[1111 EH. @5111". Jl‘ M . H . Ml. ,
l 1Hl1 ml11 llwlmnhflllJW11 11111 l1111111lfi l
”lll l “ll lllfil ‘ll glll11lll111 l .. LJJJJJllflJ.u....JJ.glllllll‘J
. lg... " 1 1
HMWWW

”11 ‘11l1llmM'Jl11llIJle JJ‘" W M ‘l'll'llg‘ J... Jii ""Ju'Al'tHH..Jml].Jl JJlJlHJ'J‘JJlJJl""JJ 1H MW. .JJJJ J‘JHJJJ ‘Jf
1 ‘1 l1111lllll111l11'111l11l 111 '1‘ 111ll1111'11 11111131 1111111 11 I.” ””1111111ll1 11 l1‘ll'£‘:ll1lJ .JJTIJE.J.'J1‘1J.?1’1
l 116ll1111111'll1‘l [p.1‘I11‘1l1l1‘ ll'lhly 1111l11l1|l 11H [3 ll: l! “:1“ 31 H1 ’111111l‘sgl311 " "~‘.' "I '1:.‘
11 ‘1 llll'lJ1 lll11111|1ll llll. 11ml l1' 11l1l'1ll11'11'11 11111l 111 11111llJI‘JHl llJJ 1'1!ng .l” 'lJlJJJlll'I' 1l1l‘ l1” 1

:l’ l , ml Ji .JJl H.Il Ml ”11i1111' m 1.11111 r J .2
2 J lkW J

 

J
Mlmll ’11 1JJI“ llllfll £11111“ ‘l"lll1‘1llllJl'll 'J‘ll‘”

        

lfil131fi “l1qug filifln M.“
llg‘ll 1" H i 1'11llhld
11'1” 11'11‘16111

J $111111 Jl lll‘l 13% l]

 

 

          
    
  
   
 

 

        
  
  
  

 

    

 

W

3' “11m lJJJll llll‘l 1
lllJ ll lJ1111n 'J. JR 1ll111l1111111ll111111 ‘1‘1‘1‘ll1’11: 11.21.11“. 1111111“ l11l1l11l1111 -.

'13J1l‘l3lJll, JJ’JlJ lllll1JJ11111JJJ‘Elml-li‘JJJlJJJJl'llllll’JJ1‘1 J 51.1”1! “ll“ .
‘=‘JJ‘JJJ‘lJJ‘J‘JiJlJ‘J‘ 3‘1‘2’JJJ‘l=‘J‘JJlllJ‘JJJJ‘J‘JJJJ‘J‘JJJJ‘J J.JJ:‘JJ.‘..J
J lJJJJ’lfl11llWJJ.” ' ll" BJJ . ~Jxfifiihll“JJ‘JuJ'[ J , lJJlxll”“‘J“‘“JJ”J*”“lJt
-‘J .‘J.':5J.llll ,J l5} “I llll lllllJlll ls“ 1"1113'f'lfll1‘illl' 1? .ll h .llzill‘i ll‘lfil (l‘ll1ll1‘JH1J11‘115W’f! l l l
t'lllllFJlJJ"PlullpfllthfillllufillndJJJJlJlJlJJ J2 J‘JJJJJJ JJ J..l ‘ ‘ {iii J
‘J 1m. . ' 1 ‘S: 1,! "1‘5l 1ll1111 11.l1l 11l1ll11slkllll1 J 1-l1lll1all§1ll 411%.. Ell1Jl0l 11""1‘l1111 1.91
:mfifiF&WWF%lfllel “ll“fllfir wdlflMflwwmlfiJfimlflwfiflfi JJJWJJ .;
J l; J l J‘lJ l lll l ll l i } lwlzllllmngl11 lill? lliii lzllJl’Il.!$?};l.ll 1L1l1ll‘llli‘l'1l JzJI;J' .ll'JJ;J.Jfl
11 1111l1lllllul“! l1l1ll“l 111“ HEW" lJJ lllp‘l‘ll‘l 1‘1‘1‘55’1311;'illlltJalJJJJJJ‘ltllfl’li-Jf .JJJJ’J‘JVUJ.
J quJ ,,. H J n' “’ la. ”lJ ?Jl‘$ .JiflquJJJ-JJ‘u-J -§~":J.. JJJ
J'Jl1bvmh k1; l l Wl gl“lU $lflmflkdflkaylbggle5fiyw
. J J’JlJl‘JJ-JJJJJ‘JJ‘JJ.‘ ‘J‘ J l. ., JJJJ .. J3...l..l1i.‘;ll.i.JJ’J,Jll.JJ1JJ‘1JJJ‘.li'J§1JJ¥Z‘JlJiééfilléllltfll‘g‘g?J5 . J; J
'"’zJJllliliélilil‘géllQJllgl{f J. l.m J-J..J..JJ‘.JJ\JJJ§J‘.JJ Jllll. J; l... .llJl‘ltJ . . '..2J:J.JJJJJJJJ
JEJJJE‘J’JJJJJJ JJJJJJ J J3... ' ’li‘JJJJ’J‘Jll‘l‘JiJJ‘ “Jill ‘ 1““l'1'1J‘J‘JJ?‘JJ‘J‘J‘“-“15‘J’J'Ji
-' -' ‘l‘llllllll‘l’lll‘l 1l1 lllgtélllll‘.JJJJJJ§J£§l§.J.l%,..J:".:;l.‘...2JilllJJiJJ“lJ‘J‘J lJJ'Jj
,. J J. J ,, 'J Jun. ifi." 19:43”. JTa‘J‘Jitz". .. .., 1.":
J ll .‘J‘JJllJ‘JlJllellJll JJJ ’l“‘Jill;JlllllfilJJ‘JJlJJJi‘J"‘Jll‘JJJ‘JJJ‘E‘JJ‘JJJJ‘:JJ‘J.‘J ;. J JJ‘J;
. '.2~l.;.lll}ll§ llllll.l.15§ii’lll ‘- . gilllll:,tsflll1tvlllllllllllllJll‘l 'Jlliiil‘lJ}§l.J..Jl§§Jl'l...J ,4 ‘gl 1
. , r. - :,.J....J;..........JJJ.mg. WJJJJ. J‘sJ‘JJlJJJgJ’JJJJJJJJJ.J..J.J..J.‘.J§J.J‘ J... J :JJJJ
' w l. J (ll! ' l J. JJ JJJ J
JJJJllTllJlJl‘JJl JJJ’? ‘ ‘ JJ ll? i‘Jl‘l‘J’JJ‘J'JJll J"l§l‘l1lJ15J1J111f???‘ . 'JJ’J‘JJJJ
.= ‘JJJJ'JlilllJllJJW J. c 1...... J..JJJ J... JJJJJJJJ J. :J
fiJJ'JJthJ'JJ‘ E J.‘ " ,Fég1.'1'1,ili 5.3.5? :l‘ In} ,JJJl “ 1‘ M1? 113
l... .JJJJJ. . - .JJJ,lJ.'.‘.JJJ 'J.l1..§? ‘ lJJJJ.J J... 2...... ... ..:.JJ
l%l%¥ l”W$%fiM#fifilflWfilWW¥
I JJJJJllJlJJJ‘ JJ‘JlJlJ *;-~ .‘JJ’JJ J ‘ J... J..J..,JJ Jill

" 3......“ 13115.“. tug.

 

xltJJlJ § Jr “3 , . 'h‘g'lz‘l 'lllzi..1'1:.-Jiftill.s:1=‘i'
llllllllll .J J..lJ.l.l‘l‘JJ..‘.f.‘“J.
illfiflfi‘l'1 1ft , :;L&l%. { ”1% “fly
JlJJJJlJJ JJ. ‘13‘JJJJJ ,...J..

hm hm JlJI1 m.

‘ 111111111111111111111111111

31293107731121

 

LIBRARY 1
Michigan State'
University

 

 

 

This is to certify that the

dissertation entitled

THE CONTRIBUTION OF SCHEMATA TO THE READING
COMPREHENSION OF EAST ASIAN READERS
OF ENGLISH AS A SECOND LANGUAGE

presented by

Susan Kathleen Kitao

has been accepted towards fulfillment
of the requirements for

PhD degree in Commum’ cati on

/ Major profcgg

 

 

Date November 18, 1987

 

MSU i: an Affirmative Action/Equal Opportunity Institution 0-12771

 

 

 

MSU RETURNING MATERIALS:
Place in book drop to
LIBRARIES remove this checkout from
—_ your record. FINES will
- ~ be charged if book is
returned after the date
stamped beiow.

 

 

gm'04.fiai
WM

300 0264

m 1 319’}-
1(38'-u

“‘1
"‘ .} O o
“J ' 7.,
1}: F
4

 

 

THE CONTRIBUTION OF SCHEMATA TO THE READING
COMPREHENSION OF EAST ASIAN READERS
OF ENGLISH AS A SECOND LANGUAGE

By

Susan Kathleen Kitao

A DISSERTATION
Submitted to
Michigan State University

in partial fulfillment of the requirements
for the degree 0?

DOCTOR OF PHILOSOPHY

Department of Communication

1987

ABSTRACT
THE CONTRIBUTION OF SCHEHATA TO THE READING
COMPREHENSION OF EAST ASIAN READERS
OF ENGLISH AS A SECOND LANGUAGE
By
Susan Kathleen Kitao

In this paper, I discussed schema theory. especially
as it relates to reading, and then discussed the
application of schema theory to second language reading.
I also discussed the role of lexical knowledge and issues
related to research on second language readers.

In order to better understand of the process of
second language reading comprehension as related to prior
knowledge and lexical proficiency. I did a study using
ninety-six nonnative speakers of English from East Asia.
I measured their prior knowledge on three topics and their
comprehension of three reading passages related to those
topics. I also measured their lexical and reading
proficiency.

I proposed two hypotheses. The first stated that
there would be a correlation between prior knowledge and
comprehension. Significant correlations were found
between prior knowledge and comprehension for the two
passages that required greater prior knowledge. Uith a
median split for reading proficiency, more correlations

were significant for readers with low proficiency than

readers with high proficiency. These results indicate

that second language readers use prior knowledge.

The second hypothesis stated that readers with high
background knowledge would have a lower correlation
between reading comprehension and reading proficiency than
readers with low background knowledge. Two of the nine
pairs of correlations were significantly different, though
differences for five others were in the predicted
direction. while the hypothesis was not strongly
confirmed, there was some support for it.

In addition, I addressed two research questions. The
first was about the relationships among prior knowledge,
reading comprehension, and lexical proficiency.
Correlations were found between lexical proficiency and
reading comprehension. Participants with higher prior
knowledge had more significant correlations. This appears
to indicate that readers depend on vocabulary, together
with prior knowledge, to achieve comprehension.

The second research question asked about differences
among reading passages that required different amounts of
background knowledge for comprehension. Both background
knowledge and lexical proficiency were more important for
passages that require a degree of background knowledge.

I concluded with a discussion of implications for the
second language classroom and recommendations for future

research.

ACKNOULEDGMENTS

I would like to thank my dissertation director, Dr.
James Stiff, the chairperson of my committee. Dr. Jack
Bain, and the members of my committee, Dr. Mary Bresnahan
and Dr. Paul Munsell. for their helpful advice,
assistance, support. and direction. I would also like to
express appreciation to my research assistants. Mike
Chiado, Mike Israel, Angenette Lynch, Susan Matthews,
Phillip Peters. Tom Slater, and Lorenzo Uhite for all
their help and especially Angenette and Phil for listening
sympathetically to my problems. I would also like to
thank a number of other people, including Chi-min Hu, Dawn
Eskew, Larry Shu. Dr. James Stalker and the graduate
assistants of the English Language Center, Mary Rhodes of
the Haslett adult basic education program, and Kimiko
Akita. In particular, I would like to thank the
participants in this study, who took time out of their
busy schedules to help me. Last, I would like to thank my
husband Kenji. without whose support and encouragement
this would not have been possible, and my parents, who

always believed that I had it in me.

iii

TABLE OF CONTENTS

LiSt 0? Tables 0 o o o o o o o 0
Chapter I--Review of Literature

Introduction . . . . . . .
Schema Theory . . . . . . .

The Application of Schema Theory to Reading
Content Schema and Second Language Readers

Measurement Issues . . . .
Methodological Standards .
Conclusions . . . . . . . .

Chapter II‘-neth0d5 o o o o o 0

Overview . . .
Instrument . .
Participants .
Procedures . .
Measurement . .
Analyses . .

O O O O O O
0 O O O O O
O O O O O O
O O O O O O
O O O O O O
O O O O O 0

Chapter III--Results . . . . . .

Initial Analyses .

Consistency Checks .
Pearson Correlations
Hypotheses . . . . .
Research Questions .
Conclusion . . . . .

O O O O O O
O O O O O O
O O O O O 0

Chapter IV—-Discussion . . . . .

Introduction . . . . . . .
Findings of the Study . . .
Implications for the Second
Limitations of the Study .

O 0 O O O O O O 0 O
O O O O O O O O O O
O O O O O O O O

O O O O O O O O

O O O O O O O 0

O O O O O O O
O O O O O O O

O O O O O O O 0

O O O O O O
O O O O O 0
O O O O O O
O O O O O O
O O O O O O
0 O O O O O
O O O O O O
O O O O O O
O O O O O O

O O O O O O
O O O O O O
O O O O O O
O O O O O O
O O O O O O
O O O O O O
O O O O O O
O O O O O O
O O O O O O
O O O O O 0

Language Classroom

0 O O O O O O O O 0

Suggestions for Future Research . . . . . . . .

Conclusion . . . . . . . .
Appendix 0 O O O 0 O 0 O O O O 0

List of References . . . . . . .

iv

U! h.
A) (fl

0-
1:

LIST OF TABLES

Table 1
Order Effects

Table 2

Cell means with number of idea units recalled as D.V.
Table 3 ‘

Gender Effects

Table 4
Effects of Nationality

Table 5
Pearson correlations among variables (Gone with
the Uind passage)

Table 6
Pearson correlations among variables (wedding
passage)

Table 7
Pearson correlations among variables (baseball
passage)

Table 8
Third order partial correlations between prior
knowledge and measures of comprehension (Gone with
the Uind passage)

Table 9
Third order partial correlations between prior
knowledge and measures of comprehension (wedding
passage)

Table 10
Third order partial correlations between prior
knowledge and measures of comprehension (baseball
passage)

Table 11
Third order partial correlations between prior
knowledge and measures of comprehension. with median
split according to prior knowledge (Gone with the
wing passage)

Table 12
Third order partial correlations between prior
knowledge and measures of comprehension. with median
split according to prior knowledge (wedding passage)

Table 13
Third order partial correlations between prior
knowledge and measures of comprehension, with median
split according to prior knowledge (baseball passage)

Table 14
Third order partial correlations between proficiency
and measures of comprehension (Gone with the Uind
passage)

Table 15
Third order partial correlations between proficiency
and measures of comprehension (wedding passage)

Table 16
Third order partial correlations between proficiency
and measures of comprehension (baseball passage)

Table 17
Third order partial correlations between proficiency
and measures of comprehension, with median split
according to prior knowledge (Gone with the Uind
passage)

Table 18
Third order partial correlations between proficiency
and measures of comprehension, with median split
according to prior knowledge (wedding passage)

Table 19
Third order partial correlations between proficiency
and measures of comprehension, with median split
according to prior knowledge (baseball passage)

Table 20
Third order partial correlations between lexical
proficiency and measures of comprehension (Gone with

the wing passage)

vi

Table 21
Third order partial correlations between lexical
proficiency and measures of comprehension (wedding
passage)

Table 22
Third order partial correlations between lexical
proficiency and measures of comprehension (baseball
passage)

Table 23
Third order partial correlations between lexical
proficiency and measures of comprehension, with
median split according to prior knowledge (Gone with
the Uind passage)

Table 24
Third order partial correlations between lexical
proficiency and measures of comprehension. with
median split according to prior knowledge (wedding
passage)

Table 25
Third order partial correlations between lexical
proficiency and measures of comprehension. with
median split according to prior knowledge (baseball
passage)

vii

THE CONTRIBUTION OF SCHEMATA TO THE READING
COMPREHENSION OF EAST ASIAN READERS
OF ENGLISH AS A SECOND LANGUAGE
Chapter I
REVIEU OF LITERATURE
Introduction
In looking at comprehension, whether reading or
listening comprehension. the traditional assumption is
that these are skills in which the reader/listener
passively ”takes in" what the writer/speaker has produced.
However, over the past fifteen to twenty years, it has
come to be recognized that what are traditionally referred
to as passive skills do require the active participation
of the comprehender. The reader/listener’s background
knowledge interacts with input for the reader/listener to
arrive at comprehension. Starting with such theorists as
Goodman (1967) and Smith (1971), the movement has been to
include the reader in textual interpretations. This
movement has not been restricted to psycholinguistics but
is also reflected in counterparts in philosophy,
psychology, communication and literary criticism. Sghgmg
theory has been developed to help explain how
readers/listeners interact with the text for
comprehension.

In this chapter, I will look at schema theory. in

particular content schemata. I will discuss how these
affect reading, and how they affect the reading of
nonnative speakers of English in particular. I will
review in detail studies that have been done on nonnative
speakers and will discuss schemata and reading in second
language learners. I will make recommendations on how
what is known about schemata can be applied in the second

language classroom.

Schema Theory

Background

As far back as Kant (1781), it has been asserted that
background knowledge plays a part in comprehension.
Bartlett (1932) developed schema theory to explain how
background knowledge is used by a reader/listener to
understand and recall a text. Bartlett found that when
participants read a story from an unfamiliar culture,
their memory of the story changed over time to fit
schemata from their own culture. Yet it was not until
four decades later that there was wide interest in
Bartlett’s theories.

Literary criticism is one field in which there has
been interest in the contribution that the reader makes to
comprehension. The school called reader response

criticism in particular has emphasized this aspect of

reading. Jackson (1947) dealt extensively with the
concept of the reader as ”artist", and discussed reading
in terms of the contribution that the reader makes to the
creative process. Iser (1978: 24), an influential thinker
in reader response criticism. wrote:

Although it is clear that acts of comprehension are

guided by the structure of the text. the latter can

never exercise complete control....(The reader)
participates in both the production and comprehension
of the work’s intention.

In the mid-1970’s. there was a revival of interest
among psycholinguists and scholars in other areas in
Bartlett’s research and theories. Researchers again began
looking at the influence of background knowledge and the
organization of texts on comprehension, particularly
reading comprehension.

Definition of Schema

Schemata are highly organized, generic knowledge
structures composed of ”slots” or ”placeholders” for each
component (den Uyl and van Oostendorp, 1980; Anderson,
Reynolds, Schallert. and Goetz, 1977). Two broad types of
schemata are content schemata and textual schemata.
Content schemata contain general or specific information
on a given topic. Textual schemata contain information

about how rhetoric is or ought to be organized.

In addition to slots. a schema includes information

about constraints on what can normally fill a particular
slot (Anderson, 1978) and what the relationships are among
the slots (Graesser, Uoll, Kowalski, and Smith, 1980). A
schema includes the network of associations that a concept
has (Pearson and Spiro, 1982). This process of filling
the slots is called instantiation of a schema (Anderson,
1978). In addition to information about constraints on
what can fill a slot, schemata include ”default values”
for each slot. Therefore, if no information is available
to fill a slot, the reader/listener fills the slot with a
value that he/she knows to be typical in that slot.
Schemata are hierarchically organized, with most important
information at the top, down to the least important
information (Anderson, 1978).
Event Schemata

One specialized type of content schema is the gxgfli
schema. also referred to by Shrank and Abelson (1977) as a
script. Like other schemata, event schemata are knowledge
structures composed of slots labeled according to what may
or must fill the slots. Event schemata contain
information about stereotypical events or situations. such
as eating in a restaurant or visiting the office of a
medical professional. The unique characteristic of event

schemata is that they are organized temporally (Schank and

Abelson. 1977; Abelson, 1981). Event schemata specify the
expected time ordering of events. I will use event
schemata to explain how schemata are believed to function.

An event schema (script) for going to a restaurant
would include information about servers, menus, paying the
bill, and so on (Bower. Black, and Turner. 1979). The
event schema would include the information that there must
be a server; that the server can be male or female: that
there must be a menu, whether it is a printed menu given
to customers or a sign posted on the wall; that the menu
will most likely have prices. In addition to a basic
restaurant schema, there are restaurant schemata for
specific types of restaurants, for example, fancy
restaurants. cafeterias. and fast food establishments.
These specific event schemata might include the
information that in a fancy restaurant, the server is
likely to be male and that the prices might be omitted. on
at least some of the menus.
Uses of Sghemata

Schemata are used mainly in two ways (Bower, Black,
and Turner, 1979). The first is in guiding actions in
typical situations. For example. a customer entering a
restaurant knows how to behave. based on event schemata.

The customer knows that, for example, when eating at a

sit-down restaurant. the server will bring a menu and then
leave, that the customer should look at the menu and
decide what to eat, and that the server will be back to
take the order.

For the purpose of this paper, I am interested in the
second use of schemata--comprehension. Schemata allow the
reader/listener to make inferences and fill in information
not explicitly included in the text (Anderson, 1978).

This makes it unnecessary for a writer/speaker to include
every detail that the reader/listener needs to know. The
reader/listener makes inferences based on the information
that is given and on information from the schema, whether
from default values or from relationships that are
specified among slots. For example, when eating at a
restaurant is mentioned in a text. the reader/listener
uses schemata to fill in details that are not specifically
mentioned and to make inferences based on the information
that is given. If a writer/speaker mentions that there
are no prices on the menu. the reader might infer that the
restaurant is a very fancy. expensive one.

In summary, schemata are highly organized, generic
'knowledge structures composed of "slots" or "placeholders”
for each component, including information about

constraints on what can fill the slot and the

relationships among the slots. Schemata are
hierarchically organized, from the most general, important
pieces of information to more specific, less important
pieces.

As part of the process of comprehension. a reader/
listener fills in the slots in the schema with information
from the text and, for slots not explicitly mentioned in
the text, with default values or values inferred from
other slots. This process is referred to as instantiation
of the schema. It is necessary for comprehension, since
the writer/speaker does not specify every piece of

information necessary for comprehension of a text.

The Application of Schema
Theory to Reading

One area in which schema theory has excited a great
deal of interest is in the area of reading comprehension.
First, I will discuss the traditional concepts of reading,
which conceptualize reading as a primarily unidirectional
process, and then more recent concepts, which conceive of
reading as an interactive process between the text and the
reader, including evidence of interaction between the text

and the reader’s knowledge on various levels.

Inadeguacies of Traditionsl Concspts of Resding
Models of reading (e.g., Gough, 1972; LaBerge and

Samuels. 1974) have been proposed which conceptualize
reading as being a unidirectional process from the printed
page into the mind of the reader. These models propose.
basically, that reading is a process in which symbols are
built up into words. words into sentences. and sentences
into overall meaning. These reflect traditional attitudes
toward reading. This conceptualization treats reading as
what is sometimes referred to as a bottom-up procsss,
because the reader begins with the lowest level, features
of symbols, from which the symbols are identified.
Strings of symbols are then analyzed into morphological
clusters, from which words are recognized, and then
strings of words are analyzed into phrases and sentences.
The reader thus works up from marks on a page through
words and sentences to meaning (Thomas, 1980).

However, many studies of the reading process have
produced results that would be impossible to explain in
terms of such a unidirectional model. On the symbol
level, there are a number of studies that have indicated
that identification of letters is influenced by their
semantic or syntactic context. For example. Nash-Ueber

(1975) showed that an ambiguous symbol (which, in

isolation, could be identified as either ”w” or ”ev”)
would be identified according to the context of the
sentence. Also. Reicher (1969) found that a letter is
perceived more accurately when it is part of a word than
when it is part of a string of unrelated letters. Even
orthographic knowledge plays a role in letter
identification. In a study reported by Rumelhart (1982),
participants were presented with strings of letters, some
of which had clusters of letters that were legal in
English and some of which had illegal clusters. Uhen
asked to name the letters in the string, participants
showed a tendency to transpose illegal clusters to make
them legal. though they rarely transposed legal clusters
to make them illegal. These studies and others indicate
that identification of letters depends on higher level
knowledge.

Further, perception of words is facilitated by their
syntactic and semantic environments. Kolers (1970)
analyzed substitution errors (errors in which an incorrect
word is substituted for a correct one) made by adults in
reading aloud. If the reading comprehension process in
unidirectional, it would be expected that errors would be
strongly influenced by visual similarity to the correct

word. However, Kolers found that, in 702 of the cases,

10

the substitutions errors were the same part of speech as
the correct word. (Eighteen percent would be expected
by chance.)

In similar studies, Ueber (1970) found that, in 90%
of the cases. first graders’ reading errors fit
grammatically with the text up to that point. and Stevens
and Rumelhart (1975) found that 98% of the substitution
errors recognizable as words were grammatically correct in
the sentence. Also. Meyer and Schvaneveldt (1971) found
that participants identifying pairs of words could
identify the second word faster if it was semantically
related to the first word, (e.g., "nurse" was easier to
identify if it followed ”doctor“ than if it followed
"bread"). The results indicate that word recognition is
facilitated by higher level knowledge.

Lastly, sentence comprehension is facilitated by
knowledge of semantics and the context in which it is
found. Schank (1973) found that ambiguous sentences
tended to be interpreted based on the reader’s semantic
knowledge (e.g., "I saw the Grand Canyon flying to New
York" was interpreted as "I saw the Grand Canyon ghilg I
gss flying to New York" rather than "I saw the Grand
Canyon ghigh was flying to New York.”) Uords with

multiple meanings are interpreted in terms of the context

11

in which they appear (Rumelhart. 1982). Bransford and
Johnson (1973) found that a sentence was much more likely
to be remembered if it was meaningful in the context in
which it was found than if it was meaningless. They
interpreted these findings to indicate that context is
used in interpreting sentence meaning, and a sentence that
does not fit the context may be difficult or impossible to
understand. even if it is syntactically and semantically
correct.

In summary, all of these studies show that lower
level processes--the identification of words and letters--
are influenced by intermediate and higher level
processes--sentence comprehension and knowledge about the
context. A conceptualization of reading must take into
account these indications that the reading process makes
use of top-down processing. As Thomas (1980: 34) stated,
”what readers gain from a text, then. is only part of
comprehension. Readers also integrate this information
with the existing knowledge in their heads."

Interactive Procgssiflg_in_flg§giflg

Goodman (1967, 1970) was among the first to spark the
recent interest in reading as an interaction among levels
of processing. rather than a bottom-up process. In this

conceptualization of reading, the process is viewed as one

12

in which the reader makes use of higher-level knowledge in
order to comprehend lower-level structures. Goodman
characterized reading as a "psycholinguistic guessing
game" in which the reader makes predictions and then
samples input to confirm or disconfirm the predictions.

The reader is not confined to information he receives

from a half inch of print in clear focus....The

reader uses syntactic and semantic information as
well. He predicts and anticipates on the basis of
this information, sampling from the print Just enough
to confirm his guess of what’s coming. to cue more
semantic and syntactic information. Redundancy and
sequential constraints in language. which the reader
reacts to. make this prediction possible (Goodman,

1970: 131).

According to Goodman, what happens "behind the eye“
is as important as what is on the printed page. Goodman
focused on how the reader’s syntactic and semantic
knowledge assists in the process of reading comprehension.
More recently, researchers have expanded this focus to
include ways that both background knowledge and knowledge
of discourse constraints assist comprehension.

Smith (1982) has been another strong and influential
proponent of an interactional conceptualization of
reading. Though. like Goodman, he did not deal directly
with schema theory, he emphasized the role of what he
referred to as ngn-visgsl informstig . Non-visual

information is information that does not come from the

printed page. including background knowledge and

13

linguistic knowledge. Smith characterized the role of
non-visual information as reducing alternatives in
reading. This allows the readers to make predictions
about what will be coming next, based on orthographic,
syntactic, semantic. and background information.

Uhat we have in our heads is a theory of what the
world is like, a theory that is the basis of all our
perceptions and understanding of the world....If we
can make sense of the world at all, it is by
interpreting our interactions with the world in the
light of our theory....Anything that I cannot relate
to the theory of the world in my head will not make
sense to me....The system of knowledge that is the
theory of the world in our heads has a structure Just
like any other theory or system of organizing
information. Information systems have three basic
components-~a set of categories, some rules for
specifying membership of the categories, and a
network of interrelations among categories (Smith,
1982: 54-57).

Smith emphasized the value of using this theory of
the world in prediction while reading.

An important difference between a skilled driver and
a learner is that the skilled driver is able to
project the car into the future while the learner’s
mind is more closely anchored to where the car is
now--when it is usually too late to avoid accidents.
The same difference tends to distinguish skilled
readers from beginners. or from anyone having
difficulty with a particular piece of reading. In
fluent reading the eye is always ahead of the brain’s
decisions. checking for possible obstacles to a
particular understanding. Readers concerned with the
word directly in front of their nose will have
trouble predicting--and they will have trouble
comprehending (Smith, 1982: 61).

Smith asserted that the predictions that readers make

on various levels help them avoid being overloaded by

14

information from the printed page. Uithout some
expectation for what was coming. a reader would be
overwhelmed with new information.

A considerable amount of research has been done over
the past decade on how reading comprehension is influenced
by content schemata. A number of studies have been done
on native English speakers showing that readers
demonstrate significantly better comprehension of texts
for which they have greater background knowledge (e.g.,
Spiro. 1980: Taylor. 1979: Langer and Nicolich, 1981:
Langer, 1984: Kintsch and Greene. 1978: Hare, 1982), or
which fit schema that they have (e.g., den Uyl and van
Oostendorp. 1980: Haberlandt. Berian, and Sandson, 1980:
Anderson, Spiro. and Anderson, 1978). Bransford and
Johnson (1972) found that. when readers had no schema for
a text. they recalled it significantly less well than if
they did. Kozminsky (1977) showed that, when the title of
a text caused the reader to activate one particular schema
rather than another. readers tended to comprehend the text
from the perspective of the schema suggested in the title.

Taylor (1979) and Eamon (1978-1979) did studies on
differences between the ways that good and poor readers
use schemata. Using third and fifth graders, Taylor found

that both good and poor readers had better comprehension

15

of familiar material. However, the difference was
significantly greater for unfamiliar material. This
indicates that poor readers are able to use top-down
processing when material is familiar. but that they resort
to bottom-up processing when the material is unfamiliar.
Taylor did not speculate on how good readers were
processing the texts. However, assuming that there
actually were no differences in the familiarity of the
texts, it appears that good readers were actually using
some form of interactive processing. Possibly they were
using either more general content schemata or textual
schemata.

Eamon (1978-1979) found that good readers were able
to recall more directly relevant information than
information of less relevance. In contrast, for poor
readers, there was less difference between memory for
relevant and irrelevant material. This indicates that
good readers use schemata better than poor readers do.

All of these studies demonstrate the importance that

content schemata have in reading comprehension.

Content Schema and
Second Language Readers
As with first language readers. the traditional

emphasis has been on the language to be comprehended and

16

not on the person doing the comprehending.

In these essentially ’linguistic’ views of

comprehension each word, each well-formed sentence.

and every well-formed text ’has’ a meaning. Meaning
is conceived to be ’in’ the text, to have a separate,
independent existence from the reader. Failures to
comprehend a nondefective text are viewed as being
due to language-specific deficits--perhaps a words
was not in the reader’s vocabulary. a rule of grammar
was misapplied. an anaphoric cohesive tie was

improperly coordinated. and so on (Carrell. 1984:

332).

The recent interest in reading and schema theory
among first language reading specialists has attracted the
attention of second language reading teachers and
researchers. Second language reading theorists are coming
to recognize that what the reader knows is as important as
what is on the page.

Research on Sscond Language Learners

A number of studies have been done with second
language learners in relation to reading and schema
theory, using both content and textual schemata. The
results of these studies are sometimes difficult to
compare. because of the differences in participants. sets
of variables, and techniques used. Also. none of these
studies have been replicated. However. they can hopefully
shed some light on how or whether second language learners

make use of schemata in reading. In this section, I would

like to discuss the studies and the implications they have

17

about second language readers’ use of schemata.

A study by Steffensen. Joag-Oev. and Anderson (1979)
did not use language specifically as a variable. since the
participants were American native English speakers and
Indians (natives of India) whose English proficiency was
very high. However, the study did address the issue of
cultural differences in background knowledge and how this
knowledge assists or interferes with comprehension(/A

The researchers had American and Indian participants
Tread descriptions of weddings from their own culture and
:from the other culture and then write descriptions of the
Rweddings. Members of both groups read the description
from the other culture more slowly than the one from their
own. They not only remembered more details of the wedding
in their own culture and remembered them more accurately,
they were able to draw correct inferences from the
description of the wedding in their own culture. This
would tend to indicate that the readers were making use of
a schema that is specific to their culture for weddings in
storing the text in memory and in interpreting the meaning
of the text.

Participants also often drew incorrect inferences
from the description of the wedding from the other

culture. based on the schema for weddings in their own

18

culture;’ For example, the description of the American
wedding included a mention that the bride wore her
grandmother’s wedding gown. An Indian participant
reported. in the recall protocol, that the dress was.
unfortunately, old and out of fashion. a comment which
would fit the Indian view that it is important for the
bride’s family to show their economic status by providing
a new, fashionable wedding sari for the bride to wear.
American readers, in contrast, emphasized the aspect of
family tradition involved in the bride wearing her
grandmother’s dress. which fits with the schema for
American weddings. Apparently when readers do not have an
appropriate schema (as the Indian readers did not have a
schema for American weddings) they interpret input
according to the most similar schema that they do have
(i.e., the schema for Indian weddings).

In addition. recall protocols were often vague in
areas where readers did not have schemata. For example.
the text on the Indian wedding contained details about the
gifts exchanged by the families. Indian respondents not
only tended to remember what the gifts were. they
commented on the significance of the gifts. In contrast.
descriptions of the gifts by Americans were vague and did

not include any information about the significance of the

19

gifts.

This study is an indication that, even in the native
language or a strong second language, the reader’s
knowledge of the schematic background of a text greatly
aids comprehension. In addition, lack of an appropriate
schema for a text hampers accurate comprehension. It is
likely to be even a more serious problem for readers of a
second language who have less-than-adequate language
proficiency.

/

//One problem with this study is that the researchers
did not measure the prior knowledge that the participants
had about weddings in the other culture. Uhile most
Americans probably do not know much about Indian wedding
customs, Indians living in the US may have had the
opportunity to attend an American wedding or may have
learned about them from other sources. /Participants who
knew about wedding customs from the other culture may have
produced more accurate recall protocols than those who did
n o t .//

// This issue raises an important question in the
intercultural study of schemata and comprehension. It is
not clear whether the process of comprehending is

different when making use of schemata from one’s own

culture than when making use of schemata learned from

20

another culture. The question is complicated by the fact
that schemata can overlap between cultures. Members of
one culture do not necessarily share all of the same
schemata. By the same token, members of different
cultures can share the same schemata. For example,
medical professionals in different cultures may share
medical content schemata with each other that they do not
share with other members of their own cultures.
fThough some second language reading teachers have

long recognized that their students were better able to
deal with culturally familiar material (Carrell, 1983c),
it was not until the 1980’s that this was empirically
tested. Studies of second language readers’ ability to
comprehend a text have traditionally focused on such
factors as lexical difficulty and sentence length and
complexity. However, when Johnson (1981) investigated the
effects of text complexity and cultural background on
comprehension, she found that, for second language
readers, cultural background had a greater effect on
ability to understand the text than did semantic and
syntactic complexityr/

I //Johnson had American participants and intermediate
and advanced Iranian participants read simplified or

unsimplified versions of two passages, one based on

21

American folklore and one on Iranian folklore. For
Iranians, she found an interaction between cultural
background and language complexity for the total number of
events recalled. More events were recalled from the
simplified American story than from the unsimplified one,
though there was no difference for the Iranian story.
However, she did not find an interaction between language
complexity and cultural background or a main effect for
language complexity when she measured inferences made from
the texts. Iranian participants produced more inferences
from the culturally familiar story than from the
unfamiliar one, regardless of language complexity.

Americans remembered more propositions from the
American story, and more from the unsimplified stories.
raising the possibility that the structure of adapted
texts is somehow different from the text structure that
native speakers are accustomed to. Like the Iranians,
Americans produced more inferences from the culturally
familiar story than the unfamiliar one.

The content schema variable may have been confounded
in this study by a textual schema variable. Possibly
knowledge of the textual schemata, the conventionalized
arrangement of the events and themes in a story, had

something to do with the fact that readers understood the

j!

22

folktale from their own culture better.

The results of this study indicate that Iranian

lsecond language readers do make use of schemata, whether

i content schemata or textual schemata, at least in reading

narratives. Also, content or textual schemata seem to be

1 more important than language complexity in making

Kinferences from the text, though simplified language seems

to be helpful in comprehending the events in the story.

“Carrell (1983b) looked at the effects of three
components of background knowledge: prior knowledge
(familiar vs. novel), prior cues to the content of the
text (context vs. no context) and the degree to which
lexical items reveal the content area (transparent vs.
opaque). Her participants, who were native speakers of
English and high-intermediate and advanced nonnative
speakers, included students from a variety of countries.
but they were primarily Malaysians. In a 2 x 2 x 2
design, Carrell used the variables of context (providing
or not providing and title and picture), transparency

(presence or absence of vocabulary that provided cues to

3the content area) and familiarity (presence 0” absence °I

prior knowledge about the content):/ For the advanced
group of students, familiarity was the only significant

variable. More was recalled from the novel text than the

23

familiar text. For the high-intermediate students.
significantly more was recalled from the transparent
version than the opaque version. All three variables
affected the recall of native speakers.

Carrell concluded that native speakers make use of
contextual cues to do top-down processing and lexical cues
to do bottom-up processing. and that, at least in short
term memory, novel information is more memorable. The
nonnative speakers in Carrell’s study, on the other hand,
are not efficient about making use of context or textual
cues. They are not particularly proficient at either top-
down or bottom-up processing. If this study has any
external validity at all. it raises the question of how
_ the advanced students, who were attending regular
\university classes, were managing the reading required in
‘ihose classes}?

One flaw in this study is operationalization of the
familiar/novel variable. As a familiar text, Carrell used
a description of the procedure for washing clothes. The
largest number of participants in this study were
Malaysians. However, according to a Malaysian informant,
some of the steps in the procedure as it is described in
the text would not have been familiar to Malaysians. For

example. the writer of the text mentioned the possibility

24

of going somewhere outside of the home to wash clothes.
Malaysians do not use laundromats in Malaysia. Also, the
writer mentioned dividing clothing into different groups
according to color. This is another thing that Malaysians
do not do. except in fairly rare cases of clothing whose
colors might run (personal communication, A. Ibrahim, June
1, 1987). Therefore, the supposedly familiar text may
have been at least partly novel. This may be a case where
the nonnative readers were using an event schema that was
different in content than the one that the writer
intended, rather than one where they are not able to
activate a schema that they already had.
1 This points up a problem in doing schema research
‘with nonnative speakers. when measurement indicates that
1 second language readers have not made use of schemata, the
I question arises of whether the problem is that they do not
t have the necessary schema, that they have not activated
fithe schema that they have. or that they have activated an
inappropriate schemaff This will be discussed in detail in
the following sections?

Another weaknes; of the study was that there was no
manipulation check on the prior knowledge manipulation.
It is possible that some of the participants did not

understand the significance of the picture that was

25

presented to them. so that participants in the prior
knowledge condition may not have had prior knowledge.

//Hudson (1982) looked at the effects of different
prereading exercises intended to activate schemata on the
reading performance of beginning, intermediate and
advanced students. primarily IranianSy//In one condition,
students were shown pictures related to the text, which
they discussed. Afterwards, students privately generated
predictions about the content of the text. In the second
condition. students were given a list of vocabulary words
and definitions. Both of these groups answered questions
about the content of the text. In the third condition,
students read the text, answered questions about it, read
the text again, and answered the same questions again.
Scores were significantly higher for the first condition
for beginning and intermediate students. There was no
difference for advanced students. Hudson interpreted this
to indicate that advanced readers were able to activate
schemata without outside help, but at the lower levels,
some help in activating schemata was necessary.

Uhile this study does indicate that prereading

exercises are of some use to beginning and intermediate
students, it is not clear from the study what aspects of

the prereading exercise were effective in activating the

26

schema.r In the first condition. participants saw pictures
related to the story, they discussed what was in the
pictures. and they generated predictions. This raises a
number of questions about the necessary and sufficient
conditions for activating schemata using prereading
exercises. Uas it one of these three steps that activated
the schema, or an interaction of the three? Uould, for
example. studying the pictures without discussing them
have produced the same results? was generating
predictions a necessary part of the process? Uould
generating predictions have been sufficient alone?

Further research to clarify the answers to these questions

would be useful.

”—

/

f”fAlso, since there was no control group that did not
reéeive any treatment, we do not know how the treatments
Hudson used compared with no treatment at all, though the
assumption probably would be that groups with no treatment
would not have done as well as students in the first
condition. If that assumption is correct, this study
indicates that it is inadvisable to let students at the
intermediate level or below read a text ”cold".

Use of Lexigal Knoglsdg . Several studies have been

done on the degree to which readers depend on the lexicon

as they read. Cziko (1980) analyzed errors made in oral

27

reading by intermediate and advanced readers of French as
a second language and by native French speakers. He found
that intermediate readers made errors that indicated
greater reliance on graphic information than on contextual
information, indicating a bottom-up strategy dependent on
lexicon. Advanced readers, on the other hand, depended
more on context, though not as much as native speakers
did.

UliJn (1981, 1984), using native Dutch speaking
readers of French as a second language, found that more
misunderstandings were caused by problems with vocabulary
than by problems with syntax. This may indicate that
readers are more dependent on their knowledge of
vocabulary than on their knowledge of syntax.

Yorio (1971) did a survey of native Spanish speakers
studying English as a second language and found that they
felt that their lack of vocabulary was their greatest
handicap. Lack of knowledge of grammar was considered a
considerably less serious problem.

These studies indicate a pattern of dependence on
vocabulary by second language readers. what this seems to
indicate is that readers depend on bottom-up processing in
reading in their second language. while Cziko (1980) and

Ulijn (1981, 1984) did not appeal to schema theory in

28

in their explanation, it does seem to fit with the results
that they found. Second language readers appear to
identify content words and make use of schemata to
identify relationships among those words.

Implicstions of Research

Based on these studies, it appears that concern about
linguistic difficulty of texts is at least partially
misplaced. In addition to lack of adequate linguistic
proficiency, lack of appropriate schemata related to a
text. or inability to make use of the schemata, can be
sources of serious problems in comprehension. Second
language readers seem to make some use of the schemata
that they have in second languages, though they seem to do
so less well than native speakers do. At intermediate
levels of reading proficiency, readers seem to be using a
bottom—up strategy of processing, in which they identify
content words and use schemata to determine how the words
should be related.

Hudson (1982) found that a prereading technique
iinvolving looking at pictures, discussing the pictures,
iand privately predicting what a text will be about seems
'to be useful in activating schemata for nonnative
speakers); However, due to confounding of the variables in

this study, it is not possible to determine which of the

29

three steps, or combination of steps, in the prereading

activity produced the effect.

Much more research is needed in the area of second
language reading and schema theory and how these apply to
the classroom before definitive conclusions can be reached
in this area.

Hypotheses and Resgsrch Qggstions

Based on the results of previous studies with native
and nonnative speakers, I propose the following hypotheses
for this study:

H : A positive correlation will exist between background
knowledge and reading comprehension, when lexical
proficiency and reading proficiency are held constant.

For the purposes of this study, Lssding comprehension
refers to understanding the meaning of a specific reading
passage, while readingiproficiency refers to general
reading skill. Studies with native speakers have
demonstrated the correlation between background knowledge
and reading comprehension fairly clearly. Uhile the
results of studies with nonnative speakers have been
mixed, there is some indication that background knowledge
improves reading comprehension.

H2: Readers with high background knowledge will have a
lower correlation between reading comprehen51on and
reading proficiency than readers with low background

knowledge.

This hypothesis is based on the previous hypothesis.

30

If prior knowledge improves comprehension, it follows that
there will be a lower correlation between reading
proficiency and reading comprehension for readers with
high prior knowledge than for readers with low prior
knowledge. The comprehension of readers with high prior
knowledge is improved by their prior knowledge. Readers
with low prior knowledge, on the other hand. depend more
on their reading proficiency, therefore their reading
comprehension depends more on their level of reading
proficiency.

In addition. I will be looking at the following
research questions:

R1: Uhat are the relationships among prior knowledge,
reading comprehension and lexical proficiency?

Research (see citations in the "Interactive
Processing in Reading" section) indicates that second
language readers depend on lexical knowledge to a greater
extent than they do on grammatical knowledge. Readers
with high prior knowledge are able to make better use of
their lexical knowledge, because they should know what
relationships to expect among concepts in the reading.
Therefore, it could be expected that readers with high
prior knowledge would be able to make better use of their
lexical knowledge than readers with low prior knowledge

and would therefore have a higher correlation between

31

lexical knowledge and prior knowledge than readers with

low prior knowledge.

R : How does reading comprehension differ for reading
passages that require different amounts of background
knowledge?

Bransford and Johnson (1972) found that readers
comprehended passages for which they had schemata better
than passages for which they had no schemata. However,
the passages used were not based on authentic passages and
seemed somewhat contrived. Carrell’s (1983b) replication
of this study using second language learners had a
possible problem in that the participants may not have had
the schema for washing clothes that she assumed that they
had. However. it seems logical that the greater amount of
background knowledge a passage requires, the more
difficult it would be for a reader with low background
knowledge related to that passage (i.e., without a well

developed schema) to comprehendIand the easier it would be

for a reader with high background knowledge to comprehend.

Measurement Issues
In reading and schema research with nonnative
speakers, there are special issues related to measurement.
These include: 1) how to measure background knowledge,
2) how to measure comprehension of the text, and 3) how to

determine whether a schema has been activated. Another

32

measurement problem, though one not limited to research
with nonnative speakers. is how to measure the amount of
background knowledge necessary to comprehend a passage.

In this section, I would like to look at how research done
on both native speakers and nonnative speakers applies to
this issue. Uhile it is necessary to be cautious in such
applications, results from research with native speakers
can at least give some indication of possible solutions
for the problems peculiar to nonnative speakers. I will
also make some suggestions about methodological standards
for research done in the area of reading and schema theory
with nonnative speakers.

Measuring background knowledge and text comprehension
are issues that researchers who use native speakers have
dealt with. However, for the most part, studies with
nonnative speakers have used participants where the
background knowledge of the participants could be assumed
to be either high or low. Uhile this strategy avoids
problems inherent in measuring the prior knowledge of
nonnative speakers, it may allow for rival hypotheses
(e.g., Carrell, 1983b), since the assumptions about
participants’ background knowledge might be wrong. Being
able to measure background knowledge would help eliminate

rival interpretations in such studies. In addition to

33

mistaken assumptions about levels of background knowledge,
manipulating background knowledge by using different texts
may result in confounding variables, such as the
difficulty level of the text.

Studies using a second language. especially at
beginning or intermediate levels of proficiency, present
problems not faced in studies using the participants’
native language. It goes without saying that in measuring
background knowledge or comprehension of a text, it is
necessary that the method avoid confounding the variables
involved. Nonnative speakers present special problems in
this area. For example, if a method of measurement
requires writing skill or reading skill on the part of the
participant, participants of low proficiency may appear
less knowledgeable than they actually are or may appear to
comprehend the text less well than they actually do.

flgthods of Assessing Background Knowledg . Since
language ability is likely to be an issue, one possibility
would be to test participants’ background knowledge in
their native language. Uhile this would eliminate the
possibility of participants’ second language proficiency
interfering with their ability to express what they know,
there are likely to be practical problems. If all

participants are from the same native language background,

34

and if help is available for writing material and for
coding in that language. this is a possibility. However,
there are often not enough participants available from one
language background, so participants from several language
backgrounds are used. This makes it impractical to test
participants’ background knowledge in their native
language. In such cases, or if, for some other reason,
the researcher chooses not to measure background knowledge
in the participants’ native language(s), a method of
measurement must be chosen that minimizes confounding of
the measurement of background knowledge with level of
reading or writing proficiency.

Taylor (1979) and Stevens (1980) used multiple choice
questions to assess background knowledge of participants.
Spilich, Besonder, Chiesi, and Voss (1979) used completion
items. These both are possibilities for testing the
background knowledge of nonnative speakers. They share
the advantage of being both easy to respond to and simple
to score. However, one problem is that good multiple
choice and completion questions are difficult to write.
Multiple choice questions require carefully written
alternatives so that the correct response is clearly
correct but that distracters sound plausible. The stems

of completion items must be written so that it is clear

35

what the expected answer is. A valid, reliable instrument
also requires extensive pretesting (Heaton, 1975).

Another problem is that unless the researchers are
very careful to make sure that the language is within the
reading ability of the participants. the issue of whether
reading ability influenced the results is likely to arise.
For completion items, responses must not require active
vocabulary beyond participants’ proficiency. If there is
a high correlation between the measure of background
knowledge and of comprehension of the text, this might be
attributed to the possibility that both measures are
measuring reading ability.

The free association method of assessing background
knowledge was originally developed by Langer (1980, 1981)
as part of a prereading activity called the Pre-Reading
Plan (PReP). Langer and Nicolich (1981), Langer (1984),
and Hare (1982) further developed the procedure as a
measure of prior knowledge. In this procedure, the
participants are given three key content words or phrases
related to a topic and asked to write anything that comes
to mind when they hear each word or phrase. Responses are
categorized by coders into three categories according to
the amount of prior knowledge that they indicate--much (3

points), some (2 points) or little (1 point), with

36

specific criteria for each category. (Langer originally
used twelve levels of prior knowledge but found that these
could be collapsed into three categories.) Scores were
averaged for the three words or phrases. Langer and
Nicolich (1981) reported intercoder reliability of .82.
They found that these measures of prior knowledge were
highly correlated with comprehension of the text and had
no significant relationship with measures of IQ. Hare
(1982) used the same procedure and found similar results.
Hare reported an interrater reliability of .90.

Uhile this procedure has some drawbacks for measuring
the background knowledge of nonnative speakers, it also
has some advantages. A major advantage of this method is
that it requires very little reading. As long as care is
taken to choose key concept words that the participants
will know, reading ability should not influence the
results. (If this is a concern, key concept words might
be translated into the native language(s) of respondents.)
Another advantage is fairly high intercoder reliability,
which might be difficult to achieve in methods that
require more writing.

This method of measuring background knowledge is more
cumbersome to score than multiple choice or completion

items. The main drawback is that it does require writing.

37

However, the writing requirements can be fairly minimal,
with only a few phrases or sentences for each key concept
word. If coders can Judge the quality of the responses of
nonnative speakers, independent of their writing ability,
this problem may be at least partially overcome. The most
serious problem would still be likely to be with beginning
readers, who may not be able to express their knowledge of
the concepts at all, resulting in lower ratings than they
should receive.

Uhile Langer’s free association method has not, to my
knowledge, been used as a measure of background knowledge
with nonnative speakers, I think it is a promising
measure, at least for participants with an intermediate to
advanced level of proficiency in the target language.

Assessing Reading Comprehension. Methods of
assessing reading comprehension have the same problems as
measures of background knowledge. Stevens (1980, 1982)
and Langer (1984) used multiple choice questions. and
these have the same advantages and disadvantages mentioned
above. Hare (1982) used short answer questions. For
nonnative speakers, these would have the disadvantage of
requiring writing and of being more difficult to score
than multiple choice or completion items.

Nunan (1985) used a cloze procedure to test

38

comprehension. He does not go into detail about choice of
words for deletion. Obviously, the choice of words for
deletion would be very important. An advantage of using
the cloze procedure is that comprehension of different
types of things could be checked by deleting different
types of words. For example, cohesive ties could be
deleted to test comprehension of cohesive relationships.
A disadvantage, however, is that it would be difficult.
though probably not impossible, to test participants’
inferences from a text using the cloze method. Since
drawing inferences seems to be a particular problem in
material from another culture, it is important that the
method of assessing comprehension be able to assess
ability to draw inferences from a text.

A more customary way of assessing comprehension in
schema theory research is through recall protocols (e.g.,
Fass and Schumacher. 1981: Spilich, Besonder, Chiesi, and
voss, 1979: Kintsch and Greene. 1978). For high
intermediate and advanced participants. this is probably
not a problem, as long as coders are not influenced by
writing ability in their Judgments about comprehension.
For participants with lower proficiency. this is more of a
problem, since they may be entirely unable to express some

of the concepts that they actually did understand.

39

Assessing Schema Activstion. In doing research

related to reading and schema theory with nonnative
speakers, questions of whether a schema was activated, or
which schema has been activated. may be raised. Results
of studies using native speakers provide some information
about how schemata operate when they are activated. Some
of these studies may be useful in developing measures of
whether schemata have been activated.

Studies by Pichert and Anderson (1977) and Anderson
and Pichert (1978), and Pass and Schumacher (1981) suggest
a technique that might be useful in answering this
question. They had participants read a story in which a
house was described. Half of the participants were asked
to read the story as if they were potential homebuyers:
the other half were asked to read it as if they were
burglars planning on breaking into the house. when asked
to recall the passage, participants recalled more pieces
of information that were important from the perspective
that they had been told to use in encoding the passage
(e.g., for the homebuyer, that the basement was damp and
musty and the house needed a new roof: for the burglar,
that the house was not visible from other houses and that
a side door was left unlocked during the day).

Similarly, Kozminsky (1977) had participants read

40

passages with different titles. Participants tended to
recall more propositions from the point of view suggested
by the biasing title.

The results of these studies might be applied to the
question of whether a schema had been activated by
nonnative speakers using a similar method. Participants
could be instructed to encode a passage from one
perspective or another. Readers who recalled
significantly more propositions related to the perspective
they had been assigned would be considered to have
activated that schema. If there was no difference in the
number of propositions for the two perspectives. the
schema was not activated. Researchers might even be able
to analyze the propositions that were produced to see if
they did match a particular schema. Uhile it might be
difficult to identify this schema with certainty, this
analysis might at least point the researchers in certain
directions. Since this method of research has never been
applied to nonnative speakers. research is necessary to
see how nonnative speakers do perform in these
circumstances.

Findings of Bransford and Johnson (1972, 1973)
suggest another possible technique. Bransford and Johnson

(1972) did a series of studies in which they presented

41

participants with a paragraph with a disambiguating
context (in the form of a picture) given before reading,
after reading or not at all. Participants who did not
have a context or who were given the context after reading
the passage remembered the passage less well than those
who were given the context before reading the passage.
Participants who were not given a context from the
beginning reported that, as they read, they did attempt to
search for some schema that would make the paragraph
meaningful.

Bransford and Johnson (1973) presented participants
with identical paragraphs with two different titles. One
sentence in the paragraph was meaningless from the
perspective suggested by one of the titles but meaningful
from the other context. This sentence was found
significantly more frequently in the recall protocols of
participants with a meaningful context than in the
protocols of the participants that did not have a
meaningful context.

Background Knowledge Reguireg to Understand a

Pssssg . A measurement problem for researchers with both

 

native and nonnative speakers is how to objectively
measure the amount of background knowledge necessary to

comprehend a passage. Uhile there are various ways to

42

measure the difficulty of reading passages that take into
account such factors as grammatical complexity, sentence
length, and lexical difficulty, there does not appear to
be an established method for measuring the amount of
background knowledge necessary to comprehend a passage.

One possibility would be to ask readers familiar with
the topic of a passage to list the pieces of background
information that they felt would be necessary for full
comprehension. Uhile this might not provide an absolute
measure, it should demonstrate the relative differences
between passages.

Another possibility is to ask proficient readers with
little background knowledge to rate passages according to
their difficulty. Again. this would not provide an
absolute standard, but it would indicate the relative

contribution of background knowledge to comprehension.

Methodological Standards
Because of the complex issues related to measurement
in research in the area of schema theory and reading with
nonnative speakers, it becomes important to set
methodological standards for this research. In this
section, I would like to propose some standards suggested

by the considerations discussed in the previous sections.

Measuring Backgrognd Knowledg . The first criterion

43

is that researchers should measure background knowledge,
or determine it by some other method, rather than assume
the background knowledge of participants from other
cultures.

In the case of research with native speakers,
especially related to event schemata, background knowledge
is sometimes assumed (e.g., Bransford and Johnson, 1972:
Taylor, 1979). In studies using event schemata with
members of one culture or subculture, this may be
Justified. since event schemata seem to be highly
conventionalized within a culture (Bower, Black, and
Turner, 1979). However, assumptions about content
schemata are more difficult to make when using
participants from different cultures (or even subcultures
of one culture). Therefore, it is important for
researchers to get information about the schemata of the
members of other cultures.

This may be done in a variety of ways, for example,
by using questions. by using Langer’s free association
technique. by asking participants to list the steps in an
event schemata (as Bower, Black and Taylor [1979] did), or
by use of informants from that culture. (In the latter
case, researchers must be careful to use informants from

the same ethnic and social group as the participants for

44

the research, since schemata might differ among different
groups.) Researchers should also be cautious about
combining participants from different cultures together,
since their schemata may differ. (This may be a practical
problem for researchers in English-speaking countries,
since it is often difficult to find a sufficient number of
participants from the same country.) Also, if the
researcher manipulates background knowledge, as Carrell
(1983b) did, there should be manipulation checks.

Measurement Instruments. The second criterion is
that instruments used to measure schemata and reading
comprehension should take into account the language
proficiency of the participants.

As with any other measures used in research,
measurement instruments used in reading and schema theory
research with nonnative speakers should be reliable and
valid. For nonnative speakers, the language proficiency
of the participants is of particular concern in assuring
that measures are reliable and valid. It goes without
saying that measures should be pretested to help insure
this. For example, a researcher using multiple choice
questions must be concerned about the reading proficiency
required to answer the questions. while one using recall

protocols must be concerned about writing proficiency,

45

both in terms of the participants’ ability to express
themselves and the coders’ ability to Judge content apart
from grammatical errors.

Measuring Activation of Schemsta. The third
criterion is that researchers should consider instruments
that include some measure of whether the appropriate
schema has been activated.

In the previous section, two methods were suggested
by which activation of schemata could be measured. Since
there is still some question about the degree to which
nonnative readers make use of schemata, researchers might
include measures of whether readers have activated a
schema. (In addition, more research is necessary on the
measures of schema activation.) This would hopefully lead
to a greater understanding of how nonnative speakers use
schemata.

Prereading Activities. The fourth criterion is that
researchers doing studies related to prereading activities
in the classroom should avoid confounding the treatment
conditions that they use.

Uhile Hudson (1982) made an important contribution to
understanding of the effect of prereading exercises,
questions remain because it is not clear which of the

elements of the prereading activity caused the effects

46

that Hudson found. Researchers who do future studies in
this area should operationalize the variables and should
design prereading activities in a way that would allow
them to determine which element(s) of the activity caused
the results that were found.

Maasuremani af Language Praficiangy. The fifth
criterion is that researchers should measure the reading
proficiency of participants and use it as a variable.

Previous studies have shown differences in results
for readers of different levels of proficiency (e.g.,
Carrell, 1983b: UliJn, 1984: Hudson, 1982: Cziko, 1980).
Future researchers should measure the reading proficiency
of participants and use it as a variable in the analyses.

Measuring Reguireg Background Knowladg . The last

criterion is that researchers comparing two or more

 

passages should take into account the relative amount of
background knowledge necessary to comprehend the passage.

Again, more research is needed on this issue to
identify methods of measurement and demonstrate their
validity and reliability. However, the demonstrated
importance of background knowledge to comprehension makes
it necessary to differentiate the amount of background

knowledge required by different passages.

47

Conclusions
Imglications for the Second Language Classrogm

Research on schema theory in general and its
application to reading in a second language in particular
has important implications for the second language
classroom. Studies show that readings with an unfamiliar
background, including an unfamiliar cultural background.
are more difficult to understand. Understanding
inferences and relationships among different pieces of
information is a particular problem, for both native and
nonnative speakers.

In second language reading classes, teachers and
materials developers should consider findings related to
reading and schema theory. (However, as Carrell [1986]
pointed out, most of the research that has been done has
been in the area of schema theory and second language
learning has not been done in classroom settings, and that
which has been done has not been applied to a variety of
settings. Therefore. caution is necessary in applying
conclusions to classroom teaching.) The findings have
implications in three major areas: 1) choice of
materials, 2) helping students activate schemata, and
3) helping students deal with reading texts independently.

I will discuss each of these briefly.

48

Choice of Materials. Johnson (1981) found that
familiarity of the cultural background of a text had a
greater influence on comprehension than did language
complexity. Uhile it is not appropriate to ignore
language complexity entirely in choosing materials, this
indicates that familiarity of the background of materials
should be a maJor factor in decisions about choices of
reading materials, especially at lower levels of language
proficiency.

Heleing Students Activate Schemata. Students may
have some difficulty activating their schemata
independently. Hudson (1982) found that students at the
beginning and intermediate levels who experienced a
prereading activity specifically intended to activate a
schema did better than students who did other prereading
activities. These results indicate that prereading
activities are important, especially at the beginning and
intermediate levels. However, more research needs to be
done to refine knowledge about what is necessary in such
prereading activities and what other activities might be
useful. Uilson (1987) has suggested a number of
prereading activities, including having students work in
groups to list the things they already know about a topic,

making predictions about a reading, then confirming their

49

predictions. and discussing the historical background of a
text. Uilson’s suggestions, along with others that have
been made (e.g., Hudson, 1982: Langer, 1981), can form the
basis for future research on prereading activities and
schema activation.

Preearing Stugents to Reag Indeeendently. The
previous section raises another issue--how well do
techniques for activating schemata in a classroom setting
transfer to independent reading? It is vital that the
reading classroom prepare students for reading outside of
the classroom. Can second language students be taught to
do prereading preparation that will help them activate the
schemata that they need to use? To answer these
questions, more research is needed, preferably over a
period of time. Hudson’s prereading technique might be
modified so that students are taught to look at the title,
subheadings, etc., from a text, think about what they know
about the subject, and make predictions about what is
likely to be included in the text. Carrell (1985) found
that students could be taught to identify the textual
schema for a text, and that this helped them comprehend
the text better. Possibly a similar technique would work

for content schemata.

Chapter II
METHODS
Overview
In this study, I looked at how East Asian nonnative
English speakers make use of schemata when they read
English. In particular, I compared amount of prior
knowledge, reading proficiency, and knowledge of

vocabulary to reading comprehension.

Instrument

The instrument was pretested with twenty participants
from Japan, China, Taiwan, and Korea. Eight passages were
pretested. from which three (Kitao, et al., 1983: 137:
Lubin, 1983: 89; and Reichler, 1987: 93) were chosen.
Passages were chosen on which nonnative speakers showed a
range of responses for number of pieces of information
recalled and prior knowledge. The passages chosen showed
a high degree of variance on the amount of prior knowledge
and recall demonstrated by participants and in the amount
of prior knowledge required for comprehension.

The first section of the final measurement instrument
consisted of nine phrases, all of them translated into the
native languages of the participants. Respondents were
asked to write down three things that they knew about each

phrase, in order to demonstrate their knowledge about that

50

51

topic. Next. participants read a passage. After reading
the passage. respondents answered vocabulary items.
Participants were next asked to write down fifteen pieces
of information that they remembered from the passage,
provide a title for the passage, and state the main idea
of the passage in one or two sentences. Asking
participants to write down what they remembered from the
passage was intended to measure their comprehension of the
details of the passage. Asking them to suggest a title
and state the main idea was intended to measure their
comprehension of the overall meaning of the passage.
These measures were repeated for three passages. (See
appendix.) The order in which the passages was presented
was counterbalanced.

Passages were used that required different amounts of
background knowledge for comprehension. The amount of
background knowledge required to comprehend a passage was
measured by asking native English speakers with knowledge
of the topic of the passage to list the pieces of
background information, excluding lexical and grammatical
knowledge. necessary for understanding that passage.
(Before doing this the respondents were given an example.)
The responses were evaluated by blind coders. who counted

the number of pieces of information that were necessary

52

for comprehending the passage. The number of pieces of
information considered to be necessary was compared using
t-tests.

In the third section. to measure reading proficiency,
participants were given a passage with blanks in place of
some words. They were asked to fill in the blanks with
appropriate words from alternatives provided. In the last
section of the measure, participants provided such
demographic information as age, nationality, and length of

time spent in English-speaking countries.

Participants

Participants were ninety-six natives of China, Japan,
Korea, and Taiwan living in or visiting the United States.
mostly students and housewives. (Participants from these
countries were used because they learned English in their
native countries using similar methods [grammar-
translation method], they had little exposure to English
in the native countries, and none of their native
languages are related to English.) Fifty-three
participants were regular Michigan State University
students. Thirty-one were English Language Center
students. The remainder were spouses of students or
students in the Haslett adult basic education program.

Volunteers were solicited in language classes and

53

contacted by phone through the membership lists of the
Japanese. Taiwanese and Korean clubs, through lists of
international students maintained by the dormitories, and

through personal contacts from those countries.

Procedures

Participants were asked to come to a meeting room or
classroom at specified times to fill out the measurement
instrument. Before filling out the measurement
instrument. participants were asked to sign a consent
form. No instructions were given other than those
included in the instrument. except to state the time
limits for each section. However, participants were
allowed to ask questions of the researcher if any of the
instructions were unclear.

For the first section (prior knowledge), participants
were allowed as much time as they needed to finish the
section. However, they were encouraged to try to finish
it within fifteen minutes. Participants were allowed
three minutes for each reading passage and four minutes
for each vocabulary section. Eight minutes were allowed
for participants to write down what they recalled from the
passage and answer questions about the title and main idea
of the passage. If participants had not started on the

last two questions after six to seven minutes, they were

54

reminded of them. Participants were warned when one minute
remained. No time limit was put on the reading
proficiency test or questions on demographic information.

The time limits were based on pre-testing. A
sufficient amount of time was allowed for most

participants to comfortably finish each section.

Measurement

The amount of prior knowledge was rated by blind
coders on a scale of zero to three. based on the system of
coding developed by Langer (1984). A rating of three
points was awarded for a response that indicated much
prior knowledge (e.g., superordinate concepts, precise
definitions. or detailed and accurate explanations). A
rating of two points was given for a response that
indicated some prior knowledge (e.g., examples or
attributes of the concept). A rating of one point was
given for responses that indicated little prior knowledge
(e.g., first hand experiences or sound alikes). A rating
of zero was given for irrelevant responses or no
responses.

Participants’ comprehension of each passage was
measured in three ways. It was measured according to
1) blind coders’ ratings of the quality of the title

provided, 2) coders’ ratings of the quality of the

55

statement of the main idea of the passage, and 3) coders’
count of the number of idea units from the passage
included in participants’ summaries of the passages.

The quality of the title and summary was rated on a
scale from zero to three. A rating of three points was
awarded for titles and summaries that expressed the main
idea of the passage particularly well and insightfully
(e.g., Scarlett O’Hara--Her Life and Character). A rating
of two points was awarded to titles and summaries that
included information from the passage but did not seem to
express the intent of the passage as a whole (e.g.,
Baseball is very popular in the United States). A rating
of one point was awarded for titles or summaries that
covered only a small portion of the passage (e.g.,
Reconstruction Atlanta). For missing or irrelevant titles
and summaries, no points were awarded.

Idea units were specified a priori from the reading
passages (Carrell, 1983b). They were words or phrases
that corresponded generally with subjects, obJects, verb
phrases, prepositional phrases, etc. Short idea units
were chosen in order to allow fine distinctions among the
number of idea units recalled.

Participants were assigned scores for reading

proficiency and vocabulary proficiency, based on the

56

number of correct responses to the reading and vocabulary

tests respectively.

Analyses

The effect of order of presentation of the passage
was checked, using analysis of variance with comprehension
measures as dependent variables and order and reading
comprehension as independent variables. Effects of
nationality were checked using measures of reading
comprehension as dependent variables and nationality and
reading comprehension as independent variables.
Differences between male and female respondents were
checked, using reading comprehension measures as dependent
variables and gender, vocabulary comprehension, and
reading comprehension as independent variables.

Intercoder reliability was evaluated for ratings of
background knowledge and reading comprehension.

In order to test the hypotheses, partial correlations
were calculated between measures of reading comprehension,
controlling for length of time spent in English-speaking
countries, reading proficiency and vocabulary proficiency
(H1). In order to get additional insight into the
contribution of prior knowledge to comprehension, in
relation to reading proficiency, two additional

calculations were performed. A partial correlation was

57

calculated for reading proficiency, controlling for prior
knowledge, vocabulary proficiency, and length of time in
English-speaking countries. Using a median split for
reading proficiency, partial correlations were calculated
between prior knowledge and comprehension, controlling for
length of time in English-speaking countries, vocabulary
proficiency, and reading proficiency.

Using a median split for background knowledge.
correlations between measures of reading comprehension and
reading proficiency were calculated and compared for
participants with high and low background knowledge (H2).
Using a median split for prior knowledge, correlations
between vocabulary proficiency and reading comprehension
were compared for readers with high and low proficiency

(R1).

Chapter III
RESULTS
Initial Analyses

Intercoder Reliability

Correlation coefficients were calculated among six
coders for each of the four types of coded responses
(degree of prior knowledge, number of idea units, quality
of title. and quality of statement of main idea). The
mean correlation coefficients were .87 for prior
knowledge. .92 for idea units, .88 for titles, and .91 for
main ideas. These correlations were used to correct for
attenuation due to measurement error.
Reliability of Maasuras

Chronbach’s alpha for the lexical proficiency
measure was .76. Chronbach’s alpha for the reading
proficiency measure was .67. These values were used to
correct for attenuation due to measurement error.

Background Knowledge Reguired for Comprehension

In order to compare the amount of background

 

knowledge required to comprehend the three passages,

fifty-three native English speakers, familiar with one of
the three topics, were given one of the passages and were
asked to list the pieces of background knowledge necessary

to comprehend the passages. Raters listed more pieces of

58

59

background information as necessary to comprehend the Gaga
wiih the Mind passage than the wedding passage (R = 3.87
for Gone with iha Uind passage: R = 1.53 for wedding
passage: t = 3.30, df = 30, p < .05, r = .52).
Significantly more pieces of information were listed for
the Gone with the Mind passage than for the baseball
passage (R = 3.87 for Gone with the Uind passage: R = 0.48
for baseball passage: t = 5.88. df = 34, p < .05,

r = .71). A significant difference was also found between
the number of pieces of information listed for the wedding
and the baseball passage (R = 1.53 for wedding passage:

R = 0.48 for baseball passage: t = 2.70, df = 36, p < .05,
r = .41).

Consistency Checks

anar of Prssentation of Passages

Since the order in which the passages were presented
to participants was counterbalanced, three 2 x 2 x 3
analyses of variance were calculated to determine whether
the order of presentation affected responses. The order
of presentation, reading proficiency, and prior knowledge
(using median splits for the latter two variables) were
used as independent variables: each of the three measures
of comprehension recalled were used as the dependent

variables. The results for the main effects for order

60

appear in Table 1.

 

 

Reading Passage
(dependent F df p n 2
M113)
Gone with
ihe Uind
(idea units) 1.851 2.95 > .05 .04
(title) 1.778 2.95 > .05 .04
(main idea) 2.794 2.95 > .05 .06
Uedding
(idea units) 0.971 2.95 > .05 .02
(title) 1.096 2.95 > .05 .03
(main idea) 1.175 2.95 > .05 .03
Baseball
(idea units) 1.431 2.95 > .05 .03
(title) 0.185 2.95 > .05 .00
(main idea) 0.581 2.95 > .05 .01

 

Table 1--Order Effects

 

No main effects were found for the order of
presentation of the passages. Of four possible
interaction effects (three two-way interactions and one
three-way interaction), one interaction effect was found.
An interaction between prior knowledge of nge with iha
.Giag and the order of presentation was found (F = 150.74,
df = 2.95, p < .05. n2 = .07) when number of idea units

was the dependent variable. Cell means appear in Table 2.

61

Higher numbers indicate more idea units recalled. Cell
means indicated that, for participants with low prior
knowledge of Gone with the Uind, fewer idea units were

Prior Knowledge

 

 

 

Low 1 High
1 l 10.81 i 9.18 l
Form 1 l 1
2 l 6.85 l 10.96 1
3 l 8.93 l 10.37 i

 

Table 2--Cell means with number of idea units recalled
as DOV.

recalled when that passage was presented first. However.
since there was only one interaction effect, and since the
effect size was small, order was not treated as a
variable, and all of the responses to each passage were
combined, regardless of whether the passage was presented
first, second, or third. It did not make a difference in
the results of the measures of comprehension whether a
particular passage was presented first, second, or third.
Knowing what the tasks (recalling pieces of information
from the passage, suggesting a title, and stating the main

idea) would be did not greatly help participants do better

62

on those tasks.
Gander of Participaai

In order to see whether the gender of the participant
made a difference in the responses to the tasks. 2 x 2 x 2
ANOVAs were calculated. using each of the three measures of
reading comprehension as the dependent variables and
gender, reading proficiency, and prior knowledge as
independent variables. with median splits for the latter

two variables. The results for main effects for gender

appear in Table 3.

 

Reading Passage 2
(dependent F df p I)
variable)

Gone gith

the Uind

(idea units) 2.273 1.95 > .05 .03

(title) 0.506 1.95 > .05 .00

(main idea) 0.004 1.95 > .05 .00
wedding

(idea units) 1.915 1.95 > .05 .02

(title) 0.719 1.95 > .05 .01

(main idea) 1.691 1.95 > .05 .02
Baseball

(idea units) 0.865 1.95 > .05 .01

(title) 0.000 1.95 > .05 .00

(main idea) 0.340 1.95 > .05 .00

 

Table 3--Gender Effects

 

63

No main effects were found for gender. Also, no
interaction effects involving gender were found.
Therefore, gender was not considered as a variable and
results from male and female participants were combined.
‘flationajity of Particigant

In order to determine whether the nationality of the
participants made a difference in their responses, 2 x 2 x
2 ANOVAs were calculated for each passage. using each of
the three measures of reading comprehension as the
dependent variables and nationality, reading proficiency,
and prior knowledge, with median splits for the latter
two. as independent variables. The results for the main

effects for nationality appear in Table 4.

64

 

Reading Passage

 

(dependent F df p n 2
__¥§£l§hle)
M
the Mind
(idea units) 0.041 3.95 > .05 .00
(title) 0.155 3.95 > .05 .00
(main idea) 0.109 3.95 > .05 .00
Uedding
(idea units) 0.226 3.95 > .05 .00
(main idea) 0.203 3.95 > .05 .00
Baseball
(idea units) 1.597 3.95 > .05 .01
(title) 1.703 3.95 > .05 .02
(main idea) 2.063 3.95 > .05 .03

 

Table 4--Effects of Nationality

 

No main effects were found for nationality, and no
interaction effects were found involving nationality.
Therefore, nationality was not treated as a variable, and
responses from participants of different nationalities
were combined.
Pearson Correlations

Pearson correlations were calculated among the

variables of reading proficiency, lexical proficiency.

length of time in English-speaking countries, prior

65

knowledge. and the three measures of reading comprehension
(number of idea units recalled and ratings of the title
and statement of the main idea) for each reading passage.

The results appear in Tables 5, 6, and 7.

 

Title Idea Profic- Vocab- Prior Length

Units iency ulary Knowledge

Idea (.54)* (.33)* (.34)* (.47)* (.42)* (.32)*
Tltie 022* 018* 018* 029* 020*
(.24)* (.22)* (.22)* (.33)* (.21)*

Idea Units .22* .32* .21* .13
(.28)* (.38)* (.24)* (.14)

Proficiency .48* .10 .32*
(.67)* (.11) (.39)*

Vocabulary .05 .37*
(.05) (.43)*

Prior Knowledge .04
(.04)

 

Table 5--Pearson correlations among variables (Gone with
the Uind passage)
( ) = corrected for attenuation
p < .05
96

11
N

 

66

 

Title Idea Profic- Vocab- Prior Length

 

Units iency ulary Knowledge
Main .34* .37* .10 .25* .27* .16
Idea (.38)* (.41)* (.13) (.30)* (.31)* (.17)
Title 000 “017* 003 019* '004
(.00) (-.22)* (.03) (.22)* (-.04)
Idea Units .33* .50* .20* .24*
(.41)* (.60)* (.22)* (.25)*
Proficiency .48* .05 .32*
(.67)* (.05) (.39)*
Vocabulary -.04 .37*
(-.04) (.43):
Prior Knowledge .10
(.11)
Table 6--Pearson correlations among variables (wedding
passage)
( ) = corrected for attenuation
* = P < 005
N = 96

 

67

 

Title Idea Profic- Vocab- Prior Length

 

Units iency ulary Knowledge
Main 036‘l 039* 018* 026* 017 009
Idea (.40)* (.43)* (.23)* (.31)* (.19) (.09)
Title .18* .12 .12 .10 .02
(.20)* (.16) (.15) (.11) (.02)
Idea Units 024*l 019* 021* 011
(.30)* (.23)* (.24)* (.11)
Proficiency .48* .28* .32*
(.67)* (.30)* (.39)*
Vocabulary .19* .37*
(.20)* (.43)*
Prior Knowledge .18*
(.19)*
Table 7--Pearson correlations among variables (baseball
passage)
( ) = corrected for attenuation
* = P < 005
N = 96

 

For all three passages, significant correlations were
found between prior knowledge and the number of idea units
recalled. For the Gone with the Uind and wedding
passages, there were significant correlations between
prior knowledge and the statement of the main idea. The
correlations among the main idea, title, and number of
idea units were significant, except for the correlation
between the title and the number of idea units for the

wedding passage. Both lexical and reading proficiency

68

were significantly correlated with the length of time
participants had spent in English-speaking countries.
Prior knowledge of baseball was significantly correlated
with length of time in English-speaking countries, but
prior knowledge of American weddings and nge with the
Giaa was not.

Hypotheses

Hyeoihesis 1

Prior Knowledge and Comprghension. The first
hypothesis stated that there would be a correlation
between prior knowledge and reading comprehension. In
order to test this hypothesis, partial correlations
between prior knowledge and each of the three measures of
reading comprehension were calculated. These partial
correlations appear in Tables 8, 9, and 10. These are
third order partials, controlling for reading proficiency,
lexical proficiency. and length of time in English-

speaking countries.

69

 

Title Main Idea
Idea Units

Prior .28* .37* .20*
KnOUIedge (032)* (042)* (022)*

 

Table 8--Third order partial correlations, controlling
for reading proficiency. lexical proficiency,
and length of time in English-speaking countries
(Gone with the Mind passage)
( ) = corrected for attenuation

 

 

* = P < 005
N = 96
Title Main Idea
Idea Units
Prior .20* .30* .29*
Knowledge (.23)* (.34)* (.33)*

 

Table 9--Third order partial correlations, controlling
for reading proficiency, lexical proficiency,
and length of time in English-speaking countries
(wedding passage)

 

 

( ) = corrected for attenuation
* = P < 005
N = 96
Title Main Idea
Idea Units
Prior .08 .13 .17
Knowledge (.09) (.15) (.19)

 

Table 10--Third order partial correlations, controlling
for reading proficiency, lexical proficiency.
and length of time in English-speaking countries
(baseball passage)
( ) = corrected for attenuation
* = p < .05
N = 96

 

For both the Gone with the Uind passage and the

wedding passage, third order partial correlations between

70

prior knowledge and all three measures of comprehension,
controlling for reading proficiency. lexical

proficiency, and length of time in English-speaking
countries, were significant. However, none of these
partial correlations were significant for the baseball
passage. Apparently, the significant Pearson correlations
were spurious for the baseball passage and can be
attributed to the correlation that both have with reading
and lexical proficiency. One possible explanation for

the lack of significant correlations is that, since the
baseball passage required little prior knowledge for
comprehension, prior knowledge was of little assistance in
comprehension, and, conversely, lack of prior knowledge
was little impediment to comprehension.

These results support the hypothesis for passages
requiring a degree of prior knowledge for comprehension.
However, for the passage that requires the least
background knowledge, the hypothesis is not supported.

flaaian Split for Prior Knowledg . In order to
further assess the effect of prior knowledge for readers
of different levels of proficiency, partial correlations
between prior knowledge and reading comprehension were
calculated using median splits for proficiency,

controlling for reading proficiency. lexical

71

proficiency, and length of time in English-speaking

countries. The results appear in Tables 11. 12, and 13.

 

Title Main Idea
Idea Units
Prior .10 .18 .07
Knowledge (.11) (.20) (.08)
(High Proficiency [N = 49))
Prior .38* .51* .41*
Knowledge (.43)* (.58)* (.46)*

(Low Proficiency [N = 47])

 

Table 11--Third order partial correlations, controlling
for reading proficiency, lexical proficiency.
and length of time in English-speaking countries.
with median split according to prior knowledge
(G one with the wing passage)
) = corrected for attenuation
* = p < .05

 

For the Gone with the Uind passage, for participants
with high proficiency. none of the correlations were
significant. For participants with low proficiency, all
of the correlations were significant. The difference
between the correlation between prior knowledge and the
main idea and for idea units were significant at the .05
level. As discussed above, the Gone wiih iha Uind passage
requires the most prior knowledge for comprehension. The
results of these calculations indicate that readers with
high proficiency are able to overcome, to some degree,
their lack of prior knowledge. However, readers with low
proficiency appear to depend on their prior knowledge to

overcome deficits in their reading proficiency.

72

 

Title Main Idea
Idea Units
Prior .17 .18 .31*
Knowledge (.20) (.20) (.35):
(High Proficiency EN = 49])
Prior .21 .48* .22
Knowledge (.24) (.55)* (.25)

(Low Proficiency [N = 47])

 

Table 12--Third order partial correlations, controlling
for reading proficiency, lexical proficiency,
and length of time in English-speaking countries,
with median split according to prior knowledge
(wedding passage)
( ) = corrected for attenuation
*=P<005

 

For the wedding passage, the correlations between
prior knowledge and idea units for participants with high
proficiency and between prior knowledge and the main idea
for participants with low proficiency were significant.
The difference between correlations with the main idea was
significant at the .05 level. The interpretation of these
results is less clear than for the Gone with the Uind
passage. The wedding passage required less prior
knowledge than the Gone with iha Uing passage. In this
case. participants with higher proficiency were more able
to make use of their prior knowledge on the idea units
measure of comprehension than the participants with lower
proficiency. However, in the case of the main idea,

participants with lower reading proficiency were better

73

able to make use of their prior knowledge.

 

Title Main Idea

Idea Units

Prior .13 .04 .17

Knowledge (.15) (.05) (.19)
(High Proficiency EN = 49])

Prior .02 .18 .10

Knowledge (.02) (.20) (.11)

(Low Proficiency [N = 47])

 

Table 13--Third order partial correlations, controlling
for reading proficiency. lexical proficiency,
and length of time in English-speaking countries.
with median split according to prior knowledge
(baseball passage)
( ) = corrected for attenuation
* = p < .05

 

For the baseball passage, none of the correlations
were significant. None of the differences between
correlations were significant. Of the three reading
passages, this one required the least background
knowledge. In this case. at least, participants were not
helped by their prior knowledge in reading a passage where
little prior knowledge was required for comprehension.

Another possible explanation is that the information
in the baseball passage was so basic that it was within
the range of knowledge of almost all of the participants,
although the other two passages included information that
was outside of the range of prior knowledge of some of the
participants. In other words, the baseball passage may

have included only "old" information for most of the

74

participants. while, in contrast, the wedding and Gaga
with the wind passages may have contained mostly new
information for some participants. some new and some old
information for some participants, and mostly old
information for some participants. If so, it would not be
unexpected that prior knowledge would affect comprehension
differentially in the latter case but not in the former
case.

Proficiancy and Comprehension. In order to see what
contribution reading proficiency makes. in comparison to
prior knowledge. partial correlations were calculated
between proficiency and the three measures of reading
comprehension. These controlled for prior knowledge.
lexical proficiency, and length of time in English-

speaking countries. The results appear in Tables 14, 15,

 

and 16.
Title Main Idea
Idea Units
Proficiency .06 .06 .06
(.06) (.06) (.06)

 

Table 14--Third order partial correlations. controlling
for prior knowledge, lexical proficiency, and
length of time in English-speaking countries (Gane
with ihe Uind passage)

( ) = corrected for attenuation
* p < .05
N 96

 

75

 

Title Main Idea

Idea Units

Proficiency -.19* -.03 .08
(-020)* (-003) (008)

 

Table 15--Third order partial correlations, controlling
for prior knowledge, lexical proficiency, and
length of time in English-speaking countries (wedding

 

 

passage)
( ) = corrected for attenuation
*=P<005
N = 96
Title Main Idea
Idea Units
Proficiency .06 .04 .09
(.06) (.04) (.09)

 

Table 16--Third order partial correlations, controlling
for prior knowledge, lexical proficiency, and
length of time in English-speaking countries
(baseball passage)

( ) = corrected for attenuation
* P < 005
N 96

 

Except for the negative correlation between
proficiency and the title for the wedding passage, none of
the partial correlations between proficiency and the
measures of comprehension were significant. The partial
correlation between the title and proficiency was
negative. These results indicate that, for these
passages, participants’ reading proficiency is not related
to their ability to comprehend the reading passages. This

does not necessarily mean that reading proficiency is not

76

reading passages was within the range of reading
proficiency of most of the participants. More difficult
passages might have produced significant results for the
correlations between reading proficiency and
comprehension.
Hypothesis 2

Proficiancyiand Prior Knowledg . The second
hypothesis stated that readers with higher prior knowledge
would have lower correlations between reading
comprehension and reading proficiency than readers with
low prior knowledge. In order to test this hypothesis, a
median split was used to divide participants into high and
low proficiency groups. Partial correlations were then
calculated between proficiency and the measures of
comprehension for each group, controlling for prior
knowledge, lexical proficiency, and length of time in
English-speaking countries. The results appear in Tables

17. 18. and 19.

77

 

Title Main Idea

Idea Units

Proficiency .16 -.07 -.15

(High Prior (.17) (-.07) (-.16)
Knowledge EN = 50])

Proficiency .02 .13 .13

(LOU Prior (002) (014) (014)

Knowledge [N = 463)

 

Table 17--Third order partial correlations, controlling
for prior knowledge, lexical proficiency, and
length of time in English-speaking countries, with
median split according to prior knowledge (Gone with
the Uind passage)
( ) = corrected for attenuation
* = p < .05

 

For Gonevyith the wind. none of the partial
correlations between proficiency and the measures of
reading comprehension were significant. Differences

between the correlations were not significant.

 

Title Main Idea

Idea Units

Proficiency -.35* -.08 .06

(High Prior (-.37)* (-.08) (.06)
Knowledge [N = 48])

Proficiency -.11 .02 .13

(Low Prior (-.12) (.02) (.14)

Knowledge [N = 483)

 

Table 18--Third order partial correlations, controlling
for prior knowledge, lexical proficiency. and
length of time in English-speaking countries. with
median split according to prior knowledge (wedding

passage)
( ) = corrected for attenuation
# = p < .05

 

For the wedding passage, there was a significant

78

negative correlation between proficiency and title for
participants with high prior knowledge. None of the other
correlations were significant. There were no significant
differences between correlations for readers with high and

low knowledge.

 

Title Main Idea

Idea Units

Proficiency -.09 -.19 .09

(High PPIOI“ (-010) (’020) (009)
Knowledge [N = 52])

Proficiency .24 .24 .02

(Low Prior (.26) (.25) (.02)

Knowledge [N = 44])

 

Table 19--Third order partial correlations, controlling
for prior knowledge, lexical proficiency, and
length of time in English-speaking countries, with
median split according to prior knowledge (baseball

passage)
( ) = corrected for attenuation
*=P<005

 

None of the partial correlations were significant for
the baseball passage. The difference between the
correlations between proficiency and the main idea was
significant at the .05 level.

The hypothesis that participants with high prior
knowledge would have a lower correlation between
proficiency and comprehension than participants with low
prior knowledge was only partially supported by the
results of this study. Out of nine comparisons, two were

significant and in the direction predicted. Of the

79

remaining seven comparisons, five of the differences were
in the direction predicted. but the differences were not
significant. Uhile these results do not unequivocally
support the hypothesis, they do suggest that it is true.
One possible explanation for the weak support of the
hypothesis is the relatively small number of participants
in each group when a median split is used. Had a larger
number of participants been used, more of the differences
might have been significant. Uith this relatively small
number of participants, the difference would have to be

quite large to be significant.

Research Questions

Research Question 1

yaaabulapyiand Comprehansion. The first research
question asked about the relationships among prior
knowledge, reading comprehension and lexical proficiency.
Previous research has indicated that second language
readers depend on lexical knowledge to a greater extent
than on grammatical knowledge. If this is so, it would be
expected that participants with better vocabularies would
have better comprehension, regardless of their reading
proficiency. In order to test this, partial correlations
were calculated between lexical proficiency and the three

measures of comprehension, controlling for proficiency,

80

prior knowledge, and length of time in English-speaking

countries. The results appear in Tables 20, 21, and 22.

 

Title Main Idea

Idea Units

Lexical .07 .24* .24*
Proficiency (.07) (.25)* (.25)*

 

Table 20--Third order partial correlations. controlling
for prior knowledge, reading proficiency, and length
of time in English-speaking countries (Gone with iha
Uind passage)

 

 

( ) = corrected for attenuation
*=P<005
N = 96
Title Main Idea
Idea Units
Lexical .09 .21* .45*
Proficiency (.10) (.22)* (.47)#

 

Table 21--Third order partial correlations, controlling
for prior knowledge, reading proficiency, and length
of time in English-speaking countries (wedding

 

 

passage)
( ) = corrected for attenuation
*=P(005
N = 96
Title Main Idea
Idea Units
Lexical .04 .17 .10
Proficiency (.04) (.18) (.10)

 

Table 22--Third order partial correlations, controlling
for prior knowledge, reading proficiency, and length
of time in English-speaking countries (baseball

passage)
( ) = corrected for attenuation
* P<005

N 96

 

81

For the Gone pith the Uind and wedding passages.
there were significant partial correlations between
lexical proficiency and the main idea and between lexical
proficiency and idea units recalled. There were no
significant correlations for the baseball passage. These
results provide some support for the idea that second
language readers depend to a greater extent on lexical
knowledge than on grammatical knowledge (as measured by
the test of reading proficiency).

Uhile there are significant correlations between
comprehension and lexical proficiency for the Gone gith
the wind and wedding passages, there are no significant
correlations for the baseball passage. The most likely
explanation for this is that the vocabulary in the
baseball passage is easier than that in the other two
passages. If the lexical difficulty of the passage is
within the range of even the participants with lower
lexical proficiency, a significant correlation would not
be expected.

It is not clear why correlations between lexical
proficiency and the main idea were significant but
correlations between lexical proficiency and the title
were not. Presumably these two questions measure the same

thing--the extent to which the readers understand the

82

overall idea of the passage. They should both be
correlated with lexical proficiency. There are at
least two reasons why they might not be. One is that it
may be more difficult to express the overall idea of a
passage in the few words allowed by a title. The question
about the statement of the main event may have been easier
to respond to, since the participants were allowed up to
two sentences, rather than being limited to a phrase.

Another possible explanation is that participants
were sometimes unable to answer one or the other of the
questions about the title and main idea due to lack of
time. Though they were reminded to answer the last two
questions. some participants did not allow enough time to
answer both, and one could have been left blank and
received a rating of zero due to lack of time rather than
lack of comprehension. This would have caused the title
and main idea responses to be less well-correlated than
they should have been.

flaaian Splits for Prior Knowledg . If second
language readers depend on vocabulary for comprehension,
it would be expected that readers with high prior
knowledge would be able to make better use of their
lexical knowledge, because they should know what

relationships to expect among the words that they know.

83

In order to test this relationship, and to better
understand the relationship between prior knowledge.
lexical proficiency and reading comprehension,
participants were divided into high and low prior
knowledge groups using median splits. The results appear

in Tables 23, 24, and 25.

 

Title Main Idea

Idea Units

Lexical .05 .14 .32*
Proficiency (.05) (.15) (.33)*

(High Prior
Knowledge [N = 50])

Lexical .12 .31* .18
Proficiency (.13) (.33)* (.19)
(Low Prior
Knowledge [N = 46])

 

Table 23--Third order partial correlations, controlling
for prior knowledge, reading proficiency, and length
of time in English-speaking countries, with median
split according to prior knowledge (Gone pith ths
wind passage)

( ) = corrected for attenuation
* = p < .05

 

For the Gone with the Uind passage, for participants
with high prior knowledge. the correlation between idea
units and lexical proficiency was significant. For
readers with low prior knowledge, the correlation between
the main idea and lexical proficiency was significant.

Differences between correlations were not significant.

84

 

Title Main Idea

Idea Units

Lexical .32* .33* .56*
Proficiency (.34)* (.35)* (.58)*

(High Prior
Knowledge [N = 483)

LeXica] "006 008 031*
Proficiency (—.06) (.08) (.33)#
(Low Prior
Knowledge [N = 483)

 

Table 24—-Third order partial correlations. controlling
for prior knowledge, reading proficiency, and length
of time in English-speaking countries. with median
split according to prior knowledge (wedding passage)

( ) = corrected for attenuation

 

For the wedding passage, for participants with high
prior knowledge. correlations with all three measures of
prior knowledge were significant. For participants with
low prior knowledge, only the correlation with between
lexical proficiency and idea units was significant. The

differences between correlations were not significant.

85

 

Title Main Idea
Idea Units
Lexical .23 .290 .14
Proficiency (.24) (.31)* (.15)

(High Prior
Knowledge [N = 523)

Lexical -.22 .05 .11
Proficiency (-.23) (.05) (.11
(Low Prior
Knowledge [N

)
443)

 

Table 25--Third order partial correlations, controlling
for prior knowledge. reading proficiency, and
length of time in English-speaking countries, with
median split according to prior knowledge (baseball

passage)
( ) = corrected for attenuation
* = p < .05

 

For the baseball passage, for participants with high
prior knowledge, the correlation between proficiency and
the main idea was significant. None of the other
correlations were significant. The difference between
correlations between lexical knowledge and the title was
significant at the .05 level.

Again, the results do not unequivocally indicate that
participants with higher prior knowledge make better use
of their knowledge of vocabulary than participants with
lower prior knowledge. (This study was not, in any case,
designed to deal specifically with this question.)
However, there is some support for that position. Five of

the correlations were significant for readers with high

86

prior knowledge, as opposed to two for readers with low
prior knowledge. There was only one case in which a
correlation was significantly higher for participants with
high prior knowledge than with low prior knowledge.
However, of the remaining eight comparisons, six were in
the predicted direction, even if the difference was not
significant. This is an indication that participants with
high prior knowledge are better able to make use of their
lexical knowledge than participants with lower prior
knowledge.

A study designed to answer this particular question
might produce clearer results. For example, if
participants were tested on knowledge of specific
vocabulary words that appear in the passage. rather than
on their lexical knowledge in general, a clearer link
between knowledge of vocabulary might be able to be
established.

Also, with a larger number of participants. more
significant differences might be found. with this
relatively small number of participants, the difference
between two correlations would have to be very large to be

significant.

.‘_,_,. a-..- _

87

Research Question 2

Summary of Related Rasults. The second research
question was related to the differences in comprehension
between reading passages that require different amounts of
prior knowledge. No specific analyses were done to answer
this question. However, analyses done to test the
hypotheses and answer the other research question do
address the issues raised in the second research question.
These have been dealt with in various places in this
chapter. In summary, prior knowledge appears to have a
greater effect on comprehension when a reading passage
requires more background knowledge. This tendency is
especially marked when second language readers have lower
reading proficiency. In such cases, it appears that
readers use prior knowledge to compensate for their lack
of reading proficiency.

It was also speculated that lexical proficiency
and comprehension should be correlated, because previous
research has shown that second language readers tend to
depend on their lexical knowledge. This seems to be
particularly true for reading passages that require

prior knowledge.

88

Conclusion

The findings of this study supported Hypothesis 1,
which stated that there would be a correlation between
prior knowledge and reading comprehension, for reading
passages that required more background knowledge for
comprehension. Prior knowledge was a particularly strong
influence for readers with lower proficiency. Reading
proficiency, on the other hand, did not have a strong
influence on participants’ ability to understand these
passages.

The second hypothesis stated that readers with high
background knowledge would have a lower correlation
between reading comprehension and reading proficiency than
readers with low background knowledge. There was some
support for this hypothesis, though it could not be
unequivocally confirmed.

The first research question asked about the
relationships among prior knowledge, reading
comprehension, and lexical proficiency. The results
indicated that there was a relationship between lexical
proficiency and reading comprehension, especially for
readers with high prior knowledge, for readings that
required more prior knowledge. This indicates that second

language readers tend to depend more on their knowledge of

89

vocabulary than on grammatical knowledge.

The second research question asked about the
differences among passages that require different amounts
of prior knowledge. The results indicated that when a
reading requires more prior knowledge. prior knowledge has
a greater effect on comprehension. They also indicated
that differences in lexical proficiency were more

important for readings that required more prior knowledge.

Chapter IV

DISCUSSION

Introduction

The main purpose of this study was to look at the

influence that content schemata have on the reading
comprehension of readers of English as a second language,
specifically those from East Asia. A secondary purpose of
the study was to examine the relationship between
vocabulary and reading proficiency. In this chapter, I
will discuss the findings of this study in relation to
previous studies and the implications of this study for
the second language classroom. I will also discuss the
limitations of this study and make some suggestions for

future research.

Findings of the Study

Prior Knowledge and Comprehension

 

The maJor finding of this study was that second
language readers (within the limitations discussed in the
sections on limitations of this study) do make use of
their prior knowledge in reading a second language
passage. They appear to use background knowledge to
compensate for inadequacies in reading proficiency. The
findings of this study indicate that nonnative speakers

make use of their prior knowledge in comprehension, Just

90

91

as native speakers have been shown to do (e.g., Spiro,
1980; Taylor, 1979: Langer and Nicolich, 1981: Langer,
1984: Kintsch and Greene. 1978: and Hare. 1982). The
results of this study also support the findings of Johnson
(1981) and Nunan (1985), whose studies indicated that the
cultural background of a reading passage influenced the
comprehension of nonnative readers of English.

However, these results contradict the findings of a
maJor study by Carrell (1983b), who found no effects for
prior knowledge. The differences in the results of
Carrell’s study and this study may be explained in at
least two ways.

As mentioned in Chapter I, there may have been
problems with the operationalization of the familiar/novel
variable in Carrell’s study, so that the passage that was
assumed to be familiar in content may have been at least
partly novel. Also, the operationalization of prior
knowledge for these two studies was different. Carrell
used a dichotomous operationalization in which background
knowledge was manipulated by presenting or not presenting
a picture and title related to the passage. However,
Carrell did not use any manipulation checks, so it is not
clear whether there were actual differences in background

knowledge. If the participants did not understand the

92

significance of the picture and title, they would not have
had the background knowledge that the researcher assumed
that they did, and there would have been participants in
the background knowledge condition who did not have
background knowledge. In contrast, in this study,
participants’ background knowledge was measured by asking
them to write down what they know about various topics.
Four levels of background knowledge were identified from
these responses. The participants’ background knowledge
was measured rather than assumed. The difference in
operationalizing background knowledge may help explain the
differences between the results of these two studies.
Vocabulary aad Comprehension

Though the study was not specifically designed to
test the relationship between lexical knowledge and
comprehension, the study does support previous findings by
UliJn (1981, 1984) and Cziko (1980). It appears that
second language readers tend to depend on lexical
knowledge, in combination with their knowledge of the
relationships among lexical items specified by the
schemata that they have for those content areas.

A study designed to answer this particular question
should produce clearer results. For example, if

participants were tested on knowledge of the specific

93

lexical items that appear in the passage, rather than on
their lexical knowledge in general, a clearer link between
knowledge of vocabulary and comprehension might be able to

be established.

Implications for the
Second Language Classroom

This study has implications for the classroom in two
areas--teaching background knowledge and teaching reading
skills.
Teaching Reading Skills

The results of this study indicate that participants
were able to use background knowledge in reading
comprehension. However. they may not be using it as
effectively as they might be. Readers with low reading
proficiency seem to make better use of background
knowledge than readers with high proficiency, as if use of
background knowledge was a last resort when reading
proficiency failed. (In contrast, studies by Eamon [1978-
1979] and Taylor [1979] seemed to indicate that native
language readers with high proficiency made better use of
schemata than readers of low proficiency.) It appears
that second language readers would benefit from some
systematic teaching in the use of background knowledge in

comprehension in second language reading.

94

Though there is no direct empirical evidence on this
point, it is doubtful that the participants have been
taught in any systematic way to use background knowledge.
Foreign language classes in their native countries
generally use grammar-translation method, which does not
emphasize use of background knowledge. The reading
courses that these participants took at the English
Language Center sometimes made use of prereading
exercises, and the teachers of these classes are generally
aware of the importance of background knowledge. However,
there has been little systematic effort to make students
aware of how they use background knowledge outside of the
classroom.

Uhile it is easy to state that teaching students to
use background knowledge in comprehension is a good idea,
it is more difficult to know whether this is actually
effective and, if it is, how to go about doing it.

Carrell (1985) showed that teaching students about textual
schemata did improve their comprehension of reading
passages with those textual schemata. However, I could
not find any studies that dealt directly with teaching
students to use background knowledge. Though Hudson
(1982) showed that certain prereading activities (looking

at pictures related to the passage, discussing with other

95

students what they know about the topic, and privately
making predictions about what the passage will be about)
can improve comprehension, it is not known whether
students can apply similar techniques to independent
reading. Further research is needed on this topic.

Even if students can benefit from being taught to
better use background knowledge, it is not known how these
skills can best be taught. One possibility is to try to
teach students to independently use a technique similar to
the one that Hudson used. Again, further research is
necessary to identify other possible techniques and to
compare them to find out which is most effective.

A related area where further research is vital is
materials development. Because the methods that teachers
use depend heavily on the materials that they use,
teaching materials that help students learn to use their
prior knowledge need to be developed and tested in the
classroom.

Taaching Background Knowlegge

Stevens (1982), in a study using American high school
students reading in English, showed that teaching
background knowledge improves reading comprehension.
Though I did not find any studies on nonnative readers,

Carrell and Eisterhold (1983), Clarke and Silberstein

96

(1977). and Krashen (1981) advocate teaching background
knowledge in order to improve reading comprehension.

Since this study shows that readers do make use of
background knowledge for comprehension. it seems logical
to teach background knowledge in second language programs.
Krashen (1981) advocated what he called aarroy raading,
that is, reading within one topic area. One possibility
is to present reading passages in a graded order so that
students read those that require little background
knowledge first and those that require more background
knowledge later. The passages that are read first would
provide background knowledge for later passages.

If the teacher does not have the opportunity to do
narrow reading because the textbook used in the classroom
contains readings on a wide variety of topics, the teacher
should try to be aware of the background knowledge that
might be required in a particular reading. If the
students would not be expected to have that knowledge, it
should be included in the students’ prereading
preparation. However, knowing what background students
already have may be particularly difficult in classes with
students from various countries.

Background information can be presented in various

IJaYS, depending on the amount of information involved and

97

the complexity of the information. I do not know of any
research on how it is best to present such background
knowledge. However, based on my teaching experience, I
think that a large amount of complex information might be
best presented in written form. Simpler information could
probably be presented orally. If some of the students
have more information on a topic than others, the

information might be brought out through class discussion.

Limitations of the Study

One limitation of this study is that it takes place
in a testing situation. specifically a testing situation
that required recall as opposed to recognition. The word
"test” was avoided in hopes that participants would not
try to memorize the reading passages as they read.
However, since they were required to answer questions
about the reading passages. the testing situation was
unavoidable. It is uncertain whether the same results
would apply to other situations, for example, reading for
pleasure or reading instructions.

Also, as part of the testing situation. participants
were first asked to write down what they knew about the
topics of the reading. This may have the effect of
activating the schema. as Hudson’s (1982) prereading

activities did. However, this situation may not be

98

entirely lacking in external validity. Readers may be
reminded of what they know about a topic when they read
the title of a reading passage. even if they do not write
it down.

Another limitation is that the participants did the
tasks within a relatively short time after reading the
passage. we do not know how the passage of time might
have affected the results. However, previous studies with
native speakers have indicated that the effect of schema
on recall strengthens over time. For example, Pass and
Schumacher (1981) found that after 24 hours had elapsed,
the effects of schema on recall were even stronger than
they were immediately after reading the passage.

Graesser, Uoll, Kowalski. and Smith (1980) found that
after one week had elapsed, schema effects were stronger
than they were after thirty minutes had elapsed.
Therefore. it is possible that the effects found in this
study would have been even stronger after a period of time
had elapsed.

Another limitation on this study is that only
participants with relatively high proficiency in English
were used. Only students in the three highest levels (out
of five levels) at the English Language Center

participated. Most of the remainder of the participants

99

were regular university students. The mean number of
months that the participants had been in English-speaking
countries was eighteen. These participants had had
considerably more exposure to English, both in written and
spoken form. and had a higher proficiency in English. than
the average person in their native countries. Therefore,
these results can only be generalized to nonnative
speakers with relatively high levels of proficiency.

Another limitation is the fact that most of the
respondents were graduate students. Uhile their reading
skills in their native languages were not actually tested,
graduate students probably have above average reading
skills in their native languages, which they can apply to
reading in a second language. Readers with average or
below average reading skills in their native language
might not be as successful in using their prior knowledge
in reading in a second language.

Also, caution should be exercised in generalizing
these results beyond the national groups represented in
this study. Students from Taiwan, Japan, Korea, and China
were chosen because their native languages are not related
to English, they have little exposure to English in their
native countries, and, for the most part, they learned

English using the grammar-translation method. The results

100

may not generalize well to national groups with other
characteristics. For example, peoPle whose native

language is related to English may respond differently.

Suggestions for Future Research

The results of this study suggest a number of
directions for future study in the area of schema theory
and comprehension for nonnative speakers.
Reagersggith Logs; Proficiancy

As mentioned above, both the English proficiency of
the participants in this study and their reading
proficiency in their native languages were relatively
high. More studies should be done with participants of
lower reading proficiency in English and lower reading
proficiency in their native languages to see how well they
use prior knowledge in reading comprehension. One
difficulty with studying reading comprehension and prior
knowledge with readers of low reading proficiency is that
it is more difficult to measure their comprehension and
prior knowledge in English. Also, it would be difficult
to use the same measures across a broad range of reading
proficiency.
Lon it dinal t

One characteristic of this study, which it shares

with almost all studies of schema and reading

101

comprehension, is that it examines the relationships among
the variables of reading and vocabulary proficiency,
comprehension, and length of time in English-speaking
countries at one point in time. Using this method, it is
difficult to answer questions about how ability to use
background knowledge develops and changes over time. In
the future, it may be useful to study the process of
reading as proficiency develops over a period of time.
Uhile it may not be practical to follow participants’
progress over a period of years, it might be possible to
study them during, for example, the first several months
that they reside in an English-speaking country.
Textual ang Content Schemata

Another area of interest that has not been studied
empirically to any great extent is the interaction between
textual and content schemata. Many studies related to
schemata have dealt either with textual schemata or
content schemata, but not both. It would be useful to
know to what extent content schemata contribute to
comprehension of rhetorical patterns, and to what extent
information from rhetorical patterns contributes to

comprehension of content.

102

Classroom Applicaiion
As Carrell (1986) pointed out, little of the research

done on schema theory and second language reading has been
done on classroom issues or under classroom conditions.
As I discussed in the section on implications for the
classroom, more research is necessary in the classroom to
see whether teaching background knowledge and teaching
students to use background knowledge is helpful. One
particular area where more research is necessary is that
of materials development.
W

'flaasurement of Requinad Background Knowledg . The
method of measuring background knowledge required for
comprehension of a passage used in this study should be
compared with other methods of measuring background
knowledge. As mentioned in the first chapter. another
possible method would be to ask proficiency nonnative
readers with little knowledge of the topics of the

passages to rate them according to how difficult they

 

seemed.
Cpmparisons of Ranges of Background Knowledge and
Inf rm n 'n Passa e . In this study, an alternative

explanation for the lack of correlation between prior

knowledge and comprehension for the baseball reading was

103

that the information was basic enough that it was within
the range of knowledge of most of the participants. One
simple way to measure this would be to have raters
underline the pieces of information in the passage that
they already knew. The number of pieces of information
not underlined could be compared for different readings
and used as a variable in studies of prior knowledge and
comprehension.
Activation of Schamaia

Studies have been done on the activation of schemata
in reading, using native speakers (e.g., Anderson and
Pichert, 1978). but none have been done on nonnative
speakers. Using techniques that have been developed to
study reading in native speakers, it would be useful to
learn more about the activation of schemata in nonnative

speakers.

Conclusion
This study was designed to look at the relationships
between prior knowledge and comprehension for East Asian
nonnative speakers of English. The study showed
correlations between reading and comprehension for reading
passages that required background knowledge for
comprehension. There was indication that readers of low

proficiency, in particular, compensated for their lack of

104

proficiency by using their prior knowledge.

These conclusions led to suggestions for classroom
applications and future research. However, many issues in
the field of reading and schemata are unresolved,
particularly as they relate to second language readers.
This research will hopefully lead to improvements in the

teaching of reading in second language classrooms.

APPENDIX

APPENDIX
READING SURVEY

This is part of a study related to reading and recall.
First, you will be asked to write down three things you know
about each of a list of words. Then you will be asked to read
three short passages and answer questions about them. You should
not try to memorize the passages as you read them. Just read
them as you normally would.

As you finish with each page. follow the instructions at the
bottom. If you are instructed to go on to the next page, do so.
If you are instructed to stop, do not go on to the next page.

If you have any questions, ask the researcher or research
assistant.

Thank you for your help.

GO ON TO THE NEXT PAGE

105

106

PART I

The purpose of this section is to find out what you know about

various topics. Urite down three things you know about each of
the following topics. If you are not able to write down a full
sentence, use a phrase or a single word.

Goneagith the Mind
1.

2.

3.

Scarlett O’Hara
1.

2.

3.

American Civil Uar
1.

2.

3.

baseball
1.

2.

3.

(baseball) pitcher
1.

2.
GO ON

TO THE NEXT PAGE

3.

a run (in baseball)
1.

2.

3.

American wedding customs
1.

2.

3.

wedding vows
1.

2.

3.

wedding processional
1.

2.

STOP AND

107

UAIT FOR INSTRUCTIONS

108

PART IIA
Read the passage below. Read the passage as you would normally
read such a passage: try to understand the meaning, but do not
try to memorize.

A traditional American wedding usually lasts from twenty to
forty minutes. After the mother of the bride and the parents of
the groom are seated in the front pew of the church, there are
often one or more solos and a greeting by the minister. This is
followed by the processional, when the groom’s attendants, the
bride’s attendants. and. lastly. the bride. enter the sanctuary.
(The groom does not take part in the processional. He enters
from a side door.) The bride and groom exchange vows. It is
traditional to use the words, ”To have and to hold from this day
forward. t"or better, for worse, for richer, for poorer, in
sickness and in health, to love and to cherish, ’til death do us
part." However, in recent years, many couples have chosen to
write their own vows instead, using concepts that are more
personally meaningful to them. Following the vows, they exchange
rings, which are considered symbols of the vows. The newly
married couple leaves the sanctuary first. followed by their

attendants.

STOP AND UAIT FOR INSTRUCTIONS

109

For each item. choose the best answer and circle the letter of
that answer. Choose only one answer. Answer each question. even
if you are not sure of the answer.

1.

No one can survive for very long without water.
a. reproduce c. transcend
b. prosper d. exist

The assignment was to write a synopsis of our favorite novel.
a. evaluation c. critique
b. summary d. dramatization

Fashion modeling can be a lucrative business.
a. ludicrous c. profitable
b. laughable d. competitive

Swarms of locusts ravaged the crops.

a. raided c. flew over

b. landed on d. destroyed

He talked so fast that I couldn’t comprehend what he said.
a. hear c. understand

b. translate d. repeat

All of the tenants in the building complained about the lack
of hot water.

a. old people c. superintendents

b. landlords d. occupants

The royal nuptials captured the attention of the world.

a. attendance c. baptism
b. wedding d. event

STOP AND UAIT FOR INSTRUCTIONS

10.

110

Urite down fifteen pieces of information you remember from the
passage that you read. If you are not able to explain some-
thing that you remember in a full sentence, use a phrase or a
single word.

GO ON TO THE NEXT PAGE

111

11.

12.

13.

14.

15.

II. Uhat title would you suggest for this passage?

III. In one or two sentences, what was the main idea?

STOP AND UAIT FOR INSTRUCTIONS

112

PART 118
Read the passage below. Read the passage as you would normally
read such a passage: try to understand the meaning, but do not
try to memorize.

Margaret Mitchell once said that Gone gith the wind is about
the theme of survival. Because Scarlett has Gerald O’Hara’s
toughness, she will not be defeated by the cruel circumstances in
which she finds herself. She will make the best of it. Isn’t
the truth rather that the war and the breakdown of the old
plantation society liberate Scarlett? They enable her to do what
she could never have thought of doing in the pre-war plantation
society. She can live a life of her own, own property, go into
business, make money. Scarlett does not do what she does after
the fall of the South merely to make the best of a disadvantaged
situation. She does not wish to be a lady. If we compare
Scarlett of Tara with the Scarlett who married first Frank
Kennedy and then Rhett Butler, and is making money and thriving
in Reconstruction Atlanta, we can only recognize how far more
satisfactory she finds the postwar South than the old. It is
difficult to imagine what charms her life before the war would
ever have held for her. They could not have come close to

equalling the active life she leads in postwar Atlanta.

STOP AND UAIT FOR INSTRUCTIONS

10.

11.

12.

13.

14.

113

His tanacious personality made him top salesperson in the
company.

a. tenable c. persistent

b. explosive d. charming

Some voters are easily swayed by glib politicians.
a. smooth-speaking c. dishonest
b. handsome d. gray-haired

He had reached the zenith of his career.

a. ambition c. happiest moment

b. zeal d. highest point
Frequent minor ailments kept her home from work.
a. irritations c. sicknesses

b. children d. falls

Earticipants from 100 countries go to the Olympic Games.

a. people who represent c. people who take part
b. pe0ple who come d. people who are athletes

Her action infuriated him.

a. saddened c. angered

b. intoxicated d. frightened

City dwellers are exhilarated by country air.
a. amazed c. humbled

b. fanned d. stimulated

STOP AND UAIT FOR INSTRUCTIONS

114

I. Urite down fifteen pieces of information you remember from the
passage. If you are not able to explain something that you
remember in a full sentence, use a phrase or a single word.

3.

10.

GO ON TO THE NEXT PAGE

115

11.

12.

13.

14.

15.

II. Uhat title would you suggest for this passage?

III. In one or two sentences, what was the main idea?

STOP AND UAIT FOR INSTRUCTIONS

116

PART IIC
Read the passage below. Read the passage as you would normally
read such a passage: try to understand the meaning, but do not
try to memorize.

Baseball is a sport that is so popular in the United States
that it is often called the national pastime. Every spring and
summer, millions of people throughout the US play this game.
Millions also watch baseball games and closely follow the
progress of their favorite teams and players.

A baseball game is played on a large field between two teams
of 9 or 10 players each. The teams take turns at bat (on
offense) and in the field (on defense). A player of the team in
the field, called the pitcher, throws a baseball toward a player
of the team at bat, called the batter. The batter tries to hit
the ball with a bat and drive it out of the reach of the players
in the field. By hitting the ball, and in other ways, player can
advance around the four bases that lie on the field. A player
who does so scores a run. The team that scores the most runs

wins the game.

STOP AND UAIT FOR INSTRUCTIONS

15.

16.

17.

18.

19.

20.

117

They swept across Europe and rathlassly killed all in their
path.

a. without stopping c. without warning

b. without weapons d. without pity

It was inevitable that women would be sent into space along
with men.

a. unlikely c. influential

b. fantastic d. unavoidable

Americans were appalled by the latest statistics.
a. surprised c. dismayed
b. informed d. pleased

As she aged, she became more garrulou .
a. talkative c. sickly
b. gracious d. grey

Obviously Helen’s forte is chemistry.
a. fortitude c. weakness
b. talent d. fixation

 

The intarment took place last Friday.
a. festivity c. launching
b. installation d. burial

STOP AND UAIT FOR INSTRUCTIONS

118

I. Urite down fifteen pieces of information you remember from the
passage. If you are not able to explain something that you
remember in a full sentence, use a phrase or a single word.

10.

GO ON TO THE NEXT PAGE

119

11.

12.

13.

14.

15.

11. what title would you suggest for this passage?

III. In one or two sentences, what was the main idea?

STOP AND UAIT FOR INSTRUCTIONS

120

PART 111

Read the passage that follows. For each blank, select one choice
that is most natural and accurate, and circle that word. In
order to determine the correct answer, you may need to read ahead
a sentence or two or go back a sentence or two and re-read. If
you are not sure of the correct answer, try to make a good guess.
EXAMPLE:

Students from all over the
world go to universities for
advanced courses in their fields
of study. Most of these _i__ are 1. courses they students college
intelligent and eager to study.

ELEANOR ROOSEVELT

Eleanor Roosevelt was born in
New York City on October 11, 1884.
She had a rather unhappy
childhood. Her parents died, and
she lived with her grandmother, who
was _2__ strict and stern. 2. because some quite little
Eleanor had_;;__ friends her 3. from one no none
own age. Uhen she _§__ fifteen, 4. should was even are
her grandmother sent her to
Allenswood, _§__ school in 5. went summer the a
England. Eleanor worked hard
_G__ was soon ranked among the 6. for and because she
best _Z__ at Allenswood. 7. lovely teachers for students

During _G__ summer of 1902, 8. here the when lovely

Eleanor returned to New York.

Three years 9 , she married 9. there later ago from
GO ON TO THE NEXT PAGE

121

her cousin. Franklin D. Roosevelt.

In 1928. he was _19_ 10.
Governor of New York. During
their _ii_ in the Governor’s 11.
mansion, Eleanor often _i2_ 12.

speeches for her husband, and she

13 to help the unemployed 13.

obtain food and Jobs.

14 her husband was 14.

elected President of the United

States, Eleanor moved her 15 15.

into the Uhite House and took

16 the duties of First Lady. 16.

She became widely known through

her 17 and writing. 17.

Eleanor might have retired

from public 18 after her 18.

husband’s death on April 12, 1945,

19 she continued to speak and 19.

give 29 to others through her 20.

many activities.

when Eleanor Roosevelt died

on November 7, 1962, the 21 of 21.

the world mourned their loss. Her
concern for the welfare of all
people had earned her the title

"First Lady of the Uorld."

realized seemed elected sentence

son minute hello years

giving made carefully said

tried watch had been living

Therefore Uhen More Constantly

since sale family step

on similar very for

superb sing lectures TV

since because life help

also but going famous

request infer inspiration offer
people quality leader quite

GO ON TO THE NEXT PAGE

122

PART IV

Please answer the following questions.

1.

 

Are you _____?
a. male b. female

what country are you from?

a. PRC d. Japan
b. Korea e. the US
c. Taiwan f. other (specify)

 

Are you a student?

a. yes b. no

If you are an English Language Center or Haslett Adult
Education student, what level are you at?

If you are a regular university student, what is your maJor?

If you are not a student, what are you doing?

How old are you?

a. 17 to 19 years old e. 29 to 31 years old
b. 20 to 22 years old f. 32 to 34 years old
c. 23 to 25 years old 9. 35 to 37 years old
d. 26 to 28 years old h. more than 38 years old

How long have you been in the US or another English-speaking
country?

years and months

How long have you studied English (both in the US and your
native country)?

a. less than 6 years d. 8.5 to 9.5 Years

b. 6 to 7.5 Years e. more than 9.5 Years

c. 7.5 to 8.5 years

Are you ?

a. an undergraduate student
b. a graduate student

LIST OF REFERENCES

LIST OF REFERENCES

Abelson, R.P. (1981). Psychological status of the script
concept. American Psychologis , Ga, 715-729.

Anderson. R.C. (1978). Schema-directed processes in
language comprehension. In A.M. Lesgold, J.U.
Pellegrino, S.D. Fokkema, and R. Glaser (Eds.),
CognitivaiPsychology and Instruction. New York: Plenum
Press.

Anderson, R.C., and Pichert, J.U. (1978). Recall of
previously unrecallable information following a shift

in perspective. Journal of Verbal Learning and Verbal
Behavior, i2, 1-12.

Anderson, R.C., Reynolds, R.E., Schallert. D.L.. and
Geotz, E.T. (1977). Frameworks for comprehending
discourse. American Educational Research Journal, i5,
367-382.

 

Anderson. R.C., Spiro. R.J. and Anderson. M.C. (1978).
Schemata as scaffolding for the representation of
information in connected discourse. American
Educational Research Journal, la, 433-440.

 

Bartlett, F. C. (1932). Remembering. Cambridge:
Cambridge University Press.

Bower. G.H., Black. J.B. and Turner, T. (1979). Scripts
in memory for text. Cognitive Psychology, ii, 177-220.

Bransford, J.O., and Johnson, M.K. (1972). Contextual
prerequisites for understanding: Some investigations
of comprehension and recall. Journal of Verbal

Learning and Verbal Behavior, ii, 717-726.

Bransford, J.O., and Johnson, M.K. (1973).
Considerations of some problems of comprehension. In

U.G. Chase (Ed.), V'sdal information processing. New
York: Academic Press.

Carrell, P.L. (1983a). Some issues in studying the role of
schemata, or background knowledge, in second language

comprehension. Raading in a Fareign Langaage, 1(2),
81-910

123

 

124

. (1983b). Three components of background knowledge

in reading comprehension. Language Learning, GG, 183-
207.

. (1983c). Background knowledge in second language

comprehension. Language Learning and Communicatian, 2.
25-33 0

. (1984). Schema theory and ESL reading: Classroom

 

implications and applications. Ina Modarn Language
Journal, GG, 332-343.

. (1985). Facilitating ESL reading by teaching text
structure. TESOL Quarterly, 12, 727-751.

. (1986). Second language reading pedagogy in the
light of schema theory. Paper presented at the 6th
Annual Midwest Regional TESOL Conference, Ann Arbor,
MI, November 6-8, 1986.

Carrell, P.L. and Eisterhold, J.C. (1983). Schema theory
and ESL reading pedagogy. TESOL Quarterly, i2, 553-
573.

Cziko, G.A. (1980). Language competence and reading
strategies: A comparison of first-and second-language

oral reading errors. Language Learning, GG, 101-116.

Den Uyl. M., and van Oostemdorp, H. (1980). The use of
scripts in text comprehension. Poetics, 9, 275-294.

Eamon. D.B. (1978-1979). Selection and recall of topical
information in prose by better and poorer readers.
Reading Resaarch Quarterly, id, 244-257.

Fass, U. and Schumacher, G.M. (1981). Schema theory and
prose retention: Boundary conditions for encoding and

retrieval effects. Qisgourse Proaesses. A, 17-26.

Goodman, K.S. (1967). Reading: A psycholinguistic
guessing game. ournal f h Rea n S alist, a.
126-135.

Goodman, K.S. (1970). Behind the eye: what happens in
reading. In K.S. Goodman and 0.8. Niles (Eds.),

Reading: Pragess and Program. Urbana, IL: National

Council of Teachers of English.

125

Gough, P.B. (1972). One second of reading. In J.F.
Kavanagh and 1.6. Mattingly (Eds.), Langdage by Eye and
Ea . Cambridge, MA: MIT Press.

Graesser, A.C., Hall, 8., Kowalski. D.J., and Smith, O.A.
(1980). Memory for typical and atypical actions in
scripted activities. Journal of Experimental
Esychology: Human Laarnin and Memor , a, 503-515.

Haberlandt, K., Berian, C., and Sanderson, J. (1980).
The episode schema in story processing. Journal of
Verbal Learning and Varbal Behavior, i2, 635-650.

Hare, V.C. (1982). Preassessment of topical knowledge:
A validation and extension. Journai of Raading
Behavior, 15, 77-85.

Heaton, J.B. (1975). writing English language test .
Essex. England: Longman.

Hudson, T. (1982). The effects of induced schemata on
the ”short circuit“ in L2 reading: Non-decoding
factors in L2 reading performance. Language Laarning.
329 1-310

Iser. U. (1978). The act of reading. Baltimore: The

Johns Hopkins University Press.

Jackson. H. (1947). The reading of books. New York:
Charles Schribner’s Sons.

Johnson, P. (1981). Effects on reading comprehension of
language complexity and cultural background of a text.

TESOL Quarterly, 1;, 169-181.

Kant, 1. (1781/1963). Criiigue of pure reason. London:
MacMillan. (Translated by N. Kemp Smith).

Kintsch, U., and Greene. E. (1978). The role of culture-
specific schemata in the comprehension and recall of

stories. Discourse Processes, i, 1-13.

Kitao, K., Brodrick, V.. Fujiwara, B.. Inoue, M., Kitao,
S.K., and Sackett, L. (1983). American samplar:

Acguiring cultural awaraness and raading skills.
Reading, MA: Addison-Uesley.

126

Kolers, P.A. (1970). Three stages in reading. In H.
Levin and J.T. Uilliams (Eds.), Basic Studies in
Reading. New York: Basic Books.

Kozminsky, E. (1977). Altering comprehension: The effect
of biasing titles on text comprehension. Memory and
Cpgnition, G, 482-490.

Krashen, S. (1981). The case for narrow reading. TESOL
Newsletter, i§(6), 23.

 

LaBerge, D., and Samuels, S.J. (1974). Toward a theory of
automatic information processing in reading. Cognitive
Psychology. é. 293-323.

Langer. J.A. (1980). Facilitating text processing: The
elaboration of prior knowledge. In J.A. Langer and
M.T. Smith-Burke (Eds.), Reader Meets Author/Bridging
the Gap (pp. 149-162). Newark: International Reading
Association.

. (1981). From theory to practice: A prereading
plan. Journal of Reading, 2G, 152-156.

. (1984). Examining background knowledge and text

comprehension. Reading Researgh Quarterly, 12, 468-
481.

Langer, J.A. and Nicolich, M. (1981). Prior knowledge

and its effect on comprehension. Journal of Reading
Behavior, 1G, 373-379.

Lubin, L., Jr. (1983). Scarlett O’Hara and the two
Quentin Compsons. In O.A. Pyron (Ed.), Recastings Gone
with the Uind in American cdltdre (pp. 81-103). Miami:
University Presses of Florida.

Meyer, D.E. and Schvaneveldt. R.U. (1971). Facilitation
in recognizing pairs of words: Evidence of a dependence
between retrieval operations. dadrnal of Experimental
Psychdlagy. 29. 227-234.

Nash-Ueber, B. (1975). The role of semantics in
automatic speech understanding. In 0.6. Bobrow and A.

Collins (Eds.), Raprasentation and Understanding. New

York: Academic Press.

127

Nunan, D. (1985). Content familiarity and the perception
of textual relationships in second language reading.

RGLC Journal. lé: 43-51.

Pearson, P.D. and Spiro. R. (1982). The new buzz word in
reading is schema. Instructor, 5;. 46-48.

Pichert, J.U. and Anderson, R.C. (1977). Taking
different perspectives on a story. Journal of

Edusational Psychology, G2, 309-315.

Reicher, G.M. (1969). Perceptual recognition as a
function of meaningfulness of stimulus material.
Journal of Exparimental Psychology, Gi, 274-280.

 

Reichler, J.L. (1987). Baseball. In yorld Book
Encyclopedia (pp. 92-96). Chicago: Uorld Book, Inc.

Rumelhart, D.E. (1982). Toward an interactive model of
reading. In S. Dornic (Ed.), Attantion and Performance
12° Hillsdale, NJ: Erlbaum.

Schank, R.C. (1973). Identification and conceptualiza-
tions underlying natural language. In R.C. Schank and
K.M. Colby (Eds.), ComputapaModals of Thought aad
Languag . San Francisco: Freeman.

Schank, R. C., 8 Abelson, R. P. (1977). Scripts. plans.
goals. and understanding; Hillsdale, NJ: Lawrence
Erlbaum Associates.

Smith, F. (1971). Understanding reading: A psycholinguis
tic analysis of reading_and laarning to read. New
York: Holt, Rinehart, and Uinston.

 

Smith, F. (1982). Undarstanding reading; A psycholigguis
iic analysis of reading and learning to read, third

edition. New York: Holt, Rinehart, and Uinston.

Spilich, G.J.. Besonder, G.T.. Chiesi, H.L., and Voss,
J.F. (1979). Text processing of domain-related in
formation for individuals with high and low domain

knowledge. Journal af Verbal Learning and Verbal
Behav. P, l§g 275-2900

Spiro, R.J. (1980). Prior knowledge and story
processing: Integration. selection, and variation.
PgetiCSg 20 313-3270

128

Steffensen, M.S., Joag-Dev C., and Anderson, R.C. (1979).
A cross-cultural perspective on reading comprehension.

Reading Rasaarch Gdariarly, 1;, 10-29.

Stevens, A.L. and Rumelhart. D.E. (1975). Errors in
reading: Analysis using an augmented network model of
grammar. In O.A. Norman, D.E. Rumelhart, and the LNR
Research Group (Eds.), Explaraiians in Cagnition. San

Francisco: Freeman.

Stevens, K.C. (1980). The effect of background knowledge
on the reading comprehension of ninth graders. Journal
of Reading Behavior. 12, 151-154.

Stevens, K.C. (1982). Can we improve reading by teaching
background information? dournal of Reading, 2G, 326-329.

Taylor. B.M. (1979). Good and poor readers’ recall of
familiar and unfamiliar text. Journal of Raading
Behavior, 2, 375-380.

Thomas, S.K. (1980). The effect of reader-authored
materials on the performance of beginning readers.
Unpublished doctoral dissertation, Michigan State
University.

UliJn, J.M. (1981). Conceptual and syntactic strategies in
reading in a foreign language. In E. Hopkins and R.
GrotJahn (Eds.), tu ' s in an ua e T chin and

Language Acguisiiion. Bochum: Brockmeyer.

UliJn, J.M. (1984). Foreign language (FL) reading:
Conceptual and syntactic strategies and their
consequences for the role of the native language (NL).

International Raviewiof Appliad Lingdisiics, 22. 71-73.

Ueber, R.M. (1970). First graders use of grammatical
context in reading. In H. Levin and J.T. Uilliams

(Eds.), Basic Gigdias in Raadi g. New York: Basic
Books.
Uilson. M. (1987). h m -ba tra e ies for ESL

reading. Unpublished manuscript, Michigan State
University. Department of English, East Lansing.

Yorio. C.A. (1971). Some sources of reading problems for

foreign language learners. Langdage Learning, 2;, 107-
115.