• TABLE OF CONTENTS
HIDE
 Title Page
 Acknowledgement
 Table of Contents
 List of Tables
 List of Figures
 Abstract
 Introduction
 Review of the literature
 Collection and treatment of...
 Analysis of the data
 Summary conclusions and implic...
 Appendices
 Reference
 Biography






Group Title: degree of predicatability of four types of sentences as measured by the eye-voice span of graduate student readers / by Diana Lynn Lagotic
Title: The Degree of predicatability of four types of sentences as measured by the eye-voice span of graduate student readers / by Diana Lynn Lagotic
CITATION PDF VIEWER THUMBNAILS PAGE IMAGE ZOOMABLE
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00098100/00001
 Material Information
Title: The Degree of predicatability of four types of sentences as measured by the eye-voice span of graduate student readers / by Diana Lynn Lagotic
Physical Description: x, 103 leaves : ill. ; 28 cm.
Language: English
Creator: Lagotic, Diana Lynn, 1945-
Publication Date: 1977
Copyright Date: 1977
 Subjects
Subject: Reading (Higher education)   ( lcsh )
English language -- Study and teaching (Higher)   ( lcsh )
Curriculum and Instruction thesis Ph. D
Dissertations, Academic -- Curriculum and Instruction -- UF
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
 Notes
Thesis: Thesis--University of Florida.
Bibliography: Bibliography: leaves 98-102.
General Note: Typescript.
General Note: Vita.
 Record Information
Bibliographic ID: UF00098100
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: alephbibnum - 000197910
oclc - 03759507
notis - AAW4600

Downloads

This item has the following downloads:

PDF ( 5 MBs ) ( PDF )


Table of Contents
    Title Page
        Page i
        Page ii
        Page iii
    Acknowledgement
        Page iv
    Table of Contents
        Page v
        Page vi
    List of Tables
        Page vii
    List of Figures
        Page viii
    Abstract
        Page ix
        Page x
    Introduction
        Page 1
        Page 2
        Page 3
        Page 4
        Page 5
        Page 6
        Page 7
        Page 8
        Page 9
        Page 10
        Page 11
        Page 12
    Review of the literature
        Page 13
        Page 14
        Page 15
        Page 16
        Page 17
        Page 18
        Page 19
        Page 20
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
        Page 27
        Page 28
    Collection and treatment of data
        Page 29
        Page 30
        Page 31
        Page 32
        Page 33
        Page 34
        Page 35
        Page 36
        Page 37
        Page 38
    Analysis of the data
        Page 39
        Page 40
        Page 41
        Page 42
        Page 43
        Page 44
        Page 45
        Page 46
        Page 47
        Page 48
        Page 49
        Page 50
        Page 51
        Page 52
        Page 53
        Page 54
        Page 55
        Page 56
        Page 57
        Page 58
        Page 59
        Page 60
    Summary conclusions and implications
        Page 61
        Page 62
        Page 63
        Page 64
        Page 65
        Page 66
        Page 67
        Page 68
        Page 69
        Page 70
        Page 71
        Page 72
        Page 73
        Page 74
        Page 75
        Page 76
    Appendices
        Page 77
        Page 78
        Page 79
        Page 80
        Page 81
        Page 82
        Page 83
        Page 84
        Page 85
        Page 86
        Page 87
        Page 88
        Page 89
        Page 90
        Page 91
        Page 92
        Page 93
        Page 94
        Page 95
        Page 96
        Page 97
    Reference
        Page 98
        Page 99
        Page 100
        Page 101
        Page 102
    Biography
        Page 103
        Page 104
        Page 105
        Page 106
        Page 107
Full Text














THE DEGREE OF PREDICTABILITY
OF FOUR TYPES OF SENTENCES AS MEASURED BY THE
EYE-VOICE SPAN OF GRADUATE STUDENT READERS















By

DIANA LYNN IAGOTIC


A DISSERTATION PRESENTED TO TIE (RADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREFIIIETS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY















UNIVERSITY OF FLORIDA


1977








































Copyright 1977

by

Diana Lynn Laqotic








































To my husband Frank and daughter Kristic
















ACKNOWLEDGMENTS


The author is indebted to the many individuals who have assisted her

in the completion of this study. She expresses her gratitude to the

members of her supervisory committee, Dr. Henry T. Fillmer, Chairman;

Dr. Janet J. Larsen; Dr. William B. Ware; Dr. Kevin M. McCarthy; and

Dr. Robert G. Wright. Special appreciation is extended to Drs. Fillmer

and Larsen-for their encouragement and interest during the preparation

of this study.

Also deserving of recognition is Dr. Ruthellen Crews for her

contribution during the formative stages of this project.

The author thanks Mrs. Rita Crawford and Mrs. Mary Marshall for

the typing and editorial assistance they provided.

The author expresses particular appreciation to her husband, Frank,

who originally suggested the pursuance of this degree. His love, support,

and encouragement made this study possible. Finally, special thanks is

given to her daughter, Kristie, for the love and happiness she has given

her throughout this year.














TABLE OF CONTENTS


ACKNOWLEDGEMENTS

LIST OF TABLES

LIST OF FIGURES

ABSTRACT . .

CHAPTER


I INTRODUCTION ....

Statement of the Problem .

Justification for the Study

Definition of Terms .

Plan for the Study .. ..

The Sample

The Materials .....

Data Collection ..

Treatment of the Data . .

Hypothesis . . . . .

Limitations . . . .

II REVIEW OF THE LITERATURE ..

The Deep Structure Model .

The Chunk odl ......

Language Development and Rea

Levin and Kaplan's Model .


Page

iv

vii

viii

ix


dinq


I










CHAPTER


II


IVS The Measurement Tool


EVS and the Unit of Decoding . .


Related Research . . .


Summary . . . . . . .


III COLLECTION AND TREATMENT OF DATA . .


The Sample . . .


The Instrument . . .


Collection of Data . . .


Treatment of Data . . . . .


IV ANALYSIS OF THE DATA . . . ...


Findings Related to the Assumptions


Findings Related to the Hypothesis


Findings Related to the Questions .


V SUMMARY CONCLUSIONS AND IMPLICATIONS .


Summary . ........


Conclusions . . . . . .


Related to the Hypothesis .


Related to the Questions . .


Recommendations . . .


Page


. . 22


. . 25


. . 26


. . 27


. . 29


S 29


. . 30


. . 34


. . 35


. . 39


. . 40


. . 40


. . 44


. . 61


- -. 61


. . 62


. . 62


. . 63


. . 71


APPENDICES


A SENTENCE TYPES . . . . . . . . . . . .


B PARAGRAPHS CONTAINING TARGET SENTENCES AND UNRELATEDD
SENTENCES . . . . . .


C SYMMETRY OF THE VARIANCE-COVARIAINCT MATRIX .. . ....


REFERENCES . . . . . .


BIOGRAPHY . . . . . . . . . . . . . . .

















LIST OF TABLES


TABLE


1 Cell Means for the Randomized Block Factorial
Design . . . . . .

2 Summary Table for the Randomized Block Factorial
Design . . . . . . .

3 Analysis of the Interaction of Critical Point by Type
of Transformation . . . . . . . . . .

4 Difference of the T's . . . . . . .... .

5 Analysis of the Interaction of Basic Versus Transformation
by Critical Point . . . . . . . ... . .

6 Difference of the C's . . . . . . . . . .

7 Analysis of the Tnteractions of Type of Transformation
by Basic Versus Transformation ......

















LIST OF FIGURES


FIGURE Page

1 Schematic of the Randomized Block Factorial Design . .. 10

2 Mean EVS by Critical Point for the Basic and the THERE
Transformed Sentences .... .......... 51

3 Mean EVS by Critical Point for Active and Passive
Sentences ............ . . . ................ 52

4 Mean EVS by Critical Point for the Basic and the
WH-puestion Transformed Sentences . . . . .. 53

5 Mean EVS by Critical Point for the Basic and the
IT-Inversion Transformed Sentences . . . . ... 55

6 Mean EVS by Critical Point for Passive and the THERE
Transformed Sentences ..... . . . . . 58









Abstract of Dissertation Presented to the
Graduate School of the University of Florida in Partial
Fulfillment of the Requirements for the Degree of
Doctor of Philosophy

THE DEGREE OF PREDICTABILITY OF' FOUR TYPES OF
SENTENCES AS MEASURED BY THE EYE-VOICE SPAN OF GRADUATE STUDENT READERS

by

Diana Lynn Laqotic

August, 1977

Chairman: Dr. Henry T. Fillmer
Major Department: Curriculum and Instruction

Psycholinguists have offered at least three different language pro-

cessing models to explain the strategies involved in the reading process.

Two models, the deep structure model and the chunk model, offer

diametrically opposite explanations. A third model, offered by Harry

Levin and Eleanor Kaplan, suggests that readers use their knowledge of

sentence predictability in attempting to decode and comprehend sentences.

This study focused on the degree of predictability of specific types

of sentence transformations. Answers to the following questions were

sought:

1. Is there a difference in the degree of predictability as measured

by the eye-voice span between

a) a basic sentence and its THERE transformation

b) an active sentence and its passive transformation

c) a basic sentence and its WH-question transformation

d) a basic sentence and its IT-inversion transformation with

an infinitive subject nominalization?

2. Does the number of transformations within a sentence affect the

degree of predictability?








3. Does the type of sentence transformation affect the degree of

predictability?

4. Does the position in the sentence where the eye-voice span is

measured affect the degree of predictability?

Sentences containing the four types of transformations tested were

developed by the researcher. The eye-voice span was measured at pre-

determined places in each of the sentences. The study sample consisted

of 20 University of Florida Reading and English graduate students.

The data were analyzed using a randomized block factorial design

with repeated measures. Results showed that there was a significant

three-way interaction: type of transformation by basic versus transfor-

mation by critical point.

Further analysis indicated that there was a complex interrelationship

among the three independent variables --type of transformation, basic

versus transformation and critical point -on the eye-voice span

measurement.

There was no evidence from the data to support any of the language

processing models. The findings indicated that since no one variable

by itself can explain the differences in the eye-voice span measurements,

the explanations of language processing offered by the major models are

too basic.
















CHAPTER I

INTRODUCTION


Psycholinguists have attempted recently to explain the strategies

involved in the reading process through the use of language processing

models. Two major models, the deep structure mode] and the chunk model,

offer diametrically opposite explanations of the operations that occur

during language processing. A third model has been offered by Levin and

Kaplan (Levin, Grossman, Kaplan, & Yang, 1972, p. 30). Sentence pre-

dictability is central to the model. These researchers have been

concerned with determining the extent to which readers use their knowledge

of the predictability within sentences in attempting to decode and compre-

hend the sentence. Studies have shown that eye-voice span can be used to

investigate such questions.

This study focused on the degree of predictability of specific

sentence types. Since eye-voice span was used as the tool for measurement,

information about the amount of material which is picked up and processed

by the reader in each sentence was examined.


Statement of the Problem

The purpose of this study was to determine which of four different

sentence structures have greater degrees of predictability within sentences.

Answers to the following questions were sought:

1. Is there a difference in the degree of predictability as

measured by the eye-voice span between









a) a basic sentence and its THERE transformation

b) an active sentence and its passive transformation

c) a basic sentence and its WH-question Iransformation

d) a basic sentence and its IT-inverstion transformation with

an infinitive subject nominalization?

2. Does the number of transformations within a sentence affect the

degree of predictability?

3. Does the type of sentence transformation affect the degree of

predi cLability?

4. Does the position in the sentence where the eye-voice span is

measured affect the degree of predictability?


Justification for the Study

Recent studies examining language processing have been conducted to

determine whether a reader uses the knowledge of language regularities as

an aid in processing a sentence. Levin, et al. (1972) and Clark (1965)

have shown that some parts of sentences are more predictable than others.

According to Levin, et al. (1972), "sequences of words in language are

constrained by rules" (p. 30). The rules which set the constraints

are called grammar. The reader creates expectations about the language

based on the detection of the grammatical constraints in the language.

"The exact pattern or distribution of constraints appears to depend on

the grammatical structure of the sentence" (Levin, et al., 1972, p. 31).

Levin and Kaplan (196H) have suggested that the reading process may

involve "some internal decision-making mechanism, which utilizes its

knowledge of linquistic structure to determirin how the message shall be

processed" (p. 258). Cromer (1970), after reviewing the results of his

study, has proposed that one source of comprehension difficulty might be

"attributed to the way some poor readers organize reading input" (p. 471).









This study investigated further the processing of language which

takes place during reading. Specific sentence structures were used to

determine the nature of the regularities and to determine whether there

was any relationship between the degree of predictability and the type

and number of transformations in each sentence.

The eye-voice span has been shown to be readily affected by the

difficulty of the material (Anderson, 1937; Fairbanks, 1937; Levin & Cohn,

1968; Stone, 1941). With the control for word number and vocabulary,

differences in eye-voice span would indicate differences in the difficulty

of the material. Therefore, the number of words in each sentence and the

vocabulary in each sentence used in this study were controlled. A longer

eye-voice span measurement would indicate greater ease in decoding the

sentence and greater predictability.


Definition of Terms

1. critical point A pre-determined place in each experimental sentence

where the eye-voice span (EVS) was systematically measured. When a

subject reached a critical point in the sentence during his oral

reading, the screen was blackened and an EVS measurement was taken.

2. intra-sentence constraint The pattern of contingencies between

sentence parts. The predictability of sentence parts.

3. eye-voice span In reading aloud, the FTS is the number of correct

consecutive words reported by the subjectt beyond the critical point

in each sentence. "The distance, usually measured in words, that the

eyes are ahead of the voice" (Gibson i Levin, 1975, p. 360).

4. deep structure The ultimate underlying form of a sentence.

5. surface structure The way we see and hear language. The acoustic

or graphic level of an utterance.








6. transformational grammar The set of rules that lead from the deep

structure to the surface structure.

7. basic sentence A sentence which is made up of a noun phrase and a

verb phrase. A basic sentence does not contain transformations.

8. WH-question transformation A question beginning with a WH-word.

Examples of WH-words are who, what, where, when, and why. The

following is an example of a WH-question transformation:

When will firemen have the equipment for their orientation?

The sentence upon which this transformation is based is

Firemen will have the equipment for their orientation tomorrow.

9. THERE-transformation "Usually a string of words will not undergo

the THERE transformation unless it contains a form of BE. The

transformation moves the BE and the auxiliary element or elements

preceding it to the beginning of the string, then adds 'there' before

the auxiliary elements) and the BE form" (Malmstrom & Weaver, 1973,

p. 148). The following is an example of a "there" transformation:

There were men on the docks approaching the machinery.

The sentence upon which this transformation is based is

Men were on the docks approaching the machinery.

10. infinitive subject nominalization Consists of "To" plus the base

from a verb which is used in the subject position in a sentence. The

following is an example of an infinitive subject nominalization:

To install alarms throughout the house is too expensive.

11. IT-inversion A noun phrase used as the subject is moved to the end

of a sentence and the element "it" is added in the subject position.

The following is an example of an TT-inversion transformation:

It is too expensive to install alarms throughout the house.









Plan for the Study

The Sample

The study sample consisted of 20 University of Florida Reading and

English graduate students. Two requirements in the selection of subjects

were that they could read vocabulary limited to Mitzel's Functional

Reading Word List for Adults, and that they could read sentences involving

embedded transformations and transformations which expand the basic

sentence. Poor readers might not have been capable of handling either

the vocabulary or the sentences designed for the study; therefore, the

selection of subjects was limited to able readers. Any subject who had

difficulty reading the sample sentences orally was eliminated from the

study. Eliminating subjects who had difficulty with the sentence tasks

did not affect the study, since the study focused on the way able readers

attacked each of the forty sentences.

Limiting the selection of subjects to Reading and English students

was necessary to assure that there were minimal differences in the way

subjects approached the reading task.


The Materials

Sentences were developed for use in the study. The vocabulary was

limited to words from the Functional Reading Word List for Adults

(Mitzel, 1966).

The sentence types developed hy the researcher were limited to the

following:

1. A basic sentence and its THERE transformation.

2. An active sentence and its passive transformation.

3. A basic sentence and its WH-word transformation.

4. A basic sentence and its IT-inversion transformation with an

infinitive subject nominalization.








The sentence types were chosen for the following reasons:

1. Except for the active and passive sentences, all of the

sentences involve a process in which the subject is moved

from the beginning of the sentence to a position following

the verb or modal.

2. The passive transformation and the THERE transformation

involve one transformation each. The WH-question

transformation involves two transformations, and the

IT-inversion transformation with an infinitive subject

nominalization involves three transformations.

3. The active and passive sentence types were used by Levin

and Kaplan (1968) in their study. A comparison will be

made between the results found in this study using new

active and passive sentences and the results of the Tevin

and Kaplan study.

The eye-voice span was measured at predetermined places (critical

points) in each of the sentences. In both the basic sentences and the

transformed sentences, the first, second, and third critical points

followed the first, second, and third words respectively in each sentence.

The fourth and fifth critical points followed three-word phrases. The

following are examples of the positions of the critical points in the

basic and transformed sentences:

12 3 4
Basic Sentence: Exercise/is/difficult/in the beginning/for many
5
individuals/with major health problems from poor eating habits.

1 2 3 4
THERE Transformation: There/were/witnseses/near the accident/from

5
the collision/of the two trailer trucks on the nearby bridge.








1 2 3 4
Passive Transformation: The/helpless/animal/was being cornered/by the

5
dogs/in an open field near the winter cabin.

1 2 3 4
WH-question Transformation: When/will/newsmen/know the truth/about

5
the proceedings/against the five men from the new company.

1 2 3 4
IT-inversion Transformation: It/is/profitable/to outline chapters/

5
in a book/to organize your study of unfamiliar subject matter.


The second critical point followed a verb or modal in all of the sentences

except the active and passive sentences. The second critical point

followed an adjective in the active and passive sentences.

One sentence was used for each critical position within each sentence

making a total of 40 sentences (see Appendix A). At least nine words

follow the last critical point in each sentence. Exploratory data by

Levin and Kaplan (1968) and studies by McConkie and Rayner (1975), Rayner

(1975a), and Rayner (1975b) using undergraduate students indicated that

eight words past the critical point was sufficient to prevent the possi-

bility of subjects extending their EVS to the last word in the sentence.

Each of the target sentences was placed in paragraphs of three

unrelated sentences (see Appendix B). The sentences were taken from

General Reading for Understanding Cards (Thurston, 1969). Sentences were

taken from paragraphs 21-60, which cover a range of difficulty from

seventh to ninth grade.

Since exploratory data by Levin and Kaplan (1968, p. 253) indicated

that subjects scanned the first line before beginning to read aloud, the

target sentence was never first in the paragraph. The target sentence

was distributed among sentence positions two, three, or four in the








paragraph. The critical point occurred in either the first, second,

third,or fourth word of a line. Among the sentence types, therefore, the

position of the target sentence in the paragraph and the position of the

critical point were varied. There were at least nine words past the

critical point of each target sentence on the same line to control for

position effect.


Data Collection

The paragraphs were projected onto a Caramatt TT screen. The subject

sat directly in front of the machine. The following instructions were

read to each subject: "A paragraph will be projected onto the screen

directly in front of you. The paragraph will consist of four unrelated

sentences. When a slide appears on the screen, you will begin reading the

paragraph orally. At some predetermined point in your oral reading, I will

push a remote control button which will remove the paragraph and leave a

blank screen. When this occurs, you should report as much of the rest of

the material as you remember seeing beyond the last word you said. The

number of correct consecutive words reported will be taken as your EVS

measurement for that paragraph. The EVS is defined as the number of words

that your eyes are ahead of your voice. Two practice paragraphs will be

presented. The total study will consist of 40 paragraphs and, therefore,

40 EVS measures. A tape recorder will be used to record exact responses.

There are no right or wrong answers. We will start with the first practice

paragraph."

The examiner sat behind the subject. This prevented the subject from

seeing the remote control button being pushed. The order of paragraph

presentations was randomized independently for each subject. This was

done using twenty lists of forty random numbers and changing the order of


I









slide presentations for each subject. A tape recorder was used to record

exact responses.


Treatment of Data

The relative merits of the major language processing models were

examined by determining if there were significant differences in the

intra-sentence constraints or predictability of the sentence types used

in this study. A longer EVS measurement indicates greater predictability

or ease in decoding the sentence. If the transformed sentences had longer

EVS measurements than the basic sentences, the chunking model would be

supported because the chunking model states that transformed sentences are

easier to process. If the EVS measurements for transformed sentences were

shorter than the basic sentences, the deep structure model would be

supported. This model states that a basic sentence is easier to process

than a transformed sentence.

The study involved an analysis of the effects of the three independent

variables, types of transformation, basic versus transformation, and

critical point on the dependent variable, eye-voice span. The independent

variable, types of transformation, involved the following four levels:

THERE, passive, WH-question, and IT-inversion. The second independent

variable, basic versus transformation, involved two levels, the basic

sentences and the transformed sentences. The third independent variable,

critical point, involved five levels, the five positions in the sentence

where the EVS was systematically measured. The schematic given in

Figure 1 describes the design.

The data were aniayzed using a randomized block factorial design.

This design provides an opportunity to examine each independent variable

separately to determine its effect on the EVS. The interactions of the






10





0
*H )n
4-J




H 0




H
0 0
UU
U)







E C

0




E4
o N








Erl -t
0 H









in




u V
co m


41


u 0

nj N







E 1
im 0






U) N


H 0
0 U -)
U) H Q [








independent variables can also be examined to determine the effects of

the interactions on the EVS.


Hypothesis

The hypothesis tested was that there was no significant difference

in the degree of predictability among the sentence types used in this

study. The model for the design was the following:

EVSijkl = + T + j + Bk + Yl + Bjk + y + Pyk + "Ajkl + Eijk


where T = effect of subjects

S= effect of types of transformations

8 = effect of basic versus transformed sentences

Y = effect of critical points


The following hypotheses were tested at the .05 level of

significance:

H : a. = o for all j
o 1
H1: a > o for some j


H : Bk = o for all k

H Pk f c for some k
1 k


H : = o for all 1

II : y o for some 1


H : Bk =1 o for all jk

HI: (0jk / o for some jk


He: .Y o for all jk
o ik




HI: sY i i o for some j1


S: = o for all k1
o kl
H1: Yk / o for some kl


H : jnPk for all jkl
o jkl
HI: sRy jk] o for some jkl









1. There are no differences in the means of the EVS among the types of

transformations.

2. There are no differences in the means of the EVS among the basic and

the transformed sentences.

3. There are no differences in the means of the EVS among the critical

points.

4. There is no interaction between basic versus transformed sentences

and types of transformation with respect to the EVS means.

5. There is no interaction between basic versus transformed sentences and

the critical point with respect to the EVS means.

6. There is no interaction between types of transformation and the

critical point with respect to the EVS means.

7. There is no interaction among types of transformation, the critical

point, and the basic versus transformed sentences with respect to the

EVS means.


Limitations

These limitations outline the bcunaries of the study:

1. The sentence types used in this study were confined to the

following:

a) A basic sentence and its THERE transformation.

b) An active sentence and its passive transformation.

c) A basic sentence and its WH-question transformation.

d) A basic sentence and its IT-inversion transformation with

an infinitive nomina lization.

2. The sample population was restricted to graduate students.

Results from this study can be applied only to able readers.

















CHAPTER II

REVIEW OF THE LITERATURE


Before examining the degree of predictability within sentences in an

effort to understand more about the strategies involved in the reading

process, a preliminary understanding of the different language processing

models offered by psycholinguists is necessary.

Two major models which have been developed are diametrically opposite

in their theoretical positions. One model, called the deep structure

model, relates reading difficulty to the number and the complexity of the

transformations in sentences. The more transformations in a sentence and

the more complex the transformations, the more difficult the reading. The

second model, referred to as the chunk model, claims that as a sentence

becomes more compl ex, it becomes easier to read. A third model, offered

by Levin and Kaplan (1972) centers on the use of eye-voice span to deter-

mine the extent to which readers use their knowledge of the predictability

within sentences in attempting to decode and comprehend the sentence.

The theoretical positions of each of the models are reviewed in thil

chapter. In addition, the research which has been done in an effort I"'

support the mod-ls is presented.


The Deep Structure Model

The processing of verbal data according to the deep structure

position involves a one-to-one correspondence between a grammatical model

and a psychological model. The deep structure model attempts to show









a relationship between operations used in transformational-generative

grammar and operations used to do verbal processing in the mind. The

major goal is to study "the psychological processes which contribute to

the acquisition, production and comprehension of language" (Saporta,

1967, p. 7). According to Pearson (1974-75), the deep structure model

was proposed by Mehler (1963) and Miller and McKean (1964). The model

has been described by Pearson in the following way: "As the surface

structure form (the way we sec and hear language) approaches deep

structure form (the state in which we consciously or unconsciously

process and understand language in the mind), comprehension is

facilitated. This facilitation occurs because the listener or reader

must undergo fewer operations (transformations) in order to analyze or

break down the surface structure form into deep structure" (p. 158).

Since 1960 numerous studies have been conducted in an attempt to

show a one-to-one correspondence between the psychological model of the

reader and the operations of transformational-generative grammar. Evans

(1972-1973) found that problem readers could "raise their comprehension

by reading transformationally simplified prose" (p. 271). Results from

some studies have shown that sentences become more complex as more trans-

formations are added (Gough, 1966) and as different types of transforma-

tions are added (Fodor, 1967). Fagan (1971a) did not find the same

results. Sentence difficulty, according to his study, was more dependent

on the difficulty of the transformation rather than on the number of

transformations within a sentence. It has been found that the active

verb sentences are more comprrhonsible and easier to process than passives,

nominalizations, adjectivalizations, and negatives (Coleman and Blumenfeld,

1963; Coleman, 1964; Miller, l'62), and non-embedded sentences are easier

to learn than embedded sentences (Coleman, 1965). "The transformations








which make subject-plus-predicate strings into adjectivalizations,

adverbializations, and nominalizations are lnbled embedding transformations"

(Malmstrom & Weaver, 1973, p. 173). Appositives, ing nominalizations,and

common element deletions were found to be among the more difficult trans-

formations for children to read (Fagan, 1971b).

Additional results have shown that "verbal noun phrases, noun and

relative clauses, subordinate clauses, as well as the number of modifiers

embedded in any noun headword, all complicate reading comprehension"

(Evans, 1972-73, p. 274). Blumenfeld and Miller (1965) concluded that

"certain complex grammatical constraints and relationships which do seem

to affect reading ability exist in the language. These larger relation-

ships and processes can be described and taught in terms of transforma-

tional grammar, and the relationships and processes involved in them do

result in a significant improvement in reading ability" (p. 755). A

previous study by Coleman and Blumenfeld (1963) and a study by Little

(1975) supported these conclusions. Results by O'Donnell (1963) do not

show conclusive evidence for the teaching of linguistic structure as a

means of developing reading comprehension. However, O'Donnell cautioned

that the ability to recognize structural relationships could be important

in reading comprehension. He concluded that there are too many distinct

factors involved in the reading process to make definite conclusions

about the relationship between the knowledge of linguistic structure and

reading comprehension.

Coleman (1962), using cloze tests, found that a technical passage

divided into short sentences was significantly more comprehensible than

the same information stated in longer sentences. He found that shortening

clauses improved comprehinsion more effertivply than shortening sentences.

Miller and SelfridqP (1950) conducted a study which indicated that readers









found context clues more useful in the reading process as the passage

increased in length.

There is evidence that relational words (prepositions, adverbs,

conjunctions, and pronouns) are more dependent on the whole sentence

structure than are reference words (nouns, adjectives, and verbs). The

relational words appear to lose their function and identity as the

sentence is fragmented (Treisman, 1965).

Stoodt (1972) and Robertson (1968) reported a significant correlation

between a subject's ability to identify the relationship that conjunctions

signal and his reading comprehension. The easiest conjunctions were and,

how, for,and as. The most difficult were when, so, but, or, where, while,

now, that, and if.

There appears to be a relationship between a student's reading

comprehension and his ability to recover deep structure (Reynolds, 1974;

Simons, 1970). Fagan (1971a) defines deep structure as that which "allows

the learner to understand the meaning of the sentence," and surface

structure as "that form of the language to which the learner is exposed"

(p. 169). Simons (1970) using a cloze test found that recovering deep

structure was a more important aspect of reading comprehension skill than

I.Q., word knowledge, and word recognition skill. Simons suggested that

more studies were needed to discover the strategies readers use in

recovering deep structure. Meh]er and Carey (1967) found that changes

in surface structure and base structure can significantly disrupt

perception. Savin and Perchonock (1965) concluded that "forms involving

more transformations or mnre complex transformations interfered with

memory because they required additional psychological processing in order

to get a deep structure reptresnntation" (p. 348).








Miller and McKean (1964) found that the processing of active affirma-

tive sentences took consistently less time than passives and negatives.

Passive sentences required more time to process than negatives. When a

sentence contained both a passive and a negative transformation, the

sentence took about as long to process as did the sum of time required

to process a passive sentence and a negative sentence separately. Miller

and McKean suggested that the evidence supported the idea that subjects

processed separately each transformation in a sentence. Mehler (1963)

and Gough (1965) also found support For this view. Mehler found evidence

that subjects "do not recall the sentence verbatim, but rather that they

analyze it syntactically and encode it as a kernel sentence plus

appropriate transformations" (p. 350). Gouqh's results (1965) indicate

that syntax is related to speed of understanding. The "hearer of a

complex sentence transforms that sentence into its underlying kernel

and that understanding of the sentence waits upon such transformation"

(p. 110).

Epstein (1967) and Tulvinq and Potkau (1962) reported that grammar

apparently influenced memory because syntax facilitated learning. Siler

(1973-1974) concluded that syntax and semantics are apparently interre-

lated; however, syntax appeared to have a greater effect than semantics

on oral reading performance. Wang (1970) found that the presence of

either syntactic or semantic structure in a string of words facilitated

learning. There was a greater loss in comprehensibility, however, with

the removal of semantic structure. More than a 50% loss of comprehensi-

bility was involved when the semantic structure was removed. When the

syntactic structure was remnved, there was only a 35% loss in

comprehensible ty.








The Chunk Model

The chunk model is offered as an alternative view of verbal

processing by some psycholinguists. This model claims that as a sentence

becomes more complex it becomes easier to read. If the surface form of

a statement is already highly synthesized, comprehension is facilitated.

If, on the other hand, the surface structure is broken down somewhat

(is closer to its deep structure, the ultimate underlying form of a

sentence), comprehension is impeded (Pearson, 1974-75, p. 156). To get

from the deep structure to the surface structure of a sentence, certain

transformations on the deep structure must be performed. The number of

transformations necessary to get from the deep structure to the surface

structure could be considered an index of the complexity of the surface

structure (Pearson, 1974-75, p. 166). Advocates of the chunk model

contend that a complex surface structure is easier to read than the

underlying deep structure form of the sentence. This position is

diametrically opposed to that of the deep structure model.

A chunk as defined by Carver (1972-73) is "a group of words, usually

longer than a word and usually shorter than a sentence length. Each

chunk forms a meaningful and practical unit of connected discourse"

(p. 382). According to Bransford and Franks (1971), "the hearer or

reader must go through some sort of synthesizing process to cement" the

units "together or else he fails to do so and never comprehends the

relations" (p. 331).

Research on the chunk model and its relationship to reading has been

limited. In a 1970 study by Cromer, two qrouos or poor readers were

compared on comprehension tasks. Both the "difference" group, which

appeared to read word-by-woid, and the "deficit" group, which appeared

to have relatively inadequate vocabulary skills as measured by scores on








the Cooperative English Test of Readina Comprehension, were given stories

presented in four different modes. The modes consisted of regular

sentences, single words, meaningful phrases, and fragmented word groups.

Under the phrase condition, words were presented in groups based on

Lefevre's (1964) criteria (cited in Cromer 1970). Most sentences were

separated into two or more phrases. A comprehension test consisting of

twenty questions was given to each student at the end of each mode

presentation. When the material was presented in reorganized phrases,

the difference group but not the deficit group was able to comprehend as

well as good readers. When material was organized or grouped in a mean-

ingful way, the comprehension scores for the difference group improved.

These results appeared to indicate that the difference group did not

organize the reading input in a way that best facilitated good compre-

hension. Cromer interpreted the results in the following way: "One

source of comprehension difficulty can be attributed to a difference in

the way some poor readers organize reading input" (p. 471). Carver (1968),

using mature readers as subjects, examined the efficiency of chunking

reading material and found that the "spatial separation of reading

material into meaningful related groups did not improve the reading

efficiency of the mature readers" (p. 3). A study by Carver (1970b)

supported the same conclusion.

Pearson (1974-75) conducted three experiments which were designed to

determine linguistic variables which might affect the way third and

fourth grade children comprehend verbal data when they read. According

to Pearson, most of the data can be explained by the chunk model. In one

experiment, Pearson examined the preferences of children among several

syntactically different ways of communicating an idea. In a second

experiment Pearson studied the effects of syntactic complexity on the









comprehension of causal relations. The third experiment examined the

difference between the syntactic form in which statements were read and

the form in which the same statements were later recalled. The findings

suggested that children were able to handle complex forms and preferred

working with those forms. The more subordinated and longer sentences

elicited better comprehension.

Additional evidence in favor of the chunking model was found by

Bransford and Franks (1971). 3dult subjects were presented with larger

and smaller components of sentences. The largest components were con-

structed to represent the relations among four simple sentences. The

smallest components consisted of the four simple sentences upon which the

large components were constructed. The following are examples of the

sentences used in the study:

ones: The jelly was sweet.

The ants ate the jelly.

twos: The ants in the kitchen ate the jelly.

threes: The ants ate the sweet jelly which was on the table.

fours: The ants in the kitchen ate the sweet jelly which

was on the table.

Larger components were rated higher on recognition scores and on confi-

dence ratings. The confidence ratings were based on the statements by

the subjects indicating whether or not they had heard certain components.

Carver and Darby (1972-73) found evidence that the "chunk" test item

shows "promise as an indicator of information stored during reading"

(p. 285). He compared the IChunk Reading Test to the Davis Reading Test,

The Nelson Denny Reading Test, and the Tinker Speed of Reading Test. The

Chunk Reading Test was found to be a valid indicator of individual

differences on the following: (I) efficiency of comprehension, (2)








accuracy connected with the understanding, and (3) the rate at which the

thoughts are being received. The remaining three tests measured only

efficiency of comprehension validly. In an additional study, Carver

(1970a) found that comprehension was measured as well on chunked items

as on multiple choice test items.

Epstein (1967) showed that the introduction of chunking aids in the

learning process when instructions and syntactical structure favor learn-

ing. When favorable conditions do not exist, then chunks inhibit learning.


Language Development and Reading

There have been some studies conducted in which general conclusions

about language development and reading have been made. The results from

these studies do not appear to support any one language processing model.

Chomsky (1972), focusing on the language acquisition of children between

6 and 10, found that there is a "developmental sequence of linguistic

stages through which all children pass." She concluded that "the age at

which different children reach the linguistic stages varies; but the

sequence of states appears to be the same for all" (p. 2). She also

found that I.Q. is positively related to linguistic development across

all stages. Bormuth (1970) found the same relationship between 1.0. and

linguistic development. Smith (1970) showed that the syntactic level at

which a student writes influences or is influenced by the syntactic level

at which he reads.


Levin and Kaplan's Model

Another language processing model has been offered by Levin and

Kaplan (1970). This model attempts to show "that tihe amount of written

material which is picked up and processed by the reader depends on the

amount of predictability within messages" (p. 258). Fillenbaum, Jones,

and Rappoport (19')4) fourd evidence that "the kind of word that can








appear in a sequence of speech is predictable even if speech has been

grossly mutilated" (p. 186). Using the cloze procedure and coding for

correct form-class completions, the authors found evidence for a

"relatively tight, redundant syntactic organization of language" (p. 186).

According to Levin and Kaplan's model, "some parts of sentences are more

predictable than others. The exact pattern or distribution of predicta-

bility appears to depend on the grammatical structure of the sentence"

(Levin, et al., 1972, p. 31). The reader actively searches for

regularities in language structure. He forms hypotheses about subsequent

textual materials which either are confirmed or not confirmed. Reading

is more efficient when grammatical predictability leads to the confirma-

tion of correct hypotheses. "The more certain the reader is about what is

coming next, the less closely he must sample the text. With increasing

uncertainty about what is to follow, sampling must become more detailed"

(Levin, et al., 1972, p. 38).


EVS The Measurement Tool

The tool used by some researchers in measuring the effects of

predictability on the reader has been the eye-voice span. By definition,

the eye-voice span (EVS) is "the distance usually measured in words, that

the eyes are ahead of the voice" (Gibson & Levin, 1975, p. 360).

Research has demonstrated that the size of thie eye-voice span is

not constant. It fluctuates to reflect differences in the predictability

of sentences. Three experiments were conducted to determine the influence

of different orders of approximation to the English language on EVS

(Lawson, 1961; Morton, 1964a; Morton, 1964b). The construction of the

different orders of approxim.t ion was done in the following way: "A

second order list was one or which the subject added the next word to a








sentence of which he could see only a preceding word. A 12th order list

was one in which the subject added the next word to a sentence of which

he could see the preceding eleven words, and so on" (Taylor & Moray, 1960,

p. 7). Results of the three experiments indicated that EVS increased as

the order of approximation to English increased up to the eighth order.

The reason EVS did not increase after the eighth order was attributed to

the observation that the "number of words read ahead was below eight words

even in the highest order of approximation" (Lawson, 1961, p. 54). The

authors concluded that "the pattern of regularities in a passage affects

directly the amount of material in the EVS" (Morton, 1964a, p. 353).

Other experiments have been conducted which "clearly indicate that

the EVS is a flexible unit dependent partially on the regularities within

sentence structure" (Rode, 1974-75, p. 138). Levin and Turner (1968)

found that the EVS was longer for structured sentences than for

unstructured word list material. The authors concluded that all readers

must to some degree take advantage of the regularities in the material

they are reading.

Levin and Kaplan (1968) conducted research which supported their

theory that language is constrained by rules called grammar and that a

reader actively searches for these grammatical constraints or regularities

as he reads through a sentence. They based their study on findings by

Clark (1965) that active and passive sentences differ in constraint.

Clark reported t'at in passive sentences, the verb and actor are highly

constrained by the subject. This is not true of active sentences. Levin

and Kaplan (1968) found the EVS was greater in "the more highly con-

strained passive form" (p. 251). Levin and Jones (1968) did not find

the same results for active and passive sentences in one of their two








experiments. The findings from the second study supported the Levin and

Kaplan study. Levin and Jones stated that they could not account for the

differing results.

Levin, et al. (1972) and Rode (1974-75) conducted experiments which

support the findings of Levin and Turner (1963) and Levin and Kaplan (1968)

that "the syntactic structure of the stimulus materials would govern the

EVS" (Rode, p. 127). Levin, et al. (1972) using right and left embedded

sentences were concerned with examining intra-sentence constraints to

determine the extent to which these constraints are actually exploited

by the reader. Using a modified cloze test designed to measure the

constraints in sentences, the authors found that right embedded sentences

had more constraint than left embedded sentences. In three additional

experiments, Levin, et al. (1972) found that EVS was larger in the right

branching, indicating "that processing is directed by the nature of the

constraint within sentences" (p. 30). The right branching sentences,

which had been shown to have more constraint in the constraint analysis,

had longer EVS measurements.

Rode (1974-75) measured syntactic constraint in verb and noun phrases

which made up two five-word clauses in each sentence. Rode found that

"the unit of decoding is strongly affected by phrase type" (p. 135).

Using subjects from third, fourth, and fifth grade classes, Rode showed

that the type of phrase, verb or noun, has a strong effect on eye-voice

span. The EVS was found to terminate at verb phrases significantly more

often than at noun phrases. The data indicated that the youngest group

completed more phrase units than the older two groups. The older subjects

"attempted to decode a unit of meaning which was a clause rather than a

phrase" (p. 137). Since thr m,'an EVS for each group of subjects was

not markedly different (2.01 words for third grade subjects to 2.51 words








for fifth grade subjects), the older subject had to read four words in

order to extend to a clause boundary. Rode concluded that "the EVS is a

flexible unit of decoding that children uise in the reading process"

(p. 137).

The preceding studies demonstrate the sensitivity of the EVS as a

tool for measuring the amount of contextual constraint or predictability

in a sentence.


EVS and the Unit of Decoding

Studies on EVS indicate that readers attempt to 'chunk' the words

in a meaningful way. In a 1970 study by Hochberg, Levin, and Frail (cited

in Levin & Kaplan, 1970), students in grades two through five were asked

to read a text in which spaces between words were filled in by a constant

but meaningless symbol. When compared to their performance on an untreated

text, second graders showed little difference in leading either text.

Fifth graders, however, were strongly affected by the interword spaces.

The authors interpreted the results in this way: "The younger children

were readinQ the text word by word, so that the lack of space did not

hamper their relevant processing units. Older children, who are apparently

forming units that are larger than a word, are unable to use these higher

order units when important cues interword spaces are not available"

(p. 210).

Rode (1974-75) did not agree with Hnchberg, Levin, and Frail that

younger students read the text word by word. Rode suggested that the

use of active-passive voice sentences by the authors as a measure of

young readers' ability to read to phrase boundaries may have been the

reasons for the results. The younger students may not have been able to

alter their EVS due to a lai"k of reading exposure to active-passive








voice sentences. Rode found that fifth grade subjects in her study

completed more clause units than third and fourth graders, while younger

subjects completed more phrase units than older subjects. The indication

was that "children in the beginning stage of the reading process utilized

syntax to about the same degree that older more accomplished readers were

using it, presumably to guide the decoding process" (p. 138). Older

students were decoding or chunkingg' a unit of meaning which was a clause

"while younger students were completing a unit of meaning which was a

phrase" (p. 137).

Levin and Turner (1970) reported that "fast or good readers read to

phrase boundaries more often than slow or poor readers" (p. 196). Resnick

(1970) found that both eye-voice span and stops at phrase boundaries

increased from third grade to college. Levin and Kaplan (1968) showed

evidence that "phrases act as units in the perception of sentences"

(Gibson & Levin, 1975, p. 366). Levin and Jones (1968) found that the

tendency to read to phrase boundaries increases as one progresses

through the sentence.


Related Research

EVS has been shown to be readily affected by the difficulty of the

material (Anderson, 1937; Fairbanks, 1937; Levin F Cohn, 1968; Stone,

1941). The more difficult the material, the shorter the EVS. Good

readers have been shown to have longer EVS scores when compared to noor

readers (Buswell, 1920; Levin & Turner, 1968; Morton, 1964a). Levin

and Kaolan (1968) found that even noor readers make some use of sentence

structure in their reading. better readers were found to make more use

of grammatical structure in thrir reading than poor readers (Resnick,

1970; Tinker, 1965).









There is contradictory evidence on whether the position within a

line has an effect on EVS. Buswell (1920) found that EVS had no effect

on position within a line. Levin and Turner (1966) reported that J. Q.

Quantz in 1897 found the longest EVS at the beginning of a line, medium

length in the middle, and the shortest at the end of the line. Fairbanks

(1937) found that the length of the EVS was more dependent on the diffi-

culty of the material than on the position within a line. Levin and

Jones (1968) found that EVS was longer at the end of the sentence than

at the beginning.

Levin and Wanat (1967) showed that EVS varied with changes in deep

structure. EVS was found to "validly discriminate between sentences with

the same surface structure but with differing deep structure" (p. 237).


Summary

The two main language processing models, the deep structure model

and the chunk model, are diametrically opposite in their position. The

deep structure model states that the more complex a sentence 'in type or

in number of transformations, the more difficult the sentence is to

process. The chunk model takes the position that the more complex a

sentence, the easier it is to process. Levin and Kaplan's language

processing model appears to come closer in position to the chunk model.

Levin and Kaplan state that the reader searches for regularities in

language structure. The amount of language processing which occurs is

dependent on the amount of predictability or constraint in the text.

Levin and Kaplan have shown that the EVS was greater in the passive

sentence. Since the passive sentence is a transformation of an active

sentence, the findings appear to support the chunk model. The work by

Levin, et al. on right and left embedded sentences does not add support





28


to either the chunk or deep structure model since both sentence types

involve the same type of transformation. The question of which language

processing model is correct remains open to further investigation.
















CHAPTER III

COLLECTION AND TREATMENT OF DATA


The discussions in Chapters I and II indicated a need to examine the

degree of predictability of different types of sentences in an effort to

determine which one of the language processing models offers the best

explanation of the strategies involved in the reading process. The

purpose of- this study was to determine the degree of predictability of

specific sentence types.


The Sample

The participants in this study were selected from a population which

consisted of University of Florida graduate students in Reading and

English. It was necessary that the subjects selected be able to read

vocabulary from the Functional Reading Word List for Adults (Mitzel, 1966)

and to read sentences involving embedded transformations and transforma-

tions which expand the basic sentence. These two selection criteria were

testable. In order to minimize the number of subjects who would have to

be screened to determine if they could successfully perform the reading

tasks required, subjects were limited to graduate students. All graduate

students are required to score at least 1,000 or 1,100 on the Graduate

Record Exam to be admitted into one of the graduate education programs.

Because the Graduate Record Exam requires advanced reading skills,

successful performance on the exam, as measured by the College of









Education's minimum required scores, was used as the initial screening

device. The screening device was helpful since only one subject of the

total tested had to be eliminated because of an inability to perform the

reading tasks successfully. The subject had difficulty reading orally

because of numerous regressions and errors in word attack.

Of the 27 remaining subjects who participated in the reading tasks,

seven had to be eliminated. Six of the subjects were dropped from the

study because each failed to read at least one target sentence correctly

making it impossible to take an EVS measure at the critical point. Either

the subject skipped a word preceding the critical word or skipped the

critical word itself. An additional subject was dropped because the

examiner failed to blacken the screen one time when the subject read a

critical word.



The Instrument

Four sentence types were chosen to compose the design. They included

the following:

1. a basic sentence and its THERE transformation

2. an active sentence and its passive transformation

3. a basic sentence and its WII-question transformation

4. a basic sentence and its IT-inversion transformation with

an infinitive subject nominalization

One sentence type, the active versus the passive transformation, was

selected because of the conflicting results of three studies. Levin and

Kaplan, in a 1968 study usinq active and passive sentences had shown

that passive sentences had longr EVS measures than active sentences.

Levin and Jones (1968) did not find the same results in one of their two

experiments using active and passive sentences. The inclusion of active









and passive sentences in this study was done to obtain additional

information on this type of transformation.

The remaining three sentence types were chosen because each of the

sentences involves a process in which the subject is moved from the

beginning of the sentence to a position following the verb or modal.

In addition, each of the four sentence types was chosen because of

the number of transformations that are involved. The passive transforma-

tion and the THERE transformation involve one transformation. The

WH-question transformation involves two transformations, and the

IT-inversion transformation with an infinitive subject nominalization

has three transformations. Basic sentences were written for each of

these sentence transformations so that a basic sentence involving no

transformations could be identified with each sentence type. As a

result, it was possible to collect infonnation on the following

questions:

1. Is there a difference in the degree of predictability between

each basic sentence type and its transformed sentence type?

2. Does the number of transformations within a sentence affect

the degree of predictability?

3. Does the type of sentence transformation affect the degree of

predictability?

4. Does the position in the sentence where the eye-voice span is

measured affect the degree of predictability?

In order to control for vocabulary, words used in the sentences

were limited to the Functional Reading Word List for Adults (Mitzel,

1966). In developing the Functional Word List, which consists of 5,000

words, Mitzel attempted to "identify the basic words an adult needs to









know by studying all the sources of reading material to which the general

public is exposed" (p. 67). The sources included the following:

1. material issued by the U. S. Government directed toward the

general public, such as pamphlets on Social Security Benefits,

Workman's Compensation, Civil Defense, etc.

2. material issued by local governments on such issues as driving

rules and regulations, housing, etc.

3. newspapers: front page news, feature articles and classified

ads (Afro-American, News-Post, Racing Form, Midnight, Tab,

Daily News)

4. Application blanks for employment, credit, insurance, mortgage,

etc.

5. signs in grocery stores, hardward stores, department stores,

liquor stores, restaurants, etc.

6. menus

7. the Yellow Pages of a large metropolitan telephone directory

8. comic books (Herbie, Superman)

9. general advertising literature, including that which is mailed

to the home or hand distributed, billboards, newspapers and

magazine ads, match book covers

10. union literature

11. religious tracts that are widely distributed

12. TV commercials

13. political campaign literature and sample ballots

14. labels on packages

15. magazines (Jet, Awake)

According to Mitzel, an adult who mastered all 5,000 words on the

Functional Word List would be considered literate because the words are








"representative of what an adult would be able to understand in order to

make effective decisions" (p. 69).

The eye-voice span was chosen as the measurement tool based on

research which demonstrated that the EVS can measure the amount of

material which is picked up and processed by the reader.

A critical point (a predetermined place where the eye-voice span

was measured) was chosen for each of the sentences. Each basic sentence

and its transformed sentence type contained the same number of words to

control for the number of words before and after each critical point.

Except for the active and passive sentences, the first critical point

followed the introductory word in each sentence. The second critical

point followed a verb or modal. The third critical point followed a

one-word noun or adjective. The fourth and fifth critical points

followed three-word phrases. Each basic sentence and its transformed

sentence had the same number of words before each critical point and

two four-word phrases following the last critical point. One sentence

was used for each critical position within each sentence.

Each target sentence (a sentence in which an EVS measurement was

taken) was embedded in a paragraph composed of three unrelated sentences

to control for context clues. In order to control for vocabulary and

readability, the sentences used to compose the paragraphs were limited

to sentences taken from General Reading for Understanding Cards 21-60

(Thurston, 1969). These paragraphs cover a range of difficulty from

seventh to ninth grade.

Because a study by Levin and Kaplan (1968, p. 253) indicated that

subjects scanned the first line of a paragraph before beginning to read

orally, the target sentence was never the first sentence in the paragraph.

The placement of the target sentence was varied among the second, third,








and fourth positions. The position of the critical point on a line was

also varied to control for position effect. After each critical point,

there were at least nine words on the line. This was done to prevent the

possibility of subjects extending their EVS to the last word on the line.

A minimum of nine words was chosen because exploratory data by Levin and

Kaplan (1968), McConkie and Rayner (1975), Rayner (1975a) and Rayner

(1975b) using undergraduate subjects indicated that at least eight words

was sufficient. Since graduate subjects were used in this study, at

least nine words followed each critical point. This proved to be

sufficient since no subject read to the end of the line during the study.


Collection of Data

Each paragraph was projected onto a Caramatt II screen which presents

a rear projected image. The subject was asked to sit directly in front

of the screen. The same instructions were read to each subject. The

instructions were: "A paragraph will be projected onto the screen

directly in front of you. The paragraph will consist of four unrelated

sentences. When a slide appears on the screen, you will begin reading

the paragraph orally. At some predetermined point in your oral reading,

I will push a remote control button which will remove the paragraph and

leave a blank screen. When this occurs, you should report as much of

the rest of the material as you remember seeing beyond the last word you

said. The number of correct consecutive words reported will be taken as

your EVS measurement for that paragraph. The EVS is defined as the

number of words that your eyes are ahead of your voice. Two practice

paragraphs will be presented. The total study will consist of 40

paragraphs and therefore, 40 FwS measures. A tape recorder will be

used to record exact responses. There are no right or wrong answers.









We will start with the first practice paragraph." Two practice paragraphs

were presented to assist the subject in his understanding of the task.

The Caramatt IT had a remote control button which allowed the

examiner to sit behind the subject during the data collection. The

subject, therefore, was unable to see the examiner when the remote

control button was pushed. This precaution prevented the subject from

observing the actions of the examiner and visually anticipating when the

critical point would occur.

When the paragraph appeared on the screen, the subject began reading

the paragraph orally. When the subject reached the critical point in the

paragraph, the examiner pushed the remote control button and the paragraph

was removed. This removal was accomplished by leaving a blank slot in

the slide tray following each slide. The contrast between the letters

and the background was sufficiently low to eliminate any afterimage when

the paragraph was removed. The subject then reported the words he

remembered seeing beyond the last word he said when the screen was

blackened. The number of correct consecutive words reported was taken

as his EVS measure for that sentence. A tape recorder was used to record

exact responses.

To control for a learning effect because of the repeated measures

design, the order of paragraph presentations was randomized independently

for each subject.


Treatment of Data

The EVS scores were obtained for each subject by taking the number

of correct consecutive words reported after the screen was blackened. If









a subject skipped words, the EVS score was the number of correct consecu-

tive words reported before the skipped word or words. For example, if

the words that followed a critical point were, "at the scene among the

motorists" and the subject reported, "at the scene among motorists," the

EVS measurement would be four. If a subject reported a word incorrectly,

the EVS measurement would be the number of correct consecutive words

reported before the incorrect word. For example, if the words that

followed a critical point were, "is dangerous on mountain trails," and

the subject reported, "is dangerous on mountainous trails," the EVS

measurement would be three.

A few subjects had a zero EVS measurement on a sentence. This

occurred when a subject skipped the initial word following the critical

point. A zero EVS measurement occurred only three times. Each time it

was with a different subject and a different sentence so there was no

pattern to the zero responses.

The data were analyzed within a randomized block factorial design

with repeated measures. A randomized block design was used because the

design met the three conditions that are necessary according to Kirk

(1968, p. 131), for the appropriate use of such a design:

1. One treatment with K = two or more treatment levels. In this

design, there were 40 treatment levels or 40 EVS measurements.

2. 'Assignment of subjects to blocks so that the variability

among subjects within any block is less than the variability

among the blocks. The number of subjects and observations

within each block must be equal. In this design, there was

one subject in each block.

3. Random assignment of treatment levels to the experimental

units within each block.








The design was also a factorial one since a factoriall experiment

refers to the simultaneous evaluation of two or more treatments in one

experiment rather than to a distinct kind of experimental design" (Kirk,

1968, p. 171).

The statistical treatment used was an analysis of variance on the

4x2x5 randomized block factorial design. Factorial analysis of variance

is the "statistical method that analyzes the independent and interactive

effects of two or more independent variables on a dependent variable"

(Kerlinger, 1973, p. 245). The independent effects were sentence type,

basic versus transformed sentences, and critical points. The interactive

effects were the interactions between the three main effects. The

dependent variable was the EVS measurements.

The assumptions underlying a randomized block design model are that

the population covariances between pairs of treatment levels are constant

and that the population variances for each of the treatment levels are

homogeneous. The hypothesis tested was that there is no significant

difference in the degree of predictability among the sentence types used.

The following model was used for the design:


EVSijkl + [i + j + Rk + T1 + Bjk + jl + 6kl +

'FYjk + Eijkl


where 1 = effect of subjects

= affect of types of transformations

0 effect of basic versus transformed sentences

} effect of critical points









The hypotheses tested were:



HI : a.



H: k
Ho k


Hi: k
H y




110: a?
ol: Pjk

II : a .

o ik1


H1: O.lk


H1: Rk


H : fRY .k
0: jk
o' jkkl


- o for all

So for some j


o for all k

a o for some k


=o for all 1

So for some 1


= o for all jk

/ o for some jk


So for all jl

/ o for some i1


= o for all k1

/ o for some kl


= o for all jkl

/ o for some ikl


Before any of the hypotheses could be tested, it was necessary to

test the assumptions underlying the design. Chapter IV explains the

results from the tests of the assumptions and the hypotheses.
















CHAPTER IV

ANALYSES OF THE DATA


This study was concerned with determining the degrees of within-

sentence predictability among four sentence types. Graduate students in

Reading and English read orally sentences which were developed by the

investigator. Eye-voice span measures were taken at predetermined places

in the sentences. The data were analyzed using a randomized block

factorial design with repeated measures. The hypothesis tested was that

there would be no significant difference in the intra-sentence predicta-

bility of the sentence types used.

The following questions were raised in the study:

1. Is there a difference in the degree of predictability as

measured by the eye-voice span between

a) a basic sentence and its THERE transformation

b) an active sentence and its passive transformation

c) a basic sentence and its WH-question transformation

d) a basic sentence and its IT-inversion transformation with

an infinitive subject nominalization?

2. Does the number of transformations within a sentence affect the

degree of predictability?

3. Does the type of sentence transformation affect the degree of

predictability?

4. Does the position in the sentence where the eye-voice span is

measured affect the degree of predictability?








Findings Related to the Assumptions

The assumptions underlying the model for a randomized block design

are that the population covariances between pairs of treatment levels are

constant and that the population variances for each of the treatment

levels are homogeneous. To test these assumptions a test for the symmetry

of the variance-covariance matrix as described by Box (1950) was done.

The following hypotheses were Lested:

2 2
H : j and Pn = o
o 1 11

H : C2j and Po 2j # o


Because the observed value of chi-square was larger than the tabled value,

the null hypothesis was rejected. For a detailed explanation, see

Appendix C. When the data depart from the required symmetry, a conser-

vative F test is recommended.


Findings Related to the Hypothesis

Before an evaluation of the conservative F test could be done, the F

value for each of the main effects and interactions of the randomized

block factorial design had to he determined. Usinq the BMD08V packaged

program design, an analysis of variance was done. The following

hypothesis was tested: There is no significant difference in the degree

of predictability amonq the sentence types used. The following model

was used:


EVSijkl = u + Ti + lj + nk1 + Ikl + jk k + "jkl +


ijkl


where T = effect of subjects

o = effect of types of transformations

6 = effect of basic versus transformed sentences

y = effect of critical points








The hypotheses tested at the .05 level of significance were

H : = o for all j

HI: ja / o for some j


H: k = o for all k
o k
T1: Bk 7 o for some k


H : I = o for all 1
o 1
Ill: Y1 / o for some 1


Ho: ajk = o for all jk

H1: Bjk o for some jk


Ho: aYj = o for all jl

HI: ayj o for some jl


H : kl = o for all kl
o kl
I: By 1 o for some kl


Ho: .jkl = o for all jkl
HI: aBSjkl / o for some jk]


The cell means are reported in Table 1.

The residual was determined by pooling the sums of squares of all

interaction terms that contained a subject effect. This pooled residual

allowed for the control of the maximum amount of error in the design.

The summary table for the randomized block factorial design is found in

Table 2.

Using the conservative F test with 1 and 19 degrees of freedom, the

tabled value of F for the three-way interaction TxBxC was found to be

4.38. Since the F value for the three-way interaction was 7.3848184, the

conservative F test was significant. Tf the conservative F test for

treatment effects is significant, an exact test will also be significant.

(Kirk, 1968, p. 143). Thus, Ihe null hypothesis was rejected. The significant





42



TABLE 1

Cell Means for the Randomized Block Factorial Design

Type of Transformation
THERE Passive WH-Ouestion IT-Inversion
5.51 4.34 4.60 5.21

Basic vs. Transformation
Basic Transformation
5.15 5.18

Critical Point
1 2 3 4 5
5.48 5.09 5.04 5.26 4.94

Basic Transformation
] 2
THERE 5.81 5.21
Passive 5.24 5.43
WH-Question 5.47 4.72
IT-Inversion 5.06 5.37

Critical Point
1 2 3 4 5
THERE 5.13 5.68 5.58 5.58 5.60
Passive 5.50 4.63 6.05 5.80 4.70
WH-Question 5.68 4.88 4.03 4.30 4.10
IT-Inversion 5.68 5.20 4.53 4.38 5.35

Critical Point
1 2 3 4 5
Basic 5.31 4.95 5.24 5.65 4.58
Transformation 5.65 5.24 4.85 4.88 5.30

THERE Transformation


Critical Point 1 2
Basic 5.65 5.95
Transformation 4.60 5.40


6.55
4.60


5.80
5.35


Passive Transformation
Critical Point 1 2 3 4 5
Basic 5.45 4.05 5.80 6.70 4.20
Transformation 5.55 5.20 6.30 4.90 5.20

WH-Ouestion


Critical Point 1 2
Basic 5.80 4.90
Transformation 5.55 4.85


3 4 5
4.20 4.10 3.35
3.85 4.50 4.85


IT-Inversion


Critical Point 1
Basic 4.35
Transformation 6.90


2 3
4.90 4.40
5.50 4.65


4 5
6.00 5.65
4.75 5.05


-_-----------~-------~-


IT-Inversion












Summary Table for


TABLE 2

the Randomized Block


Factorial Design


Source



Type(T)

Basic vs. Transformation(B)

Critical point (C)

Tx B

T xC

B xC

Tx B x C

Subjects

Residual


Sum of Degrees of Mean Conventional
Squares Freedom Square F


95.06

.18

3.22

27.45

148.79

58.64

150.40

348.17

1257.61


31.69

.28

7.24

9.15

12.40

14.66

12.53

18.32

1 .70


18.67*a

.17

4.27

5.39*

7.31*

8.64*

7.38*

10.79*


Total 799


Astericks indicate significant F's.









three-way interaction indicated that there are significant differences

among the sentence types in the degree of predictability.

In order to determine the nature of the interaction, tests of the

simple interaction effects and simple simple main effects were done. The

hypotheses were tested at the .01 level of significance (see Table 3, 5,

and 7). Tukey tests were conducted to determine where the differences

were significant between the T means and the C means (see Tables 4 and 6).

The data indicated that a complex interrelationship exists among the

three independent variables types of sentence transformation, basic

versus transformation, and critical point.


Findings Related to the Questions

Question One

Is there a difference in the degree of predictability as measured by

the eye-voice span between

a) a basic sentence and its THERE transformation

b) an active sentence and its passive transformation

c) a basic sentence and its WH-question transformation

d) a basic sentence and its IT-inversion transformation with

an infinitive subject nominalization?

When the differences between the basic sentences and the transformed

sentences were examined, it was found that there were no significant

differences between the basic sentences and the transformed sentences in

fourteen out of the twenty CT combinations.

A significant difference at. the .01 level was found between the

basic sentence and the transformed sentence at the combination of critical

point three and the basic versus the THERE transformation (CT ). At

critical point three, the basic sentence had a significantly higher EVS








TABLE 3

Analysis of the Interaction of Critical Point
by Type of Transformation


Sum of Decrecs of Mean
Squares Freedom Squares


181.62

117.58



25.94

36.30

76.34

73.00

61.45

53.70

4.94

64.10

7.65

18.30


15.13 8.92*a

9.80 5.77*


8.65

12.10

25.45

24.33

20.48

17.90

1.65

21.36

2.55

6.10


Residual


aAstericks indicate significant F's.


Table 4 indicates where the differences were significant among the T
means.


Source


SSCT at B1

SSCT at B2


F


b
SST

SST

SST

SST

SST

SST

SST

SST

SST

SST


BC1

BIC2

B C3
1

BIC

B1C5

2C1

B2C2


B2C3

BC
2C4

B2C5
22IC

B 2CI
B23C

B 2C3
B2 C4

B 2C5


5.09*

7.13*

14.99*

14.34*

12.07*

10.55*

.97

12.59*

1.50

3.60





46


TABLE 4

Difference of the T's


T3

4.90


T2 T4
2 4

5.45 4.35



T4 T2

4.90 4.05


T1 2 4 T3

6.55 5.80 4.40 4.20



T2 T4 T1 T3

6.70 6.0 5.8 4.1



TI T T T
T4 T1 T2 T3

5.65 5.10 4.20 3.35


T2 T3 T1

5.55 5.55 4.6



T4 TI T3

4.65 4.60 3.85


B C

1 1


13

5.80


T4

6.90








TABLE 5

Analysis of the Interaction of Basic Versus
Transformation by Critical Point


Sum of Dearees of Mean
Source Squares Freedom Squares F


SSBC at T 46.10 4 11.52 6.79*a

SSBC at T2 56.42 4 14.10 8.31*

SSBC at T3 22.85 4 5.71 3.37*

SSBC at T4 83.67 4 20.92 12.32*






SSC at B T1 21.94 4 5.49 3.23

SSC at B T2 99.74 4 24.94 14.69*

SSC at B T3 68.36 4 17.09 10.07*

SSC at B T4 43.94 4 10.99 6.47*

SSC at BIT1 31.84 4 7.96 4.66*

SSC at B T2 23.16 4 8.04 4.74*

SSC at B2T 30.56 4 7.64 4.50*

SSC at B T4 67.26 4 16.82 9.91*


Residual 741 1.70



aAsterioks indicate significant F's.


Table 6 indicates where the differences were significant among the C
means.





48



TABLE 6

Difference of the C's


6.70 5.80


5.45


4.20 4.05


5.80 4.90 4.20 4.10


1 4


6.00 5.65 4.90 4.40 4.35




C5 C2 C4 Cl C3

6.1 5.40 5.35 4.60 4.60


6.3 5.55


5.20 5.20 4.90


4.50 3.85


5.50 5.05 4.75 4.65


~


2 4 6.90









TABLE 7

Analysis of the Interaction of Type of Transformation
by Basic Versus Transformation

Sum of Degrees of Mean
Source Squares Freedom Squares F

SSBT at C1 72.22 3 24.07 14.18*a

SSBT at C2b 16.57 3 5.52 1.92

SSBT at C3 36.39 3 12.12 7.14*

SSBT at C4 27.63 3 9.20 5.43*

SSBT at C, 8.36 3 8.36 4.92*


SSB at C T1 11.03 1

SSB at ClT2 .10 1

SSB at C T3 .63 1

SSB at C1T4 65.03 1

SSB at C3T1 38.03 1

SSB at C3T2 2.5 1

SSB at C3T3 1.23 1

SSB at C3T4 .63 1

SSB at C4T1 2.03 1

SSB at C T2 32.40 1

SSB at C4T3 1.60 1

SSB at C4T4 15.63 1

SSB at C T1 10.00 1

SSB at C5T2 ]0.00 1

SSB at C5T3 22.5 1

SSB at C5T4 3.6 1

Residual


Astericks indicate significant I's.

b
Since SSBT at C2 was not significant,
conducted:

SST at C2 24.69 3

SSB at C2 3.01 1

Difference of the T's at C
TI 14 2T3
5.675 5.200 4.875


11.03

.10

.63

65.03

38.03

2.50

1.23

.63

2.03

32.40

1.60

15.63

10.00

10.00

22.50

3.60







the following tests were


-.22

3.31



T2
4.625


6.50

.06

.37

38.31*

22.40*

1.47

.72

.37

1.19

19.09*

.94

9.21*

5.89

5.89

13.26*

2.12


4.85*

1.95









measurement. The EVS measurements for the basic sentence and the THERE

transformation are presented in Figure 2. Examples of the two sentences

follow. The astericks indicate the significant difference in EVS between

the sentences at critical point three.

*Significantly longer EVS
1 2 3 4
Basic: Officials/were/waiting/at the scene/among the motorists

5
near/the old road by the huge stadium.

1 2 3* 4 5
THERE: There/were/men/on the docks/near the machinery/for the

first time since the latest injury.


At critical points two and four, significant differences were found

between the active sentence and the passive sentence types (CT22) (CT42).

The passive sentence had a longer EVS measurement at critical point two.

At critical point four, the active sentence had the longer EVS measurement.

The EVS measurements for the active and passive sentences are shown in

Figure 3. The following are examples of the two sentences. The astericks

indicate the significant differences in EVS between the sentences at

critical points two and four.
Significantly longer
EVS
1 2* 3 4* 5
Active: The/security/guard/was being careless/about the locks/on

the glass doors of the main building.

Significantly longer EVS
1 2* 3 4* 5
Passive: The/young/manager/was being introduced/to the personnel/

from the retail division of the drug company.


A significant difference was found between the basic sentence and the

WH-question transformation at critical point five. The WH-question

transformation had a longer E'S measurement at critical point five.

Figure 4 shows the EVS measurements for the basic sentence and the






51





















40 44



S0 0
3 0 u2 0

a) M I
U) E- 4 a)3
44 r d C)





4 H








44J 41

















IA(
-H-HO m - C



/ 43

/ \ -P--1











/1


\ / P^



\ / (D


I_ _

























r
0




0
IJ



m o









.H u
\u o
to








to I I >
to C)-H
















\-> H C
\ 0o
SMI
14'












w td










r 444

.1 \ N 4
C-i



















m





oo
S4
-H U(











C-
\ -H I



















CU-i










C-t

H^
C4'1


~ ______ ~__






53











uo
C O U
0 44 0


H!H O



0t -0
4j


SH 0 I! U


4- 40
C -H -H 0 1 n3 *-H



f0



















u 0 c
U 7 U nj '




























r m
0 )


04f4.
Sl\-41
U o-
S0I




/ \-4







/ 4-)

















Sll
IT
S-f4



/ 5









WH-question transformation. Examples of the two sentences follow. The

astericks indicate the significant difference in EVS between the sentences

at critical point five.


1 2 3 4
Basic: Vetcrans/will/usually/receive their checks/for partial

5,Significantly longer EVS
disability/within a few days after the first of the month.


1 2 3 4
WH-question: When/will/psychologists/find an answer/for the

5*
cause/of severe emotional problems in very young

children?

At critical point one and four significant differences were found

between the basic sentence and the IT-inversion transformation with an

infinitive subject nominal ization. The IT-inversion transformation had

a longer EVS measurement than the basic sentence at critical point one.

The basic sentence had a longer EVS measurement than the IT-inversion at

critical point four. The EVS measurements for the basic sentence and

the IT-inversion with an infinitive subject nominalization are found in

Figure 5. The following are examples of the two sentences. The astericks

indicate the significant differences in EVS between the sentences at

criticalpoints one and four.

Significantly longer

1 2 3 4* 5
Basic: Exercise/is/difficult/in the beginning/for many individuals/


wilh major health problems from poor eating habits.

Significantly longer EVS
1* 2 3 4 5
IT-inversion: It/is/important/to examine lakes/in remote areas/to

identify the needs of the different fish.











































4-



c4
*H T) >




Ul Ti *
CU-]


41


Cl


C'4-4
H H0



V'
mTi


'i
ul


SA3


'Ti
4J
.CU

-,4 -r

340



ION
C In -H
0 44
-H ()I mo







Ir- *H








Question Two

Does the number of transformations within a sentence affect the

degree of constraint?

When the sentence was a basic one, there were significant differences

between transformation types at all the critical points. At critical

point one, both the basic sentence for the WH-question transformation and

the basic sentence for the THERE transformation had significantly longer

EVS measurements than the basic sentence for the IT-inversion transforma-

tion. At critical point two, the basic sentence for the THERE transforma-

tion had a significantly longer EVS measurement than the basic sentence

for the passive transformation. At critical point three, the basic

sentence for the THERE transformation had a significantly longer EVS

measurement than the basic sentence for the WH-question transformation

and the IT-inversion transformation. Also, at critical point three, the

basic sentence for the passive transformation had a significantly longer

EVS measurement than the basic sentence for the WH-question and IT-

inversion transformations.

At critical point four the basic sentence for the WH-question

transformation had a significantly shorter EVS measurement than each of

the other basic sentences. At critical point five, the basic sentence

for the IT-inversion transformation had a significantly longer EVS

measurement than the basic sentence for the WH-question and the passive

transformation. Also, at critical point five, the basic sentence for the

THERE transformation had a longer EVS measurement than the basic sentence

for the WH-question transformation.

When the sentence was a transformed one, there were no significant

differences between the trarnsformed sentences at critical points two,

four, and five. At critical point one, the IT-inversion transformation








had a significantly longer EVS measurement than each of the other three

sentence transformations. At critical point three, the passive transfor-

mation had a significantly longer EVS measurement than each of the other

transformed sentences.


Question Three

Does the type of sentence transformation affect the degree of

predictability?

When the basic sentences were examined, it was found that at critical

points one, three, four,and five there were no significant differences

between the basic sentence for the THERE transformation and the basic

sentence for the passive transformation. There were significant

differences between the other two basic sentences, and there was no

pattern evident.

When the transformed sentences were examined, it was found that at

critical points one, two, four,and five there were no significant

differences between the THERE transformed sentence and the passive

transformed sentence (see Figure 6). Each of the sentences has one

transformation. Examples of the two sentences follow. The astericks

indicate the significant difference in EVS between the sentences at

critical point two.

Significantly longer EVS
1 2* 3 4 5
THERE: There/were/men/on the docks/near the machinery/for the

first time since the latest injury.

1 2* 3 4 5
Passive: The/old/doq/was being taken/by his owner/to the

dedicated veterinarian in the old neighborhood.























0



O.,
-H m-O
41 441



0 C
0J 4-1


* / m



111
' E 0
-H CD C 43






Di H C
UD /4 a)

z C / 0n
C, a,

13




H C C



'H iN 7














N 0














tP
v-H





















0'
.r4









Question Four

Does the position in the sentence where the eye-voice span is

measured affect the degree of predictability?

Significant differences were found between the means for the

critical points at all but one of the transformation types and basic

versus transformation combinations. When the sentence was a basic

sentence for the THERE transformation, there were no significant

differences among the critical points.

There were differences between the critical point means when the

sentence was a basic sentence for the passive transformation. Critical

points four, three, and one had significantly longer EVS measurements

than critical points five and two. When the sentence was a basic

sentence for the WH-question transformation, there were significant

differences at the following critical points. Critical points one and

two were significantly longer in EVS measurements than critical point

five. Critical point one was also significantly longer in EVS than

critical points three and four. Critical point four had a significantly

longer EVS than critical point five. When the sentence was a basic

sentence for the IT-inversion transformation, critical points four,

five, and two were significantly longer in EVS than critical point one.

Critical point four also had a significantly longer EVS than critical

point three.

With the THERE transformation, critical points five and two had

significantly longer EVS measurements than critical points one and three.

The passive transformation showed that critical points three and one had

significantly longer EVS measurements when compared with critical point

four. With the WH-question transformation, critical points one, two, and

five had significantly longer EVS measurements than critical point three.





60



The IT-invcrsLion transformation showed that critical point one had a

significantly longer EVS measurement than critical points two, three,

four, and five (see Table 6).
















CHAPTER V

SUMMARY CONCLUSIONS AND IMPLICATIONS


Summary

This study examined the degree of predictability of specific

sentence types. The following questions were raised:

1. Is there a difference in the degree of predictability as

measured by the eye-voice span between

a) a basic sentence and its THERE transformation

b) an active sentence and its passive transformation

c) a basic sentence and its WH-question transformation

d) a basic sentence and its TT-inversion transformation

with an infinitive subject nominalization?

2. Does the number of transformations affect the decree of

predictability?

3. Does the type of transformation affect the degree of predicta-

bility?

4. Does the position in the sentence where the EVS measurement is

taken affect the degree of predictability?

The study sample consisted of 20 University of Florida Reading and

English graduate students.

Sentences were developed by the researcher. The eye-voice span was

used as the tool for measurement because it is sensitive to the effects

of different types of predictability on the reader. EVS can also be used









to examine the amount of material which is picked up and processed by

the reader.

A randomized block factorial design with repeated measures was used

to test the hypothesis of no significant differences in the intra-sentence

predictability of the sentence types used in this study. The data

indicated a significant three-way interaction among types of transfor-

mations, basic versus transformation, and critical points. An analysis

of the simple interaction effects showed significant differences for all

but one of the sums of squares (see Table 7). When simple simple main

effects were examined, twenty of the thirty-eight were significant.

Fifteen Tukey tests were conducted to determine which means were signifi-

cantly different.

Further analysis indicated that there was a complex interrelationship

among the three independent variables type of transformation, basic

versus transformation, and critical point on the eye-voice span

measurement.

There was no evidence from the data to support any of the language

processing models.


Conclusions

Related to the Hypothesis

The hypothesis was that there would be no significant differences in

the degree of predictability among the sentence types used in the study.

Specific sentence types were chosen, and original sentences were developed.

The vocabulary and the number of words in the sentences were controlled.

Any differences between the CVS measurements would indicate a difference

in the ability to decode the sentence and a difference in sentence

predictability.









The statistical procedure did not support the hypothesis of no

significant difference. Analysis of the data indicated that the three-way

interaction of transformation types, basic versus transformations, and

critical points was significant. Findings from the analysis of simple

interaction effects, simple simple main effects, and Tukey tests

indicated that a complex relationship existed between the three indepen-

dent variables and the eye-voice span measurements. The independent

variables could not be analyzed separately to determine the effect of each

on the eye-voice span measurements. The complexity of the relationship

among the interaction of all three independent variables requires that the

differences in the eye-voice span measurements be examined, taking all

three variables into account.

The conclusion drawn from the data is that the differences in the

degree of predictability between sentence types as measured by the eye-

voice span are dependent on the complex interrelationship among the three

independent variables sentence transformation, basic versus transforma-

tion, and critical point.


Related to the Questions

Question one: Is there a difference in the degree of predictability

as measured by EVS between

a) a basic sentence and its THERE transformation?

There was no significant difference except at critical point three

where the transformed sentence had a significantly longer EVS measurement,

indicating greater predictability only at that point in the sentence. See

Figure 2. The only conclusion which can be drawn is that at critical

point three, the THERE transformed sentence had greater predictability

or was easier to decode than the basic sentence.








Is there a difference in the degree of predictability as measured by

the eye-voice span between

b) an active sentence and its passive transformation?

There were no significant differences between the two types of

sentences except at critical points two and four.

Significantly longer
EVS
1 2- 3 4* 5
Active: The/security/guard/was being careless/about the locks/on

the glass doors of the main building.

Significantly longer EVS
1 2* 3 4* 5
Passive: The/helpless/animal/ was being cornered/by the dogs/in an

open field near the winter cabin.

At critical point two the passive sentence had a significantly longer EVS

measurement, whereas, at critical point four the active sentence had a

significantly longer EVS measurement (see Figure 3).

The results from this study do not support the findings of Levin and

Kaplan (1968). The Levin and Kaplan study was based on Clark's research

(1965) which showed that the latter part of the passive sentence, the verb

and the actor, is highly constrained by the first part, the subject. This,

according to Clark, was not true of the corresponding parts of active

sentences. Levin and Kaplan's research indicated that the EVS measurements

for the active and passive sentences followed the constraints of the two

sentences as indicated by Clark. The passive sentence was found to have

a longer EVS than the active at the point where the active and the passive

forms began to be differentially constrained.

The data from the researcher's study showed that there is a dis-

crepancy in results between the Levin and Kaplan study and this one. In

this study, at the point where the active sentence, according to Levin and









Kaplan, is suppose to have a shorter EVS, the EVS for the active sentence

is significantly longer. The critical points for the active and passive

sentences were chosen to correspond to the other three types of sentences

used in this study; however, a comparison can be made between the active

and passive sentences in this study and the Levin and Kaplan study.

Examples of the Levin and Kaplan active and passive sentences and the

critical points follow:

1 2 3 4 5* 6*
Active: The brash tall/man/was/certainly/being/loud/at the

meeting of the new group on the main campus.

Significantly longer

1 2 3 4 5*EVS
Passive: The cute chubby/boy/was/slowly/being/

6*Significantly longer EVS
wheeled/by the maid along the narrow lane to the

country store.

Examples of sentences from this researcher's study follow:
Significantly longer
1 2 3 4*EVS 5
Active: The/reckless/cop/was being impulsive/about the message/
near the murdur scene from the jealous wife.

Significantly longer EVS
1 2* 3 4*
Passive: The/last/survivor/was being examined/by hospital
5
attendants/in a temporary tent near the crash site.

In the Levin and Kaplan sentences, at critical points five and six,

the passive sentence had significantly longer EVS measurements than the

active sentence. In this researcher's study at critical point four, the

active sentence had a significnatly longer EVS measurement than the

passive. Also, this researcher found that the passive sentence had a

significantly longer EVS measiurment than the active sentence at critical

point two. Levin and Kaplan found significant differences only at points

five and six.









One aspect of the language processing model proposed by Levin and

Kaplan (Levin, et al., 1972), therefore, is not supported by these

findings. According to Levin and Kaplan, the EVS should reflect the

differences in constraints of the two sentences. This did occur.

However, instead of the passive sentence having the longer EVS, the

active sentence did. A study by Levin and Jones (1968) also showed the

active sentence having the longer EVS measurement. Levin and Kaplan

(1968) stated that they could not account for the discrepancies between

the studies.

Is there a difference in the degree of predictability as measured

by EVS between

c) a basic sentence and its WH-question transformation?

The WH-question transformation was found to have a significantly

longer EVS measurement than the basic sentence at critical point five.

The two sentences did not show significant differences at the other

critical points (see Figure 4). The only conclusion which can be drawn

is that at critical point five, the transformed sentence had greater

predictability or was easier to decode than the basic sentence.

Is there a difference in the degree of predictability as measured

by EVS between

d) a basic sentence and its IT-inversion transformation with an

infinitive subject nominalization?

Significant differences were found at two of the five critical

points. At critical point one, the IT-inversion sentence had a longer

EVS measurement, whereas, at critical point four, the basic sentence

had a longer EVS measurement (see Figure 5). Therefore, at the

beginning of the sentence, the transformed sentence had greater








predictability or was easier to decode. At critical point four, however,

the basic sentence was easier to decode.

The conclusion drawn from the data is that except for the active-

passive sentences, no general conclusions can be made. In two cases,

the further into the sentence one went, the easier the basic sentence

was to decode. This was true when the active and passive sentences were

compared, and it was true when the basic sentence and the IT-inversion

sentence were compared. In the other two cases, the opposite occurred.

The transformed sentences were easier to decode further into the

sentences. This was true with the basic sentence and the THERE trans-

formation, and with the basic sentence and the WH-question transformation.

Question two: Does the number of transformations affect the degree

of predictability?

The reason this question was asked was to determine if support could

be found for either the deep structure model, which states that the more

transformations in a sentence, the harder the sentence is to read, or the

chunking model, which offers a diametrically opposite explanation. The

data related to this question do not support either theory. When the EVS

measurements for the THERE transformation and the passive transformation,

each of which had one transformation, were compared to the WH-question

transformation, which had two transformations, and the IT-inversion

transformation, which had three transformations, no pattern was discern-

ible (see Table 4). The EVS did not increase or decrease as the number

of transformations increased. In fact, at critical points two, four,and

five, there were no significant differences in the EVS measurement among

all of the transformation types. This indicates that all of the trans-

formed sentences at critical points two, four,and five had the same degree

of predictability as measured by EVS. There was no discernible difference









between the transformed sentences at those critical points. At critical

point one, the IT-inversion transformation had the longest EVS measurement

when compared to the other three transformations. At critical point

three, the passive sentence had the longest EVS measurement.

Because a longer EVS measurement indicates greater ease in decoding

and greater predictability, the data indicate that the IT-inversion

transformation with three transformations is easiest to decode at critical

point one, and the passive sentence with one transformation is easiest to

decode at critical point three. At the remaining critical points, the

reading difficulty or predictability is the same for all four transformed

sentences.

In the case of the THERE transformed sentence and the passive

transformed sentence, each of which has one transformation, it can be

concluded that except at critical point three, the two transformed

sentences were not significantly different. This conclusion appears to

support one aspect of both the deep structure and the chunking models.

Both models contend that sentences with the same number of transforma-

tions have the same complexity.

Question three: Does the type of transformation affect the degree

of predictability?

The data related to this question indicated that in four out of five

cases, there were no significant differences between the basic sentences

for the THERE transformation and the basic sentences for the passive

transformation. Also, in four out of five cases there were no

differences between the THERE transformed sentences and the passive

transformed sentences.

An examination of the basic sentences for these two transformations

showed that except at criLical point two, the two basic sentences were









not significantly different in the EVS measurements. This conclusion

would appear to support the statement by both the deep structure and the

chunking advocates that all basic sentences have the same degree of

difficulty. However, when the basic sentence for the WH-question

transformation and the basic sentence for the IT-inversion transformation

were examined, significant differences were found between them at critical

points one, four, and five. Significant differences were found between the

basic sentence for the THERE transformation and the basic sentence for the

WH-question transformation at critical points three and four. Significant

differences were found between the basic sentence for the THERE transfor-

mation and the basic sentence for the IT-inversion transformation at

critical points one, three, and five. At critical points three and four,

significant differences were found between the basic sentence for the

passive transformation and the basic sentence for the WH-question

transformation. At critical points three and five, significant

differences were found between the basic sentence for the passive

transformation and the basic sentence for the IT-inversion transfor-

mation.

The data indicate that only in the case of the basic sentence for

the THERE transformation and the basic sentence for the passive trans-

formation were the sentences alike in predictability or ease of decoding

in four out of the five measurements. Because both basic sentences were

written for easy transformation into a sentence with one transformation,

this may account for the fact that these two basic sentences were not

significantly different from each other but were significantly different

from the other basic sentences. The other basic sentences were written

for easy transformation into a transformed sentence with two or three

transformations.









In the case of the THERE transformed sentences and the passive

transformed sentences, it can be concluded that except at critical point

three, the two transformed sentences were not significantly different in

the EVS measurements. This conclusion appears to support both the deep

structure and the chunking models. The advocates of both models contend

that sentences with the same number of transformations have the same

complexity.

The data with regard to the THERE and passive transformed sentences

indicate that except at critical point three, the two transformed

sentences were not significantly different. Because both sentences have

one transformation, the findings suggest that the type of transformation

does not affect the sentence difficulty as long as the number of

transformations remains the same. This conclusion was not supported

by the data related to the basic sentences except with regard to the

basic sentences for both the THERE and passive transformations. This

conclusion also does not appear to support the findings of Fagan (1971)

that sentence difficulty was dependent on the difficulty of the trans-

formation rather than the number of transformations within a sentence.

Except for the passive transformation, each of the transformed

sentences involves a transformation in which the subject is moved from

the beginning of the sentence to a position following the verb or modal.

As the conclusions concerning the relationship between type of

transformation and EVS indicate, there is no discernible relationship

between any of the transformed sentences except in the case of the

THERE and passive transformations.

Question four: Does the position in the sentence where the EVS

measurement is taken affect the degree of predictability?








In the case of the basic sentences, the basic sentence for the

WH-question transformation showed a shorter EVS measurement at critical

point five of the sentence when compared with critical point one. The

indication is that the sentence becomes more difficult to decode toward

the end of the sentence. Just the opposite occurred with the basic

sentence for the TT-inversion transformation.

The data for the transformed sentences showed that there was no

significant difference between critical points one and five of the WH-

question transformation. The passive transformation was significantly

different between critical points one and four, but not between points

one and five. The THERE transformation had a significantly shorter EVS

at the beginning of the sentence than at the end, indicating that the

decoding process was more difficult at the end of the sentence. The

opposite was true of the IT-inversion transformation, which showed a

significantly higher EVS at the beginning of the sentence as opposed to

the measurement at the end of the sentence.

The conclusion indicated by the data is that the critical point

has a significant effect on the EVS measurement. It does make a

difference where the EVS measurement is taken. However, there is no

pattern. No general statements can be made about the relationship

between the critical point and the EVS measurement except to state that

the critical point is a significant variable.


Recommendations

This study compared the degree of predictability among specific

sentence types. The findings suggest that a complex relationship exists

between the three variables --types of sentence transformations, basic

versus transformations, and critical points. No one variable by itself









can explain the differences in the EVS measurements. In attempting to

explain the influence these variables have on the reading process, it

is necessary to determine which transformations are involved; it is

necessary to determine whether the sentence is in its basic form or is

in its transformed state; and it is necessary to determine where the

EVS measurement is taken.

The deep structure model and the chunking model offer explanations

which are too simple according to the findings of this study. Of both

models, the only part which appears to be partially supported by the

data is the proposal that sentences with the same number of transformations

have the same decree of difficulty. It is imperative that this statement

be accepted with reservation because the data on basic sentences do not

support this theory. All basic sentences are supposed to have the sane

degree of difficulty because they do not involve any transformations.

This was not found to be true except in the case of the basic sentence

for the THERE transformation and the basic sentence for the passive

transformation.

The deep structure model proposes that a one-to-one correspondence

exists between a grammatical model and a psychological model. In order

to add support to this theory, the data should have indicated that as

the number of transformations increased, the EVS decreased or the sentence

became harder to decode. This did not occur. At critical point five,

for example, the WI-question transformation had a significantly longer

EVS measurement than the basic sentence. At the other four critical

points, there were no significant differences between the basic sentence

with no transformations and the WH-question transformation with two

transformations.








The chunking model states that as a sentence becomes more complex

or has more transformations, it becomes easier to read. This theory was

not supported by the data. In fact, the basic sentence for the IT-

inversion transformation was significantly easier to decode at critical

point four than the IT-inverstion transformation. At critical points

two, three,and five, there were no significant differences between the

basic sentence with no transformations and the IT-inversion transformation,

which has three transformations.

The findings indicate that a question about Levin and Kaplan's

language processing model needs to be raised. The model states that the

EVS will always reflect the grammatical constraints in a sentence by

showing a longer EVS where the grammatical constraint is greater. In

this study, at the point where the passive sentence was more highly

constrained, according to Clark (1965), the EVS for the active sentence

was significantly longer. The EVS did reflect the differences in

constraint, but not in the direction suggested by the model. Levin

and Jones (1968) found the same results. Levin et al. (1972), working

with right and left embedded sentences, showed that the results they

found supported their model. The research indicates discrepancies which

will require further investigation.

Because no definite patterns could be discerned among the sentences,

the concern about whether a child should be presented with a transformed

sentence or a nontransformed sentence may not be the important issue.

It appears that too many factors are involved to answer that question

with a yes or no. The data indicated that some sentences 'L th a number

of transformations were easier to read than the basic sentence while just

the opposite was true of other sentences with a number of transformations.









On the basis of this study, it appears that children need to be

exposed to as many sentence types as possible to allow the students to

develop versatility in reading. The theory that children should not

be exposed to transformed sentences until they have mastered the basic

sentences is not supported by this study.

Further research needs to be conducted on different types of

transformed sentences to determine how other variables such as the

knowledge of linguistic structure, the length of clauses, and the

inclusion of specific types of words in a sentence (conjunctions,

pronouns, prepositions, verbals) affect language processing. In

addition, research needs to be conducted to examine the relationship

between sentences with the same number of transformations. In this

study, two sentences with one transformation each showed no significant

differences in four out of the five measures. If additional research

supports this finding, equating sentences with the same number of

transformations may be possible.

The findings which indicated significant differences among the

basic sentences remain open to further investigation. All basic

sentences are defined as being equal in difficulty because basic

sentences contain the same number of transformations. The data did

not show this to be true. Only the two basic sentences which were

written for easy transformation into a sentence with one transformation

were found to have no significant differences in four of the five

measures.

Additional research needs to be conducted to investigate the

question of whether basic sentences can be considered equal in difficulty.

The evidence here is that there is a difference between the basic sentences








used in the study except the basic sentences for the THERE and passive

transformations, which were written for easy transformation into

sentences with one transformation.

The EVS measurement is a measurement of the decoding process. The

EVS offers an opportunity to investigate the way readers process sentences.

Because no measurements on comprehension were taken in this study, addi-

tional research needs to be conducted which includes an evaluation of EVS

and its relationship to comprehension. Following the EVS measurements,

comprehension questions could be given to determine the interrelationship

among length of EVS, comprehension, and sentence type. In conducting

such a study, the influence of other variables would have to be controlled.

For example, if comprehension measures were taken, the target sentences

would probably need to be placed in complete paragraphs rather than in

paragraphs with unrelated sentences. The unrelated sentences in this

study were used to control for context clues. It would be difficult to

ask questions which adequately measured comprehension when the measurement

was taken on a series of unrelated sentences.

Additional areas of study could involve a study of the relationship

among EVS, sentence type, and the less able reader. Focusing on

Chomsky's conclusions (1972) that there is a "developmental sequence of

linguistic stages through which all children pass" (p. 2), studies could

be conducted with less able readers to determine which sentence types are

easier to decode (have longer EVS measurements). A wide range of

sentence types could be examined including those involving sentences

with inverted word order.

The major conclusion from this study is that all the models offer

explanations of language processing which are too basic. The chunk model

and the deep structure model focus on the number of transformations in a








sentence. Instead of trying to determine whether a transformed sentence

is easier or harder to read, the question should be how many factors in

addition to transformations affect language processing. Levin and

Kaplan's model attempts to examine other variables in addition to the

number of transformations in a sentence. The position in the sentence

where a measurement is taken is important in the Levin and Kaplan model.

The data from this study indicated that the position in the sentence

where the EVS measurement was taken had a significant effect on the

results. There is a question about the Levin and Kaplan model as a

result of this study. The relationship between the EVS and constraint

analysis was not supported.

Research which examines the relative merits of the language pro-

cessing models need to be continued. The results from this study

indicate that language processing is much more complicated than the

explanations offered by the major models.












































APPENDIX A

SENTENCE TYPES
















APPENDIX A


Sentence Types

In order to control for vocabulary, words used in the sentences were

limited to the Functional Reading Word List for Adults (Mitzel, 1966).

Each basic sentence and its transformed sentence type contained the same

number of words to control for the number of words before and after each

critical point. Two four-word phrases followed the last critical point

in each sentence.


Basic

1 2 3 4 5
1. Supervisors/were/listening/to the decision/of their employees/

about a strike vote for more sick days.

2. Firemen were running toward the area of the fire near the front

door of the new hospital.

3. Patrolmen were waiting at the scene amount the motorists near

the old road by the huge stadium.

4. Officials were standing on the field near the player with a

broken arm from the first game.

5. Citizens were meeting at the school with county politicians

about the financial responsibilities of the area taxpayers.


THERE Transformations

1 2 3 4 5
1. Therc/were/carpenters/on the project/with the schedule/for

the entire insulation of the old building.








2. There were men on the docks near the machinery for the first

time since the latest injury.

3. There were homeowners in the vicinity with the juveniles from

the recent fight at the neighborhood pool.

4. There were witnesses near the accident from the collision of

the two trailer trucks on the nearby bridge.

5. There were experts at the meeting for the public about the

real dangers of heavy cigarette smoking.


Active

1 2 3 4 5
1. The/crude/lawyer/was being offensive/at the trial/of the senior

bookkeeper of the export company.

2. The reckless cop was being impulsive about the message near the

murder scene from the jealous wife.

3. The wise minister was being polite to the assembly of very

angry people from the reform group.

4. The security guard was being careless about the locks on the

glass doors of the main building.

5. The new superintendent was being flexible in his approach to

the proposed changes in the school budget.


Passive

1 2 3 4 5
1. The/old/dog/was being taken/by his owner/to the dedicated

veterinarian in the old neighborhood.

2. The faithful physician was being honored by his patients in

the banqut room of the Oriental restaurant.

3. The helpless animal was being cornered by the dogs in an open

field near the winlor cabin.








4. The young manager was being introduced to the personnel from

the retail division of the drug company.

5. The last survivor was being examined by hospital attendants

in a temporary tent near the crash site.


Basic

1 2 3 4 5
1. Passengers/will/soon/have their tickets/for the flight/to the

remote island off the southern coast.

2. Banks will probably have severe problems in the future with

soaring interest rates for new housing loans.

3. Veterans will usually receive their checks for partial

disability within a few days after the first day of the month.

4. Applicants will probably find some installments for medical

insurance below the proposed rates of the new company.

5. Pedestrians will generally find underground tunnels at

convenient locations in many urban areas with heavy traffic

patterns.


WH-Question Transformation

1 2 3 4 5
1. When/will/newsmen/know the truth/about the proceedings/against

the five men from the new company?

2. 'When will firemen have the equipment for their orientation to

the new system of controlled fire fighting?

3. When will psychologists find an answer for the cause of severe

emotional problems in very young children?

4. When will consumers have sufficient knowledge about the products

on most grocery shelves throughout the entire country?








5. When will people have some protection against the powers of

big criminal leaders with important political friends?


Basic

12 3 4 5
1. Climbing/is/dangerous/on mountain trails/for most experts/

during severe winter storms with below freezing temperatures.

2. Exercise is difficult in the beginning for many individuals

with major health problems from poor eating habits.

3. Flying is unpleasant during rough weather with high winds for

the few passengers with very weak stomachs.

4. Conservation is important for the survival of life forms on

this huge planet during the future decades.

5. Investing is necessary in all areas of our economy for the

continued growth of our industrial society.


IT-Inversion

1 2 3 4 5
1. It/is/wise/to deposit money/in the bank/to insure the safety

of your emergency funds.

2. It is necessary to vaccinate people for various diseases to

prevent the spread of many dangerous germs.

3. It is practical to discuss problems over the telephone to

eliminate the possibility of any fist fights.

4. It is profitable to outline chapters in a book to organize

your study of unfamiliar subject matter.

5. It is important to examine lakes in remote areas to identify

the needs of the different fish.














































APPENDIX B

PARAGRAPHS CONTAINING TARGET SENTENCES
AND UNRELATED SENTENCES
















APPENDIX B


The target sentences were embedded in paragraphs of unrelated

sentences to control for context clues. If the sentences were related,

the subject might be able to guess words based on information given in

the sentences preceding the target sentence. To control for vocabulary

and readability, the unrelated sentences used to compose the paragraphs

were limited to sentences taken from General Reading for Understanding

Cards 21-60 (Thurston, 1969). The target sentences were distributed

among sentence positions two, three, or four in the paragraphs because

a study by Levin and Kaplan (1968) indicated that subjects scanned the

first line before beginning to read aloud.

The underlined word is the critical point where the EVS measurement

was taken. To control for guessing, the position of the critical point

was varied among the first, second, third or fourth word of a line.

There were at least nine words past the critical point of each target

sentence on the same line to control for position effect.


The worth of the product is basically dependent upon the quality

of the materials from which it is made. Quakers believe that

fighting is wrong, and therefore many of them refuse to become

soldiers. The audience cannot see the small silken strings with

which the operator makes the marionette walk or dance about the

stage. Patrolmen were waiting at the scene among the motorists

near the old road by the huge stadium.








A felony is usually a crime that is punishable by death or by

imprisonment in a state prison. Instead, he seeks ways to make

his high standard of performance a goal for others to attempt to

surpass. Banks will probably have severe problems in the future

with soaring interest rates for new housing loans. To obtain

successful results in fine printing, the right quality of paper

is essential.


An examination of any beaver logging operation shows many felled

trees that have fallen away from the water. Because of the lack

of men and materials, very few roads were constructed during the

war. The old dog was being taken by his owner to the dedicated

veterinarian in the old neighborhood. The Romans built a vast

network of roads that connected their city to the farthermost

corners of their conquered possessions.


The advertisement had spoken enthusiastically of a handsome

residence set on a beautifully landscaped acre. As he came

closer still, he could see how she shrank and shivered when

the sea waves sprinkled her with cold salt spray. They make

dollars available to foreign countries which in turn can buy

more from the United States. The last survivor was being

examined by hospital attendants in a temporary tent near the

crash site.


Charles Goodyear tried to improve this material so that it would

not be affected by temperature. This water contains dissolved

minerals that harden and remain on the floor of the cave.

Conservation is important for the survival of life forms on









this huge planet during the future decades. Writers of fiction

have portrayed the gladiators who fought for their lives in the

Roman arenas as romantic and heroic figures.


The rain pounded the windows of the cockpit so hard that the

pilot could not see out. He was really quite an unusual dragon,

a very lazy dragon who disliked exercise of all sorts. There

were experts at the meeting for the public about the real

dangers of heavy cigarette smoking. I couldn't sleep after

I went to bed, for every shadow and every sound assumed a

menacing form.


Equipment for outdoor living must withstand sunshine, rain, and

snow. If only a couple of eggs are removed, the mother bird

ignores her loss and hatches the remaining eggs. The wise

minister was being polite to the assembly of very angry people

from the reform group. The dawn of civilization occurred at an

earlier time in Northern Africa than it did in Europe.


When she was grown, she was still so beautiful that even then

no one called her anything else. Most of the advances made by

mankind can be attributed to the fact that man's errors can be

corrected. When will people have some protection against the

powers of big criminal leaders with important political friends?

The bags that are used by cotton pickers are subject to a great

deal of wear.


It is seldom that one finds a leader in science who is also a

leader in politics or in philosophy. Supervisors were listening

to the decision of their employees about a strike vote for more








sick days. Our bodies are strengthened not by what we eat but

by the food that we digest. Dick wished very much that he were

his cousin Bob, who had just won the state oratorical contest.


Collectors of stamps, coins, or other objects should collect

common specimens as well as unusual ones. Certain conditions

of moisture, light, temperature, and so on, are required by

every kind of plant or animal for life. It is necessary to

vaccinate people for various diseases to prevent the spread

of many dangerous germs. When he witnessed the total destruction

of his home, he was glad that, at least, the building was insured.


They frequently die as a result of other diseases, particularly

those that affect the respiratory tract. To this poor family,

not the least of the doctor's wonderful qualities was his habit

of forgetting to send a bill. The prospector searches for gold,

and the scientist searches for a better understanding of nature.

It is wise to deposit money in the bank to insure the safety of

your emergency funds.


As soon as young seals have acquired a little strength, their

mothers leave them periodically. Thus when the hens are not

given sufficient drinking water their production of eggs is

decreased. Exercise is difficult in the beginning for many

individuals with major health problems from poor eating habits.

The pitcher plant is so named because the margins of its leaves

fold together to form a receptacle that retains rainwater.









Industries should be sure that any equipment that may come in

contact with electric current is properly grounded. Joe hated

dancing school, but each week his mother sent him anyhow. The

biggest obstruction in the path of progress is not lack of

money but lack of men. There were homeowners in the vicinity

with the juveniles from the recent fight at the neighborhood

pool.


He wanted a son, to be his companion and eventually to rule

after him. This is one reason, perhaps, that during the Civil

War the Confederacy had such excellent leadership. It is

practical to discuss problems over the telephone to eliminate

the possibility of any fist fights. At the end of the story,

we find our hero rich, healthy, and beloved.


During the harvest season the men on all the farms were working

from morning until sunset. When hostile Indians roamed through

the land, every white settlement had a fort for the protection

of the inhabitants. Citizens were meeting at the school with

county politicians about the financial responsibilities of the

area taxpayers. In the mornings they fly out to the country to

forage in grainfields and barnyards.


What we are thinking of as past is the major transition from

hand tools to machines. When will psychologists find an answer

for the cause of severe emotional problems in very young

children? When they are well fed, they become fat and produce

a larger quantity of meat. The wider lanes become faster than

the narrower lanes when the traffic becomes heavier.









The end had ceased to charm, and how could there ever again

be any interest in the means? Robert's record of faithful

service made all of us feel that his employer's harsh

treatment was hardly justified. I never read poetry just

because my friends have enjoyed it, even if I consider my

friends to be discriminating in taste. Veterans will

usually receive their checks for partial disability within

a few days after the first day of the month.


The huge bridge that spans San Francisco Bay has the same

purpose and rating as a state highway. Applicants will

probably find some installments for medical insurance below

the proposed rates of the new company. Man brings trouble

upon himself when he disturbs the balance that nature has

established. The damage appears to be caused not by

freezing of the cell but by the refrigeration.


Even though he had been elected class president, he still

longed to be a sports hero. People are now said to spend

much more time watching television than they spend listening

to the radio. When will firemen have the equipment for their

orientation to the new system of controlled fire fighting?

Soon after we began building the cabin, our group functioned

like a well-ordered machine.


It is often possible to control an epidemic when the cause

of a disease and the way in which it is spread are know.

It is profitable to out line chapters in a book to organize

your study of unfamiliar subject matter. Over a span of









centuries any language changes so much as to be unintelligible

to any but a few scholars. People in the United States have

wants that are more varied than those of native people in

parts of Asia and Africa.


Some of those who farm on such a plain do not realize that

their endeavors necessarily involve a risk. He thought of

many reasons for their unkindness and blamed them for his

misery. The great steel cities of the United States are not

only workshops of the present but also laboratories of the

future. The faithful physician was being honored by his

patients in the banquet room of the Oriental restaurant.


Evaporation is the chief process by which the atmosphere is

supplied with moisture. When will consumers have sufficient

knowledge about the products on most grocery shelves throughout

the entire country? It is commonly said that robbing a bird's

nest of some eggs deprives the world of just so many birds.

They think that a newspaper will accept any article, even

though it is carelessly written.


You can sow like an expert and have almost anything you want

for the cost of materials. There were men on the docks near

the machinery for the first time since the latest injury. He

called the natives of the island Indians, because he thought

that he had reached the East Indies. In some places a deep

layer of relatively loose dirt covers the underlying bedrock.


We need candles, so we first took some fat and melted it into

tallow. The young manager was being introduced to the








personnel from the retail division of the drug company. It

is launched by means of an elastic cord, a spring catapult,

or a tow cable. The society hopes that facilities for

having every schoolchild's hearing tested regularly can be

made available.


In some new skyscrapers, steel and glass walls an inch thick

are replacing the old-style foot thick walls. Passengers

will soon have their tickets for the flight to the remote

island off the southern coast. On any other morning the

horse would have been standing patiently at the pasture gate,

waiting to be bridled. The crews of the world-circling jet

planes saw three sunrises, but they were gone less than forty-

six hours.


A fish must be hooked at that moment when it bites. The new

superintendent was being flexible in his approach to the

proposed changes in the school budget. They choose for the

location of their homes that part of a sandy bank which has

the greatest degree of exposure. An epidemic is an outbreak

of a disease that strikes many persons at about the same time.


The earth is part of a galaxy of stars that we call the Milky

Way. It is important to examine lakes in remote areas to

identify the needs of the different fish. So that the beds

might continue to yield pearls, the divers were cautioned

against leaving the oyster to die after robbing its shell of

the pearl. Incidents of this type are occurring everywhere,

every day.




University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs