comprehension. Journal 2: Verbal Learning and
Verbal Behavior, 1974, 13, 512-521.

 

Holyoak, K.H. and Clark, H.H. The given-new strategy and the compre-
hension of relative clauses. Unpublished manuscript.

Hornby, P.A. The role of topic-comment in the recall of cleft and
pseudo-cleft sentences. Papers from the Seventh Regional Meeting
2: the Chicagg Linguistic Society. Chicago: Chicago Linguistic
Society, 1971.

Hornby, P.A. The psychological subject and predicate. Cognitive
Psych010gy, 1972, 3, 632-642.

Hornby, P.A. Surface structure and presupposition. Journal 2: Verbal
Learning End Verbal Behavior, 1974:.12! 530-538.

Jackendoff, R.S. Semantic interpretation in generative grammar. Cam-
bridge, Mass.: The M.I.T. Press, 1972.

Jackson, M.D. and McClelland, J.L. Processing determinants of reading
speed. Journal 9f Experimental PsychOIOgy: General, 1979..19§:
151-181.

Jesperson, O. ‘A modern English grammar 33 historical principles, part
VII, syntax. London: Allen and Unwin, 1961.

Lehiste, I. Suprasegmentals. Cambridge, Mass.: The M.I.T. Press, 1970.

Newman, J.E. and Dell, 6.8. The phonological nature of phoneme monitor-
ing: A critique of some ambiguity studies. Journal of Verbal

Perfetti, C.A. and Lesgold, A.M. Discourse comprehension and sources of
individual differences. In M.A. Just and P.A. Carpenter (Eds.),
Cognitive processes in comprehension. Hillsdale, N.J.: Lawrence
Erlbaum Associates, 1977. '

Shields, J.L., McHugh, A. and Martin, J.G. Reaction time to phoneme
targets as a function of rhythmic cues in continuous speech. Jour-
nal 3; Experimental PsycholOgy, 1974, 102, 250—255.

MICHIGAN ST

ll Will”

3129!

17mm