use of the cloze procedure to measure readability …
TRANSCRIPT
"NIVERSITY OF HA WAii LIBRAQ
USE OF THE CLOZE PROCEDURE
TO MEASURE READABILITY .Al~D READING COMPREHENSION
IN THE SAMOAN LANGUAGE
A THESIS SUBMITTED TO THE GRADUATE DIVISION OF THE UNIVERSITY OF HAWAII IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF ARTS
IN ENGLISH AS A SECOND LANGUAGE
DECEMBER 1981
By
Gary L. Bowne
Thesis Committee:
Charles Mason, Chairman Ted Plaister
Danny Steinberg
-
We certify that we have read this thesis and that in
our opinion it is satisfactory in scope and quality as a
thesis for the degree of Master of Arts in English as a
Second Language.
THESIS COMMITTEE
ii
--ACKNOWLEDGEMENTS
Grateful acknowledgement is given to the following
people for assistance with this project:
iii
The late Dr. Ruth Crymes, for encouragement in the early
stages.
Samoan students of Kaimuki High School in Honolulu, who
took a trial version of the multiple-choice reading test.
Staff and students of Leone High School in American
Samoa, for. continued support and cooperation.
Leonard Burns, for invaluable assistance with the computer
at the University of Hawaii.
Evelyn Nakanishi, Department of East Asian Languages at
the University of Hawaii, for expeditious typing.
iv
ABSTRACT
A series of cloze reading tests of varying deletion rates
and a traditional multiple-choice reading test wer~ adminis
tered to a group of high school students in American Samoa
in order to determine the efficacy of using the cloze
procedure as a measure of readability and comprehension in
the Samoan language.
Results indicated that there were strong correlations
of r = .80 or better between scores on the cloze reading
tests and the multiple-choice reading test. There was a
slight yet statistically significant difference between the
exact word (EW) and acceptable word (AW) scoring systems,
and a significant difference in mean scores across deletion
rates. This latter difference was largely due to a sizeable
drop in mean score between the five-to-one and four-to-one
deletion rates.
It, was concluded that cloze procedure could serve as
a measure of both readability and reading comprehension in
the Samoan language. The difference in scoring methods was
not considered crucial, thus confirming the practicality of
exact word scoring. Both four-to-one and five-to-one were
identified as functional deletion rates.
ACKNOWLEDGEMENTS
ABSTRACT . . . .
LIST OF TABLES
LIST OF FIGURE . .
THE PROBLEM
BACKGROUND .
TABLE OF CONTENTS
REVIEW OF THE LITERATURE
THE HYPOTHESES .
METHOD ....
Cloze reading tests Traditional reading test Test administration . Data analysis .
RESULTS
Hypothesis One Hypothesis Two Hypothesis Three
DISCUSSION . . . . .
Hypothesis One Hypothesis Two Hypothesis Three
CONCLUSIONS
IMPLICATIONS .
APPENDICES . .
BIBLIOGRAPHY
v
iii
iv
vi
vii
1
2
3
16
16
17 18 19 20
24
24 26 29
34
34 37 38
42
44
46
69
vi
LIST OF TABLES
Table 1. Means for Exact Word and Acceptable Synonym Scoring Method . . . . . . . 24
Table 2. Multiple-choice Reading Test Means 25
Table 3. Analysis of Variance on Multiple-choice Reading Test Scores . . . . . . . . 25
Table 4. Correlations between Multiple-choice Reading Test Scores and Cloze Deletion Rate Scores . 26
Table 5. Two-way ANOVA on Deletion Data ........ 27
Table 6. Correlations between Exact Word Scoring Method and Acceptable Synonym Scoring Method for each Deletion Rate . . . . . . . . . . . . . . 28
Table 7. Analysis of Variance and Trend Analyses on Deletion Rate--Exact Word Score Data . . . . . 30
Table 8. t-tests on Cloze Exact Word Scoring Data . 31
THE PROBLEM
Measuring readability is difficult even in a widely
studied language like English. And although there are a
variety of methods available, we have yet to determine the
best system. This thesis addresses the question of how well
current readability assessments used in English would function
in Samoan, a language not at all similar to English.
There is a need for an accurate measure of readability
for materials in the classroom in Samoa, but formal assess
ments do not often exist in languages with a limited number
of speakers. A more precise determination of readability
levels for materials under development would insure greater
sensitivity to student reading abilities, and would allow for
a more accurate gradation of materials.
Measuring comprehension of written Samoan is an equally
valid concern. Although an abundance of standardized instru
ments is available in English, no formal instrument exists
for measuring comprehension in Samoan. The preparation of
a more systematic method for measuring comprehension would
allow for the establishment of grade level expectations, the
creation of norms, and the development of an accurate system
of program evaluation.
Currently, no such formal network exists. While it is
true that there are casual and valuable assessments such as
teacher judgement, it seems worthwhile to consider the
development of a more practical and more "standardized"
system. The lack of any concrete method for measuring
2
readability and reading comprehension in Samoan prompted
a research question: what device could serve as an accurate
yet simple measure of both readability and reading comprehen
sion for texts written in Samoan?
BACKGROUND
American Samoa is an unincorporated U.S. territory in
the South Pacific. The Department of the Interior controls
funds for American Samoa and as a result, education is to
some extent influenced by the federal government. While
locally appointed school officials determine policy for the
most part, the viability of educational programs is often
determined by the availability of federal funds.
Public schools in American Samoa are modeled after the
American system, following a stateside curriculum pattern
and even a northern hemisphere calendar. The bulk of the
teaching materials and books come from the United States, as
do a fair number of contract teachers at both the elementary
and secondary level.
Students in Samoa are taught in the vernacular for the
first three years of school (a policy that is now being
implemented in American Samoa). Thus the introduction of
reading instruction will be in Samoan using Samoan materials.
But such materials are currently few in number and it will
take some time to develop a repertoire of materials for this
program. Measuring readability or comprehension must, of
necessity, be undertaken on the materials that are presently
available.
3
Although predominantly an oral culture, Samoa is
witnessing growth in printed materials in the native language.
Currently, there is little available in the form of
"literature" printed in Samoan, perhaps because the written
tradition is young by western standards. Another factor may
be that the spoken word still carries far more weight than
the written one in this island culture.
While most people in American Samoa can read in their
native language, they have a limited selection when they
choose to do so. There are several dictionaries, a grammar,
and basal readers developed by the Bilingual Education
project of the Department of Education in American Samoa. A
few stories have been translated from English into Samoan
and published in New Zealand. By far the bulk of the
written literature is composed of Bible stories and Biblical
tracts in Samoan (Marsack 1962) .
Th.is situation translates to a system of informal
assessment in the native language. With no substantial
supply of reading materials available, there has been no rush
to create formal instruments in Samoan to measure language
ability, comprehension, or the reading materials themselves.
But with renewed emphasis on the teaching of Samoan in the
primary grades, a more precise system of measurement is
indicated.
REVIEW OF THE LITERATURE
Studies of readability began in earnest in the early .
part of this century. While there are numerous factors
4
involved in measuring readability, most studies address three
things: legibility, reader interest, and ease of under
standing. The last factor was probably given the most
attention in the majority of studies.
The recent history of readability studies is basically
one of formulas used to predict reading levels. Chall (1958)
reports that there were twenty-nine quantitative studies of
readability from 1922 through 1953, and the "typical product
of each study was a readability formula." From about 1958
to the present, there has been a decline in the use of read
ability formulas but an increase in the use of such alternative
measures as the cloze procedure.
This thesis will not examine and define all of the
formulas devised in the past fifty years. For those who wish
to ,trace the development and application of these formulas,
see Klare (1963) and Gilliland (1972). It might, however,
prove useful to look at the characteristics of the readability
formulas that emerged during this period.
Each study usually resulted in a formula "based on the
counting and weighting of several significant factors in the
printed matter that would be used to predict the reading skill
needed to understand it" (Chall 1958). Each formula
typically involved a regression equation derived from a
combination of factors which had the highest multiple
correlation with criterion passages.
Although the formulas each weighted different combina
tions of factors, most were similar in that only a few factors
5
were used as predictors. Ultimately, most of these formulas
had two things in common--namely, the counting and weighting
of vocabulary difference and sentence length. The commonality
of these two factors provided a thread of continuity over
the years.
The late forties and early fifties was the last time
that any significant work was carried out in this area.
Studies by Lorge (1939), Dale-Chall (1948), Flesch (1951),
and Spache (1953) all produced formulas which differed
slightly but which shared the common features of vocabulary
difference and sentence length.
A significant shift in readability studies came about in
1953 with the formulation of the "cloze" procedure by Wilson
Taylor. His work with Gestalt closure theory represented a
significant shift away from the formula school. Taylor's
research produced results that correlated well with existing
formulas for establishing readability levels; moreover, his
was the first such "formula'1 which took the language abilities
of the reader into account.
With the advent of the cloze procedure, research with
readability formulas almost ceased. Predictive formulas
lost some of their universal appeal, even though they still
found applications. The subsequent appearance of psycho
linguistic theory and its affinity for cloze research ended
the long reign of readability formulas, with very few
exceptions.
6
One of these was Edward Fry (1968) who proposed a
simple formula which took two factors into account, average
word length and average sentence length. His formula was
one of very few to appear at this time, but it has enjoyed
some popularity because of its ease of use. The formula
involves the use of a graph, thereby eliminating complicated
calculations. The Fry formula provides a quick and easy
estimate of readability, and it correlates well with previous
formulas such as Dale-Chall and Flesch. The Fry formula
stands out as a rare exception in a forest of cloze studies
on readability.
Additional research with the cloze procedure by Taylor
(1957) indicated that cloze was more sensitive to the
nuances of style than any predictive formula. Bormuth
(1962) provided additional evidence to support the work by
Taylor, and Weintraub (1968) found cloze to accurately
measure.the readability of English prose (passage difficulty)
better than any current formula. The sensitivity of cloze
to passage difficulty seems well documented.
Whereas readability studies focused primarily on the
use of predictive formulas, the study of reading comprehen
sion has for the most part dealt with traditional multiple
choice tests. Cleland (1966) in a review of the history of
reading comprehension, states that the term "comprehension"
did not appear in the literature until 1917.
The study of comprehension has taken the form of
multiple-choice tests which generally measure the effects
7
of comprehension after a person has read a passage. In the
past, the distinction between readability and comprehension
was that readability formulas were used to predict compre
hensibility, while reading comprehension typically involved
questions to measure understanding after the fact. The first
did not involve the reader; the latter of course did.
There were no radical developments in the study of
reading comprehension from the twenties to the forties.
Research continued with traditional methods, just as the
measurement of readability at this time continued with little
change in methodology. However, as readability formulas
reached their zenith in the forties, so too the study of
comprehension changed at that time.
The emergence of psycholosical testing stimulated the
development of the statistical technique referred to as
factor analysis. This involves identifying the common
elemen~s that account for an individual's score on a given
test, and then attempting to evaluate each of the elements
to determine the relative importance of each variable.
Davis (1944) conducted a factor analysis with proficient
readers in an attempt to identify the crucial elements that
influence reading comprehension. His study produced results
which indicated that comprehension in reading was a skill
composed primarily of two factors, the understanding of word
meaning and the ability to reason verbally.
In a later study, Davis (1972) found that these same
two factors accounted for eighty-nine percent of the variance
8
in reading comprehension. At the same time he listed several
additional factors which relate to cloze procedure: concen
tration on the literal meaning of a passage, following
structure, and recognizing the mood and literary techniques
of the author. Other factor analyses have revealed different
variables but the two primary factors that commonly emerged
from these studies were a general verbal ability, and the
specific skills of vocabulary knowledge and the ability to
manipulate language (Spache 1963).
Interestingly, research in the areas of readability
analysis and comprehension study seems to have come together
in the mid-fifties. The development of the cloze procedure
and its validation as a readability measure was closely
followed by experimental studies on cloze as a measure of
reading comprehension.
The cloze procedure opened new doors for research on
readin~. The earlier validation of cloze as a measure of
readability prompted research in the comprehension field.
Bornruth (1962) and Rankin (1965) both found cloze to be a
highly efficient and reliable measure of reading comprehen
sion in their work. They reported that cloze tests produce
good correlations with criterion tests over the same material
and that cloze tests measure comprehension as well or better
than comparable multiple-choice tests. Taylor (1957) found
that cloze test scores served as accurate predictors of
multiple-choice test scores.
9
The validation of cloze procedure as a measure of both
readability and reading comprehension has resulted in a
"merging" of these two concepts, so that many studies fail
to distinguish between the two. Certainly there is a link
between the two, and it is this interrelationship which gives
cloze one of its major strengths--the capacity to match
students and materials.
Earlier work with factor analysis in comprehension
studies led to factor analysis with the cloze procedure.
Weaver and Kingston (1963) carried out a factor analysis of
cloze tests and multiple-choice tests. Subsequent work by
Bormuth (1969) along the same lines demonstrated that the
same factors which account for most of the variability in
traditional comprehension tests also account for most of
the variability in cloze tests. The two factors isolated in
earlier studies, a vocabulary factor and a verbal relation
ship f~ctor, proved to be the primary factors involved with
cloze procedure.
Research with cloze procedure as a measure of read
ability and reading comprehension in English prompted studies
in other languages. These studies reported good results in
languages similar to English, such as French (Landscheere
1972), as well as in languages unlike English such as Thai
and Vietnamese (Oller 1972) and in Japanese (Yamada 1979).
It should be noted, however, that the cloze system appears
to work better in languages that have a history of written
literature. This may be due to the fact that for these
languages a standard written form has evolved.
10
Cloze tests have also been used successfully within an
ESL context to determine proficiency levels of high school
and college students (Oller 1973, Stubbs & Tucker 1974). A
modified "matching" cloze test has been used successfully at
lower elementary levels with ESL students in Saipan to test
comprehension in English (Baldauf & Propst 1979). Thus it
appears there is a definite application for the cloze
procedure in second language reading situations.
The use of cloze procedure has stimulated studies on a
number of technical or statistical issues, one of which is
a functional deletion rate for cloze testing. In English
this functional deletion rate is the point below which there
is too little context to successfully cloze a passage. At
deletion rates of less than four-to-one for example, there
is insufficient context from which to draw inferences, so
the successful completion of one item is highly dependent
on the successful completion of a prior item (MacGinitie
1961). (See also Aborn, et al. 1959.) The extreme inter
dependence of items diminishes scores as well as the useful
ness of the results.
At deletion rates of four-to-one and larger, the
surrounding context is sufficient to allow closure without
excessive dependency on a prior item. Most research in
English seems to accept that we need about eighty percent
of the context in order to complete the other twenty percent
(Miller & Friedman 1957). The determination of this
functional deletion rate is based on the availability or
11
lack of context, and not simply on the variability of scores.
Another issue related to cloze testing is which scoring
method to use. Studies in English indicate that there is no
significant difference between the exact word (EW) and the
acceptable synonym (AW) method. Rankin (1957), Ruddell (1964),
and Bormuth (1965) have concluded that the exact word method
is preferable in testing situations, and that it provides
all essential information with far less work.
On the other hand, recent research efforts have shown
a distinct trend toward use of non-exact or synonym scoring
methods. The rationale is that by not examining incorrect
responses and synonym generation, we lose valuable information
(Asher, et al. 1976). While cognizant of the fact that exact
word scoring is more practical, this trend seems to advocate
the le~rning aspects of cloze.
The majority of studies in English conclude that the
exact word scoring method is preferable, and that there is
no practical difference in the two methods. As a result,
the exact word method enjoys greater popularity because it
is more efficient.
A related issue is the concern about the system of
word deletion in cloze procedure. A variety of deletion
styles have been used, including lexical, structural, random
(Meredith & Vaughn 1978), and strictly mechanical deletion
(every gth word). The latter has withstood the test of
time, and also produced the highest correlations with
concurrent measures of validity.
12
The success of the cloze procedure in measuring both
readability and comprehension provides a contrast to the
limitations of standard readability formulas. While these
formulas can be applied to English, the application of such
a formula to the Samoan language does not produce acceptable
results. One reason is that in Samoan there has been no
formal study of word frequency, and there are no word lists
based on frequency of use. Secondly, most factors involved
in these formulas are language-specific.
There is one readability formula that could be applied
to Samoan, the Fry formula. This is based on two factors,
average word length and average sentence length, strictly
mechanical items. Although this formula can be applied to
Samoan (in a mechanical sense), the results are extremely
variab~e. For example, a passage in English was judged to be
at roughly the sixth grade level by Samoan readers, and by
standard readability formulas such as Dale-Chall. The
equivalent passage in Samoan, also at about the sixth grade
level, was placed at the twelfth grade level using the Fry
formula.
The reason for such a discrepancy is that in Samoan
every syllable not only has a vowel; but every vowel is also
a syllable. Syllabic reduplication is also a more common
element in Samoan. While the Fry formula is easy to use,
it is not effective for determining readability in Samoan.
13
Although current readability formulas appear to have no
application in Samoan, there is a strong probability that
cloze procedure would work as a measure of readability or
comprehension. Bormuth (1968) has a procedure for assigning
grade placement levels based on students' cloze percentage
scores. The purely mathematical aspects should work for
Samoan materials as well, if cloze procedure proves to be
applicable.
Certainly there are reasons why the cloze procedure may
not work in Samoan. The written literary tradition is
evolving slowly; the culture is basically an oral one. The
system of public education is relatively new, and only
recently have the schools returned to a policy of using the
vernacular for the lower primary grades. Up to now, most
students have had as much experience with beginning reading
in English as they have in Samoan, with mixed results. All
of thes€ factors make for a complicated educational situation
in which cloze procedure may have no application.
Just as there are reasons why the cloze procedure may
not have an application in Samoan, so too there are reasons
why it should. In terms of methodological considerations,
the process of closure should be within the reach of this
group. Research with cloze in English has involved second
grade through college, and although success at lower levels
is sometimes tenuous, it has been shown that students with
basic decoding abilities can manage the cloze task (Gallant
1965).
14
A second factor that favors cloze success in Samoan
is that the relationship between oral and written language
may be a more pronounced one in Samoan than in a language
like English. Samoa as an oral culture places great emphasis
on oral language and since reading is related to oral
experience, pride in the oral culture and narrative experi
ence should lead to a positive approach to reading.
Oral experience is related to reading mastery and it
appears to be. related to cloze success as well. As reading
skills mature, students usually do better on cloze tests
(Johnson 1979). The more mature reader exhibits a keener
knowledge of the process; not only are the scores higher but
the type of error is usually of a more logical nature (Asher,
et al. 1976).
Information theorists off er various models of language
processing, including several to explain the cloze procedure
(Tuimau 1972). It appears that no one model would serve to
explain this complicated process but it does appear that
language and cloze processing are themselves similar across
languages.
Perhaps because most semantic relationships are binary,
a student's awareness of collocation and coligation becomes
a primary skill in mastery of cloze procedure. At a certain
point in his development, the individual seems to develop a
metacognitive awareness of both the tendency and the require
ment of certain words to co-occur (Pickering 1977). This
knowledge, coupled with an understanding of the redundancy
15
in language, provides the student with the ability to cloze
a passage.
The processing of written language and spoken language
is distinctly different . In a speaking situation we cannot
go back and instantly replay words, which we can do in writing.
Cloze procedure is similar to reading in that both seem to
involve a horizontal and a vertical component, according to
Weaver (1965). The horizontal aspect involves an analysis
of the structural elements of the language while the vertical
component involves a lexical search for the proper semantic
aspect at any particular blank.
Samoan students quite probably deal with cloze reading
tests in much the same way that other students do. The rich
oral environment may facilitate closure because with little
in the way of printed material in Samoan, there is a good
chance that what occurs in oral narrative patterns will also
occur ~n print. Ruddell (1964) has shown that students cloze
passages with high frequency oral patterns more easily than
those with low frequency oral patterns.
We know that oral experience and prior knowledge both
play major roles in successful reading. Samoan students
must bring these aspects into play just as any other student
would to achieve mature reading. The probability of success
with reading and cloze procedure in Samoan rests to a great
extent on the integration of oral experience and prior
knowledge by Samoan students. With cloze success in Samoan,
we may infer that it measures the same aspects in Samoan
16
that it measures in other languages. If the process of
dealing with cloze is cross-linguistic in nature, then the
product should also be similar.
THE HYPOTHESES
There is presently no formal assessment measure for
readability or reading comprehension in Samoan . Yet it
seems that some method of measuring these aspects would be
desirable for materials currently being used and those under
development at this time. The possibility of finding a
relatively simple method for measuring both readability and
reading comprehension was considered, and the cloze procedure
was thought to be a logical choice for its simplicity, ease
of use, and range of applications. To test the application
of the cloze procedure, the following hypotheses were
formulated:
H1 : Cloze procedure is an effective tool for measuring
reading comprehension and readability in the
Samoan language.
H2 : There is no appreciable difference between exact
word (EW) and acceptable synonym (AW) scoring in
Samoan cloze tests.
H3 : There is an identifiable functional deletion rate
n for cloze tests in Samoan.
METHOD
Subjects: Ninth grade students were chosen as the
sample group for this study. Significantly, ninth grade is
the first year of high school in American Samoan, and by this
17
time students should have developed reading skills which
would allow them to deal with most of the literature that
is available .in Samoan. There is reason to believe that
students in lower grade levels can also handle the cloze
task, but for this exploratory study, mature subjects were
desired.
A second important factor in favor of this level is
that all ninth grade students take Samoan Language and
Culture as a subject, while those in higher grades do not.
Although students in English classes are tracked according
to proficiency levels, Samoan Culture classes are mixed
across levels, resulting in a more random sample.
Further justification for using students from a Samoan
Culture class is the fact that the course is taught in
Samoan and involves both language and culture. Thus, both
the subject matter in the tests and the tests themselves
would cyave greater content and face validity in a Samoan
Culture class than in any other situation.
Cloze reading tests: A series of fifty-item cloze tests
with varying deletion rates were constructed from a passage
taken from Tala O Le Vavau (1976), a book of Samoan legends
and culture. A scarcity of scientific material in Samoan
effectively precludes the testing of technical prose, even
though testing a variety of prose styles would be more useful
for determining the efficacy of cloze procedure in Samoan.
Seven cloze reading tests were prepared, with deletion
rates ranging from nine-to-one (nine words and a blank) down
18
to three-to-one (three words and a blank). In each case,
the blank space was a standard length of ten type spaces.
Testing this extreme of deletion rates represents an attempt
to discover whether there is an optimal deletion rate in
Samoan for measuring readability, as there appears to be in
English.
Using a four-to-one deletion ratio, one passage can
yield five different forms of a cloze test. There is some
justification for testing each form at any particular
deletion rate to obtain a truer measure of readability.
Bormuth (1964) points out that five forms of a cloze test,
all with the same deletion ratio but each beginning with
a different word, can yield five dif~erent means. In such
a case, the mean of the means would give the best indication
of an overall score.
Creating five or more forms of each test passage for
this s~udy would have meant a total of forty-two different
tests, and administration of each form would have reduced
the number of subjects for each test to about six, an
inadequate number for drawing any substantive conclusions.
Seven tests drawn from the same passage assures that most
words in the passage will be deleted at least once.
Traditional reading test: As a cross-validation measure
to determine the relationship of the cloze tests to tradi
tional reading comprehension testing, a sixty-item multiple
choice test on the same subject matter was constructed. This
test was based on fifteen short reading passages in Samoan,
19
covered the same type of material and was taken from the
same book as the cloze passages. Each reading passage was
followed by four multiple-choice questions designed to check
the student's ability to find specific information, make
inferences, understand vocabulary, and determine the main
idea. This test was pre trialed on twenty Samoan students
at Kaimuki High School in Honolulu, following which revisions
were made prior to testing in American Samoa.
The ideal situation would have been to administer a
standardized test in Samoan, against which the cloze tests
could be measured. The lack of such an instrument in Samoan
necessitated the use of an original test. A Samoan language
test given to all seventh and eighth grade students in 1976
no longer exists, and the results of this one-time test
have disappeared. With no standardized instrument available
in Samoan, the writing of an original test became the only
reasonable alternative.
Test administration: The cloze reading tests were given
to ninth graders in the fall of 1980. Tests were prepared
in groups of forty-two, with six forms of each deletion rate
within each group. The pre-packaged tests were then distri
buted randomly in Samoan Culture classes so that in each class
there would be an approximately equal number of each cloze
form. The mix of forms both within and between classes
provided the best opportunity for random sampling.
Students were told that the passages were designed to
measure their reading and understanding of different aspects
20
of Samoan Culture. The entire forty-five minute period was
allowed for completion of the passage. Oral instructions
and a short sample test in Samoan were presented prior to the
actual test to familiarize students with the format and test
taking procedures.
The multiple-choice reading test was administered to
the same students a month after the cloze tests. Though
taken from the same book, the multiple-choice format presents
the information in an entirely different way, and tests
receptive rather than productive skills. The same format
was used in the administration of the multiple-choice test.
That is, Samoan Culture teachers gave the test to the same
group of ninth graders as those who had previously taken the
cloze tests, but students in this case were allowed two
periods to complete the tests.
Data analysis: The cloze reading tests were scored
using both the exact word (E1d) and the acceptable synonym
(AW) method. The exact word method dictates that only an
exact replacement of the deleted word counts as correct while
in the acceptable word method, a semantically and syntactically
equal word is also counted as correct. Since one aspect of
this study was to determine which scoring method should
prevail, it was necessary to score each test twice.
Analysis of the multiple-choice reading test was
conducted primarily in the interest of discovering what
revisions might be needed to improve the instrument for any
future application. The test was corrected and subjected to
21
item analysis using the biserial r method, in which scores
of the highest 27% and the lowest 27% are correlated (Garrett
1965). Any item with a correlation of r = .20 or higher is
considered valid. Test reliability was determined with the
Kuder-Richardson Formula 21, a measure of interitem
consistency in student performance (Tuckman 1978).
Results from the multiple-choice reading test reveal
that thirteen items on the test had a correlation lower than
r = .20, and at first glance this would indicate a test of
questionable validity. A close examination shows that seven
of these items were "too easy," i.e. more than seventy
percent of the sample responded correctly to the questions.
An ideal difficulty level would occur if fifty to sixty
percent of the sample responded correctly.
Five of the items present a unique case in that these
five were originally valid items. They were marked correctly
but sc9red as incorrect, because the test key was changed
after inadequate analysis of results from the pre-test group.
The incomplete analysis predicated a change in the item
distractors so that five distractors originally labelled as
incorrect were changed to become the "correct" choices, and
the items thought to be correct were changed to serve as
"incorrect" distractors.
An example may provide a clearer explanation. The
Samoan word atamai was used as one of four distractors for
a question. It was not the correct choice, but the pre-test
group almost universally chose this word, which is similar to
22
Hawaiian akamai (wise). Based on these results, the master
test was changed so that atamai would become the correct
choice.
Upon receiving the tests from Samoa, it became evident
that the pre-test group was making choices more closely
aligned to the Hawaiian experience. Students in Samoa, on
the other hand, made choices based on their own cultural
background, which is to say they generally chose the original
proper choice. However, because the test key was changed
after the pre-test group results, they were in effect being
penalized for marking correct answers. Thus in the item
analysis, the multiple-choice test appears to be less reliable
than it really is.
Acknowledging that seven items were too easy and that
five items were reversed (marked correctly but not scored
correctly) partially explains the reliability index of
r = .7i, or r = .75 based on a reliability estimate from a
table of Harris (1969). A correction factor for the
inappropriate items and the mislabelled items would produce
a better reliability coefficient. For a researcher-made
first edition, the test nevertheless appears to be reasonably
valid as a cross-validation measure with the cloze reading
tests.
The BMDS biomedical program of the University of
California (Dixon 1977) was used in conjunction with the
Statistical Package for the Social Sciences (Nie, et al. 1975)
in the computer of the University of Hawaii for the data
analysis in this study.
23
A two-way analysis of variance was carried out on the
scoring data from the seven deletion rates on the cloze
tests to discover whether there is any significant difference
in means across deletion rates, and whether there is any
significant difference between the exact word and acceptable
word scoring methods.
A one-way analysis of variance was conducted on the
multiple-choice reading test to determine any difference
between groups in the sample which might necessitate the use
of the multiple-choice test scores as a covariate in sub
sequent analyses. Additionally, Pearson product-moment
correlations were computed on both the exact word and
acceptable word scoring data of each cloze deletion rate with
the multiple-choice reading test.
()Qce it was determined that there was only a slightly
significant difference between scoring methods, a one-way
analysis of variance and t-tests were run on the exact word
scoring data to provide information regarding the point of
significant falloff in scores. Subsequent analyses refer
only to exact word data, since results are similar for either
scoring method.
An analysis of trend was run on the exact word deletion
rate scores to determine which curve model would best
describe the pattern of cloze mean scores. A series of
t-tests were then conducted on all deletion rate pairs to
24
detect any slight but possibly significant differences
between non- adjacent deletion rate scores.
RESULTS
Hypothesis One: Cloze procedure can serve as a measure
of readability and reading comprehension in Samoan.
The results of the cloze reading tests shown in Table 1
demonstrate that at larger deletion rates, the students
replaced a majority of the blanks with the exact word
deleted from the text. At lower deletion rates, there were
less than fifty percent exact word replacements. In general,
students who scored in the mid to high range also generated
more synonyms, a trend that coincides with research by
Asher, et al. (1976).
Table 1
Means for Exact Word and
Acceptable Synonym Scoring Method
. Deletion Rate N Exact Word (SD) AcceEtable Word (SD)
9 to 1 35 29.60 (3.96) 29.71 (3.98)
8 to 1 35 30.11 (4.06) 30.23 (4.09)
7 to 1 35 29.89 (4.01) 30.03 (4.03)
6 to 1 35 29.43 (4.14) 29.S4 (4.19)
s to 1 3S 28.86 (4.61) 29.03 (4.64)
4 to 1 3S 22.31 (4.19) 22.46 (4.20)
3 to 1 3S 18.80 (3. SS) 18.86 (3.61)
Overall 27.00 27.12
25
The multiple-choice reading test scores are shown in
Table 2. An inspection of these means reveals little
variation across the seven deletion rates and the result of
a one-way analysis of variance on the means was non-
significant, F (6,238) = 0.31, ns. Since there were no
significant differences among the groups on their reading
test scores, it was therefore not necessary to employ the
multiple-choice reading test as a covariate in the subsequent
analyses (Cohen & Cohen 1976). The summary statistics . for
this analysis of variance are shown in Table 3.
Table 2
Nulc.iple-cho.ice .Read.ing I'esc Neans
Deletion Rate Mean SD N
9 to 1 41.26 6.06 35
8 to 1 40.14 6.33 35
7 to 1 41.43 5.83 35
6 to 1 40.69 6.73 35
5 to 1 41. 77 5.81 35
4 to 1 41.69 6.99 35
3 to 1 40.80 6.48 35
Table 3
Analysis of Variance on
Multiple-choice Reading Test Scores
Source df MS
12.2710
40.10
F
0.31
Probability
Deletion
Error
6
238
ns
26
Correlations between the multiple-choice reading test
scores and the two scoring methods across deletion rates are
shown in Table 4. These correlations demonstrate a positive
relationship between the two types of tests. The relatively
high scores on the cloze tests, and the strong relationship
between the multiple-choice and cloze reading tests justify
acceptance of the first hypothesis.
Table 4
Correlations between Multiple-choice Reading Test Scores
and Cloze Deletion Rate Scores
Deletion Rate Exact Scoring Acce:etable Scoring
r df .E. r df .E. - -M-C test 9 to 1 .84 33 <.001 .83 33 <.001
M-C test 8 to 1 .87 33 <.001 .86 33 <.001
M-C test 7 to 1 .85 33 <.001 .86 33 <.001
M-C test 6 to 1 .88 33 <.001 .87 33 <.001
M-C test 5 to 1 .90 33 <.001 .90 33 <.001
M-C test 4 to 1 . 85 33 <.001 .84 33 <.001
M-C test 3 to 1 .83 33 <.001 .84 33 <.001
M-C test Overall .58 243 <. 001 .59 243 <.001
Hy:eothesis Two: There is no appreciable difference
between exact word and acceptable word scoring in Samoan.
A two-way analysis of variance (7 x 2) was performed
on the cloze test scoring data. The between factor was
deletion rate (9:1, 8:1, 7:1, 6:1, 5:1, 4:1, and 3:1) and
the within or trial factor was type of scoring method (exact
27
word versus acceptable word). The results from this analysis
of variance indicated that the deletion rate factor and the
type of scoring method factor were both significant,
F (6,238) = 43.04, p<.0001, F (1,238) = 33.55, p<.0001,
respectively. The deletion rate by type of scoring method
interaction was non-significant, F (6,238) = 0.41, p = .87.
Summary statistics for this analysis of variance are shown
in Table 5.
Table 5
Two-way Al-IOVA on Deletion Data
Source
Deletion Rate
Error
Type of Scoring
D x T
Error
df
6
238
1
6
238
MS
1444.97
1.84
0.023
0 . 055
F
43.04
33.55
0.41
Probability
< .0001
< .0001
.87
The correlations between the exact word scoring method
and the acceptable synonym scoring method for each of the
deletion rates is shown in Table 6. As expected, and as the
table indicates, each of these correlations is higher than
.99, demonstrating a near perfect relationship between the
two scoring methods.
28
Table 6
Correlations between Exact Word Scoring Method and
Acceptable Synonym Scoring Method for each Deletion Rate
Deletion Rate r - df Probability
9 to 1 .997 33 <.001
8 to 1 .997 33 <.001
7 to 1 .996 33 <.001
6 to 1 .997 33 <.001
5 to 1 .997 33 <.001
4 to 1 . 996 33 <.001
3 to 1 .998 33 <.001
Overall .998 243 <.001
An inspection of the means for the scoring methods
across the deletion rates reveals a similar finding. The
overall mean for the exact word scoring method was 27.00 and
for the acceptable word scoring method 27.12 (see Table 1).
An inspection of the individual cell reveals a similar slight
difference of about .1, a point favoring the acceptable
synonym scoring method across all deletion rates.
Though the type of scoring factor was significant in a
strictly statistical sense, the actual size of the difference
between the two methods (0.12) and the near perfect correla
tion between the two scoring methods (.998) raises the
question of whether this is a meaningful difference. It does
not appear that such represents a truly meaningful difference
and the subsequent analyses will only be done on the exact
29
word scoring data. The analyses of acceptable word scoring
data reveals identical results to those to be reported on
the exact word scoring method, and this points to acceptance
of the second hypothesis.
Hypothesis Three: There is a definable functional
deletion rate n for use with cloze tests in Samoan.
Table 7 shows the results of a one-way analysis of
variance and trend analysis on the deletion rate factor
versus exact word scoring data. The results of a one-way
analysis of variance indicated a significant effect for the
deletion rate factor, F (6,238) = 43.11, p<.0001. The
deletion rate factor accounted for fifty-two percent of the
variance in the exact word score data.
30
Table 7
Analysis of Variance and Trend Analyses on
Deletion Rate-.- Exact Word Score Data
Source df SS MS F p R2 ----
Deletion Rate 6 4316.496 719.4160 43.1114 <.0001 .521
Cum R 2
linear trend 1 3004.713 3004.713 180.059 <.0001 .363
I
jdeparture from !linear trend 5 1311.783 262.357 15.723 <.01 i
quadratic 1 1124.183 1124.183 67.367 <.0001 .498
departure from quadratic 4 187.600 46.90 2.81 .03
cubic 1 22.671 22.671 1.359 ns .501
departure from cubic 3 164.929 54.976 3.295 .02
quart~c 1 41.509 41.509 2.49 ns .506
departure from quartic 2 123.420 61.710 3.698 .03
quintic 1 96.991 99.991 5.81 <.OS .518
departure from quintic 1 26.429 26.429 1.584 ns
sex tic 1 26.402 26.402 1. 582 ns .521
Error 238 3971.595 16.6874
31
A series of t-tests were performed between adjacent
deletion rates in order to determine the location of the
significant differences between the various rates. There
were no significant differences between the nine-to-one
versus eight-to-one rates (t (238) = 0.53, p = .60), the
eight-to-one versus seven-to-one, (t (238) = -0.23, p = .82),
the seven-to-one versus six-to-one (t (238) = -0.47, p = .64),
or the six-to-one versus the five-to-one rates (t (238) =
0.59, p = .56).
Table 8
t-tests on Cloze Exact Word Scoring Data
9 to 1 vs. 8 to 1 t(238) = 0.53, p = .60 ns
8 to 1 vs. 7 to 1 t(238) -0.23, p = .82 ns
7 to 1 vs. 6 to 1 t(238) = -0.47, p = .64 ns
6 to 1 vs. 5 to 1 t(238) = -0.59, p = .56 ns
5 to r vs. 4 to 1 t(238) = -6.70, p < .0001 significant
4 to 1 vs. 3 to 1 t(238) = -3.60, p = .0004 significant
However, deletion rates five-to-one versus four-to-one,
and four-to-one versus three-to-one were significantly
different from each other, t (238) = -6.70, p<.0001, and
t (238) = -3.60, p = . 0004, respectively . A summary of
these t-tests is presented in Table 8. Figure 1 presents
these results in graphic form across the deletion rates.
32
Figur e l
Me .?.n Percent Corr ect Res pon ses a t the Various Deletion Rates
60 ~
u w ~ ~
0 u 50 ~
z w u ~ w a... 40 z <( w ~
30
o--~__._~~_._~~..i..-~__..~~--'-~~-'-~..-J
9:1 8:1 7:1 6:1 5:1 4:1
DELETION RATE
33
Subsequent t-tests were also performed between all pairs
of deletion rates, and these results provided the same
information. That is, from deletion rates of nine-to-one
through five-to-one there were no significant differences
between any pair. But there was a significant difference
between pairs four-to-one and above, or four-to-one and
below.
It was hypothesized that a quadratic curve model would
best describe the data shown in Figure 1. An analysis of
trend was used to test this hypothesis, and the results in
Table 7 show that a quadratic model accounted for the data
significantly better than a linear model, F (1,238) = 67.36,
p . 0001.
While the linear model accounted for thirty-six percent
of the variance (of a total of fifty-two percent), the
quadratic model accounted for an additional thirteen percent.
A test.for departure from quadratic was also significant,
F (4,238) = 2.81, p = .03, but the amount of additional
variance gained by going beyond the quadratic model is only
two percent. Thus a quadratic model best describes the
relationship between deletion rate and the exact word score
dependent measure (Kirk 1968).
The analysis of variance on the exact word scoring data
demonstrates that there is a significant difference in means
across deletion rates. The subsequent t-tests point out
exactly where that significant difference occurs. The data
leads to a confirmation of the third hypothesis that there is
' ' I
I l
34
a definable functional deletion rate n for cloze tests in
Samoan; however, it is not clear whether this functional
deletion rate is five-to-one or four-to-one, or perhaps both,
depending upon the type of reading passage used.
DISCUSSION
Hypothesis One: Cloze procedure can serve as a measure
of readability and reading comprehension.
The first hypothesis states that cloze procedure can
effectively serve as a measure of readability and reading
comprehension in the Samoan language, and the results
demonstrate that this is so. The mean percentage correct
scores on the cloze reading tests shown in Table 1 indicate
that Samoan students can use cloze procedure successfully
with all but the smallest deletion rates.
The multiple-choice reading test in Samoan was designed
to measure comprehension over the same material as the cloze
reading tests. Results on the multiple-choice reading test
also show clearly that the students performed reasonably
well with this traditional test format. Table 4 displays
strong correlations between the cloze reading tests and the
multiple-choice reading test, an indication that the two
tests are measuring a common ability.
At larger deletion rates, the cloze mean scores cluster
around sixty-percent, which is in fact the most common mean
across the five larger deletion rates. At a four-to-one
deletion rate, the mean score of forty-four percent suggests
a passage that could be handled at the "instructional" level
35
based on criteria cited by Bormuth (1967), and corresponds
to the 75% comprehension score on traditional reading compre
hension tests.
The mean score of just under thirty-eight percent at a
three-to-one deletion rate indicates that this test rate
was too difficult. But since this latter test was over the
identical passage as the tests which produced sixty percent
means, we can see that the lower score is a reflection of the
diminished context at this smaller deletion rate (Fillenbaum,
et al. 1963).
Cloze functions effectively in languages dissimilar to
English, notably in Japanese (Yamada 1979) and Vietnamese
(Klare, et al. 1971). There is no basis for assuming that
cloze somehow measures something totally different in
Samoan. The high cloze mean scores demonstrate that the
relative lack of an extensive written tradition is not a
varia~le which affects cloze testing.
The mean scores among all groups on the multiple-choice
reading test establish that these students can deal with a
traditional comprehension test. The correlations between
the multiple-choice reading test and the cloze reading tests
signify a close relationship between the two, a fair indica
tion that both tests are measuring comprehension.
Just as cloze appears to measure understanding of the
material, so too it gives an indication of the readability
of the passage. In general, a greater understanding of the
material being read will generate a higher cloze score.
36
Cloze seems to measure comprehension and comprehensibility
at the same time, which perhaps explains why many people
speak of them as almost the same concept.
A good indication of the relative readability of the
passage is that the students produced mean scores of around
sixty percent over the five larger deletion rates, with
lower means only when the context is substantially reduced.
The larger scores demonstrate that the students have a good
understanding of the material, for if it were too difficult
the scores would be lower, even at larger deletion rates.
Another indication of the readability level of the
passage used for closure is that subjective ratings by Samoan
readers placed it at the upper elementary level. A consensus
of teachers and students who read the undeleted passage was
that Samoan high school students should have no real diffi
culties with the passage. (Primary school in American Samoa
runs through eighth grade, and high school begins with the
ninth grade.)
Subjective assessment together with the high mean scores
on the cloze reading tests demonstrate the overall effective
ness of the cloze procedure as a measure of general
readability. Further, the scores on a traditional instrument
(the multiple-choice reading test) together with the strong
relationship between the two types of tests show that cloze
is a valid measure of comprehension in Samoan. The efficacy
of cloze procedure as a measure of both readability and
37
reading comprehension is substantiated by the data, and this
points to a confirmation of the first hypothesis.
Hypothesis Two: There is no appreciable difference
between exact word and acceptable word scoring methods.
The second hypothesis states that there is no appreciable
difference between the exact word (EW) and acceptable word
(AW) scoring methods with cloze tests in Samoan. While
there is a slight difference between the two scoring methods,
the results nevertheless justify acceptance of this hypothesis.
Table 5 presents the results of a two-way analysis of
variance on the scoring methods across deletion rates. The
results indicate a statistically significant albeit minor
difference in scoring systems. Table 1 illustrates that the
difference in total mean score for the two scoring methods
on the cloze reading tests amounts to 0.12, which in practi
cal terms does not constitute a crucial difference.
C9rrelations between the exact word and acceptable word
scores are shown in Table 6. These correlations are all in
excess of r = .99, nearly perfect. Since both sets of scores
were produced by the same groups, this is not unusual. But
with the total difference in means so small, and the
correlations between scoring methods so high, it becomes
clear that whatever happens with the exact word scores will
also occur with the acceptable word scores.
This duplication of results is exactly what has
transpired. The analyses of variance, correlations, and
t-tests provided exactly the same information with both
38
scoring methods. That being so, the exact word method should
provide all the necessary information with no question about
what constitutes an acceptable synonym.
In English, the bulk of the proof lies with the exact
word method. Studies by Taylor (1957) and Ruddell (1964)
state that exact word scoring provides basically the same
results as acceptable word scoring with far less work. Exact
word scoring is the preferred method in testing situations
to preserve scorer reliability and to provide economy.
Since there is a slightly significant difference in
scoring metho.ds, additional testing may prove useful.
Because a great deal depends on subjective assessment of
synonyms, it would be wise to proceed with caution before
making any conclusions in favor of the acceptable synonym
method, especially since the exact word method seems to do
the job equally well from a practical standpoint. A careful
exami~ation of the data _prompts acceptance of the second
hypothesi~.
Hypothesis Three: There is ·a definable functional
deletion rate n for cloze tests in Samoan.
The third hypothesis states that there is a functional
deletion rate n for cloze testing in Samoan. An analysis
of the data indicates that not only is this true, but the
functional deletion rate in Samoan falls within the range
of those commonly used in English, four-to-one (4:1) and
five-to-one (5:1).
f
I t i ' ' f ,
I
39
The results of a two-way analysis of variance in Table
5 indicate that there is a significant difference in mean
scores across deletion rates. A close examination reveals
that this difference is the result of a sizeable drop in
mean scores between the five-to-one and four-to-one deletion
rates, and a smaller but still significant drop between the
four-to-one and three-to-one rate. This pattern can be
seen quite clearly in Figure 1.
The t-tests shown in Table 8 confirm that there is no
significant difference in means from the nine-to-one through
five-to-one deletion rates. There is only a significant
difference between deletion rates four-to-one and above, or
four-to-one and below.
The data suggests that both a four-to-one and a five-
to-one deletion rate would be functional for cloze tests in
Samoan. Arguments can be given for both deletion rates using
the data from this study. Individual interpretation may
favor one or the other, but an absolute determination cannot
be given based on the administration of one test passage to
one sample group.
Evidence for the four-to-one functional deletion rate
comes from the fact that it falls directly between a deletion
rate that produces inadequate results, and larger deletion
rates which are equally easy to cloze. Further, the use of
larger deletion rates forces the reader to process more text
and use more context to achieve basically the same results,
so in essence this would be less efficient reading.
t
I I
i
I f !
I I I l I
I I I i
I l
40
The designation of a functional deletion rate in
English appears to be based on the amount of context.
MacGinitie (1961) has established the extreme interdependence
of items at deletion rates of less than four-to-one. Other
authorities as well accept that in English we need about
eighty percent of the context in order to complete the
other twenty percent (Bormuth 1967). ·
Figure 1 demonstrates that scores are relatively stable
from deletion rates five-to-one th~ough nine-to-one. The
fact that there is no significant difference in means for
these deletion rates indicates that increased context at
these rates does not produce a concomitant increase in mean
scores, but does require increasing amounts of context to
provide the text base of fifty deletions. This coincides
with results by Aborn and others (1959) which show that
additional surrounding context beyond four words does not
produce any significant gain in score.
We see in Table 8 that moving from a five-to-one
deletion rate to a four-to-one deletion rate does result in
a significant drop in means. Thus we have some evidence that
a four-to-one deletion rate provides less but still sufficient
context. Going from a four-to-one rate to a three-to-one
rate again results in a significant drop in mean score , and
a crucial drop in context.
At the three-to-one deletion rate. the mean percent
correct scores of just under thirty-eight percent fall within
the "frustration" level designated by Bormuth (196 7) , and
41
this would normally mean that the material is too difficult.
But other cloze tests taken from the same passage show higher
means, so the passage itself is not too difficult. Rather,
this narrow deletion rate is too difficult to process because
of the minimal surrounding context.
The reduced context at a three-to-one deletion rate
would dictate a high degree of interdependence among the
items. This means that the successful completion of one item
rests on successful completion of a prior item. The result
is that scores do not reflect the ability of the testee so
much as they reflect a significant element of chance.
An argument for the five-to-one functional deletion rate
for cloze tests in Samoan is that one must reach this point
for there to be no further change. It is true that moving
to a larger deletion rate would produce no significant
change, and that moving to a smaller deletion rate would.
This ~ine of thought relies on a statistical perspective
rather than the contextual bias generally employed with cloze
research in English.
Additional support for the five-to-one functional
deletion rate may be derived from the larger standard
deviation produced at this rate. If greater variability is
a preferred criterion, then five-to-one would stand as a
more useful deletion rate, based on results from this test
administration. Hittleman (1978) concurs that deletion
rates larger than four-to-one provide more variation.
42
The type of prose is a further consideration which
could favor one deletion rate over the other. The test
passage in this study was essentially narrative material on
relatively familiar subject matter. Narrative material
generally produces higher results than technical prose where
function words carry the logic and where fewer synonyms are
used (Smith-Burke, et al. 1978). The use of a technical
passage may show that five-to-one is a functional deletion
rate.
Where there is a desire to measure information gain,
four-to-one may be the functional deletion rate. However,
a five-to-one deletion rate seems to produce greater
variability and thus may prove to be more functional in some
instances.
It is not possible to unequivocally state which of the
two deletion rates is more functional based on the data
from one test. More comprehensive testing should demonstrate
which of the two is a functional deletion rate for cloze
tests in Samoan. The hypothesis nevertheless stands
confirmed in that an identifiable functional deletion rate
for cloze tests in Samoan is clearly in evidence.
CONCLUSIONS
The intent of this project was to determine whether
the cloze procedure works effectively in Samoan as a measure
of both readability and reading comprehension. An analysis
of the data supports the conclusion that cloze does in fact .
measure aspects of readability and reading comprehension in
43
the Samoan language. This parallels the validation of cloze
in languages dissimilar to English.
The strong correlations between the cloze reading tests
and the multiple-choice reading test over the same type of
material demonstrates that cloze tests measure comprehension
as well as readability. That cloze provides essentially the
same information and functions as a global measure of compre
hension gives weight to the argument that cloze can serve as
an effective instrument in Samoan.
The issue of which scoring method to use can be resolved
on the basis of practicality. There is a statistically
significant difference between the exact word and acceptable
word scoring methods, although this difference is marginal.
The exact word method provides essentially the same informa
tion and is far more efficient.
The optimal deletion rate appears to be either five-to
one or. four-to-one, perhaps depending upon the kind or reading
passage being used. The determination of these .rates is based
upon the significant falloff in mean scores at deletion rates
of less than five-to-one. On the one hand the four-to-one
deletion rate uses the least amount of reading context and
appears to be the point of balance between the too easy and
too difficult continuum. On the other hand the five-to-one
deletion rate functions as effectively as deletion rates of
six-to-one through nine-to-one while using correspondingly
less reading context than the higher rates. The same
44
arguments with regard to necessary surrounding context that
apply to English seem to be relevant to Samoan.
This study produced a few salient points about the
measurement of reading in Samoan. Heretofore, measurement
of readability, reading comprehension, and reading skills
has been conducted on an informal basis. The lack of a
repertoire of materials in Samoan has made the need for
creating evaluation instruments less than critical. But the
recent resurgence of efforts to create materials in Samoan
means that some form of evaluation is in order for reading
related materials. All indications are that the cloze
procedure can serve as an economical, efficient adjunct to
these efforts .
. IMPLICATIONS
A number of issues that evolved from this study warrant
further considerations. The determination that cloze
procequre is a viable measure of reading comprehension and
readability has implications for the Samoan language program
in general and the materials development/reading program in
particular. The existence of an economical evaluation
instrument for both comprehension and passage difficulty of
new texts should prove beneficial.
Several reconrrnendations seem appropriate. It would be
useful to administer cloze reading tests across grade levels
in all high schools and in upper elementary grades throughout
Samoa. The object would be to test various passages at one
particular deletion rate, either four or five to one, with
45
five forms for each test to obtain an average of the five
means. The means may vary across grade levels, but the data
would provide the basis for establishing grade placement
levels for books.
The creation of a standardized instrument in Samoan, a
Samoan language test or reading comprehension test, is
extremely desirable. There must be a way to evaluate student
progress on a yearly basis. Further, a norm-referenced
instrument would ·prove highly useful as a cross-validation
measure for cloze tests and any other tests developed in
Samoa.
Test administration on this scale requires close
cooperation at high levels and more than a degree of persis
tence. There are priorities to consider, but every school
system needs evaluation procedures to measure its own
success vis a vis the students. Cloze tests could facilitate
the ev.aluation process substantially by virtue of the fact
that they are simple to create, to use, and to correct.
A project of this scope requires careful planning and
execution, but it seems clear that the end results would
more than compensate for the effort. With little else
available in the way of formal assessment mechanisms, the
cloze procedure would appear to have the broadest application
with the lowest cost, in terms of both time and money .
System-wide testing should provide long term benefits, as
more materials and books are produced in the Samoan language.
46
APPENDIX A: INSTRUCTIONS AND SAMPLE TEST
Su'ega fa'ata'ita'i ma ni fa'amalamalamaga atili
FA I AMALAMALAff.AGA
I le fa'ai'uga o le itulau lenei, o 10 1 0 i ai se
fa'aa'oa'oga o se itua'iga su'ega fou. 0 nei su'ega taitasi
sa lolomiina mai i ni palakalafa o se tusi. 0 lo'o iai upu
o lo'o misi i laina ta'itasi.
0 lau vaega e fai, o le tusiina mai lea o le upu o 10 1 0
misi i faiupu taitasi. Tusi i le avanoa o lo'o iai le laina,
i luga o le laina.
1. Tusi na'o le upu e tasi i avanoa ta'itasi.
2. Taumafai e fa'atumu avanoa uma. 'Aua le fefe e mate.
3. Taumafai e sikipi avanoa faigata, ona toe fo'i mai
lea iai pea uma ona fa'atumu isi avanoa.
4. Taumafai e tusi le upu sa'o. E le afaiina pe a le
sa'o le sapelaina, a ia tau fetaui ma le upu sa'o.
5. 0 le tele o avanoa e taliina i upu ua tatou masani
ai, ae o nisi upu e tusiina fa'apea:
*O fainumera e pei - 3,247 po'o le $12.00 po'o 1978
*O nisi upu e ta'ilua, e pei o le
47
SU'EGA FA'ATA'ITA'I
0 101
0 i lalo ifo o le itulau lenei se su'ega fa'ata'ita'i.
Fa'atumu le avanoa i le upu ua e iloa ua le'o fa'atumuina. ·
Toe siaki lau pepa pe a uma, i lou toe vaai lelei i au tali.
0 tali na o lo'o tusiina, fa'au i luga ma lalo i le pito i
lalo o le pepa. Tusi manino mai ia mama.
Su'ega
0 Ioane o le tama i vasega amata i le
kolisi (freshman), ma lo'o maua e ia
fa'afitauli uma pei ona masani ai le
to'atele tamaiti fa'atoa amata. 0 se mea
mautinoa lava, o ia fa'afitauli uma
amata mai a'o le'i alu ese
ia mai lona aiga. E tele sa ia faia e le'i
mana'o
o ai mo le kolisi.
VC1UJ
0
. 8
• +J
f ai aua 0 le a alu
0 . L
.£
C1SC1
vs
0
vn
. 6
·~
. t
r 1.
t r ! 48
APPENDIX B: CLOZE TEST PASSAGE IN ENGLISH
TRANSFER OF LAND
The transfer of land from one f arnily to another is
justified in the following instance: a chief may be
embarrassed by the unexpected arrival of his sister or a
fellow chief and his not having a pig to welcome the visitor.
He reflects which of his neighbors--chief or orator--owns a
big porker or two. He then goes to the owner a~d says,
"Chief, I find myself in a predicament and should like to
have the pig that is wandering about your place." The other
replies, "You have done well to come to me in your trouble.
Send your young men to catch the animal."
After the visitors have departed, the chief discusses
with the members of his family how to repay his friend. They
decide that the chief go to him and speak as follows: "I
have two offers to make to reward you for your service. I
am ready to give you a boat or, if you prefer, you may have
a piece of land. II The other replies, "I thank you for corning
along. Keep the boat; your children may need it to go
fishing. Let me have the piece of land; it will be useful
to my family. Thank you." The land will belong to the new
owner for ever after.
There is another justification for the transfer of
land. Since the days of yore there have been in Samoa expert
bonito fishermen. They are adept in the art of tying the
49
bonito fishhook. The people firmly believe that if the hook
is wrongly tied, no fish will be caught.
If a chief wants to become a bonito fisher, he must
first learn how to tie the hook. He will go to the chief
fisherman and ask him to teach him the art. The fisherman
grants his request and shows him how to fasten the tortoise
shell hook to the shaft of pearl shell. As a result, the
chief catches many bonitos.
The chief then considers what return to make to the
fisherman, whether to repay him in fine mats or pigs. He
realizes that neither would do him much good because, in a
fit of temper, the fisherman might take away from him the
right of tying the hook. A piece of land, on the other
hand, will be a constant reminder of the transaction .
I
I l
r I i ~ i
50
APPENDIX C: CLOZE TEST PASSAGE IN SAMOAN
0 LE FAAFESUIAIGA FANUA
0 le faafesuiaiga o se f anua mai le isi aiga i se isi
e mafai ona faatulagaina faapea. Afai ua tigaina se alii
ina ua omai se malaga a sona tuafafine poo se tamalii, a ua
leai se manu e tai ai se sua i lena malaga, ona filifili
ai lea o ia poo ai se tulaf ale poo se alii o iai se manu
tele lava, pe tasi pe lua.
Ona ia alu· lea i le al ii o iai sana manu ma fai atu,
"Alii e, ua ou le maufautua, o lea ua ou sau ai i lau manu
o 10 1 0 taa i lou pa'epa'e ina ia ou avea e fai ai lou faa-
lavelave. II Ua tali atu le alii, "Ua le lei ua e maliu mai
lou tigaina. 0 mai ni taulelea e pue atu le manu. II
Ua toe f oi atu le malaga i lo latou nuu moni, ona
f ilif ili ai lea 0 le alii ma lona aiga uma i se mea e taui
ai le pule a le alii na aumai ai le manu. Ona latou maua
ai lea o le tonu e faapea, ia alu atu le alii i le matai
ma fai atu ia te ia. "E lua mea nei ua ou sau ma au ou te
folafolaina atu e tali ai lau pule. 0 le vaa, a e le
f inagalo iai ou te avatua le f asif anua e taui ai lou aga
lelei." Ua tali atu le tulafale, "Ua faafetai maliu mai,
tuu ai pea le vaa, e aoga i ou alo, ae aumai le fasifanua
ua e folafola mai nei e agai atu iai lau fanau. Faafetai
lava." 0 lena fasifanua o le a tumau lava i le aiga e
faavavau.
i
I I I I l
f t ~ !
51
E iai le isi ala e rnaua ai le fanua. 0 le tu rnai
le vavau a Samoa, ua iloga fale tautai o le alo atu. Sa
faufau e i latou pa e sisi ai atu. Sa faapea taofi o Samoa,
a sese ona fau o le pa, ua leai se atu e maua e lea vaa.
Afai o se alii ua manao i le fale tautai, ona manao
foi lea o ia ina ia na iloa faufau o pa e maua ai atu. Ona
fai atu ai lea o ia i le tautai sili, "Ua ou manao ia ou
iloa faufua pa; ia e alofa lava oe le tautai sili, ia
a'oa'o mai ia te au le faiga o lea mea." Ua talia o ia e
le tautai sili. Ona a'oa'o ai lea o ia i le fauga o pa.
Ua tele atu e maua i ana pa e faufau.
Ona f ilif ili ai lea o ia poo le a se mea e aoga e
ave atu i le taupo o ni ietoga poo ni puaa. Ona faapea
ai lea o lona taofi, e leai se e aveatu ai ni ietoga poo
ni puaa i le tautai, aua a ita o ia i se aso, toe fao lea
o le faufau pa. E lelei ona tuu atu ia te ia o se fasi-
fanua. faamanatu ai i le tautai ua tonu lava, ona ua tuuina
mai e ia le faufau ia te au.
r I
I
I
r I ! ' 53
FAATONUGA
0 palakalafa nei e fesoasoani ia te i matou e suesueina ai
pe faapefea le tulaga o lau faitau. I faai'uga uma o vaega
taitasi o loo iai fesili e fa e tauina mai ia te i matou pe
f aapef ea le lelei ona e
1) sueina faamatalaga moni;
2) faia faaiuga;
3) malamalama i upu;
4) sueina le manatu autu.
E sefulu ma le lima vaega i lenei suega. Ia faitau lemu i
vaega taitasi. A uma ona faitau, tali lea o fesili e fa.
A uma loa lea vaega ona faasolo lea se ia ma e'a. A amata
loa ona e galue i le suega ona, 'aua lea ona e toe fesili.
'Aua nei mataina lenei tusi. Ia faatumu le avanoa · i
talaane o le numera e fetaui ma le tali ua e f ilifilia i
le pepa mo tali; faataitaiga:
A B c D
0) 0 © Amata le suega pe f aapea mai le f aiaoga, "Amata".
54
E lua ala o toga: o toga muamua e ta'ua o le laufau (e lavalavaina e le tamaitai i lana faaipoipoga); o toga rnulimuli e ta'ua o le saga. 0 le ie na aumai ai laufau, o le ie lea o le tamasa. 0 le ie ua aumai ai le saga, o le ie lea o le a maua e le tulafale e ana le togafiti i lona fale; a o tulafale uma e tofu lava i latou ma le ta'itolu pe ta'ifa ie toga ma o latou siapo e faalagolago i le tele o siapo o le nunu.
1. 0 le ituaiga toga e lavalavaina e le tamaitai i lana f aaipoipoga o le
1. saga 2. laufau 3. laufala 4. siapo
2. 0 le ala e lavalavaina ai ele tamaitai ona o le
1. mativa 2. tulaf ono 3 . mo'omo'oga 4. aganuu
3. 0 le upu i le palakalaf a e uiga maua
1. ala 2. faapitoa 3. togafiti 4 . tofu
4. 0 le manatu taua o lenei vaega o le .•
1. ituaiga o ietoga 2. tofusia o ietoga 3. lalagaina o ietoga 4. teuina o ietoga
r 55
0 nofoaga pe a fai ni saofaiga. E le so'ona nonofo i le fale, aua ua iloga nofoaga. Afai o se alii, e nofo tonu i le matuatala po o pou o le tala. Afai o se tulafale, e nofo i le pou i le pepe o le fale ma · le atualuma foi. Ua ta'ua o va-i-pou po'o va-i-matai tulafale ua le maua ni o latou pou, aua e gata ia i latou e sili ona maua pou.
5. I lenei palakalafa, 0 ai tagata e le nof o i se pou?
1. · alii 2. f aipule 3 . tulaf ale 4. kovana
6. 0 tagata i le f ono a matai e
1. maf ai ona filifili lona nofoaga . 2. iloa lona nofoaga. 3. le iloa lona nofoaga. 4 . le mautinoa lona nofoaga.
7. 0 le upu i le palakalaf a e uiga rrrua:mua
1. pou 2. nof oaga 3. maua 4 . sili
8. 0 le manatu sili 0 lenei palakalaf a o le
1. nofoaga sao 2. sili o le alii 3. fono a matai 4. pou o le fale
56
0 tulafale sa faia lea filifiliga latou te su'esu'eina lenei tamaitai pe faapefea ona itu, pe tutusa itu e lua, e lelei le itu o le tama, lelei foi le itu o le tina. E faapea foi le aiga o le tamaitai ma lo latou nuu, latou te su'esu'eina le tama poo le alii, pe tutusa ona itu pe leai. Afai ua latou iloa e le tutusa itu o le alii ma lo latou tamaitai, ona te'ena lea o le aumoega. Afai o se tamaitai e le tutusa ona itu, e le mafai ona alu i ai se aumoega a se nuu atoa; e gata lava i le pitonuu e i ai le alii ona o atu e aumomoe i lea tamaitai.
9. Tasi le ala e lelei lava e taofi ai se talanoaga o se faaipoipoga:
1. e le aulelei le tama. 2. e le masani le tama. 3. e lei faa~ua le tama. 4. o le tagataese le tama.
10. 0 le faiva o le aumoega o le
1. tausi le aumoe. 2. vaavaai le aumoe. 3. puipui le aumoe. 4. filifili le aumoe.
11. 0 le upu i le palakalafa e uiga agatonu
1. motu 2. sosoo 3. gata 4. tutusa
12. 0 le manatu sili o lenei palakalafa o le
1. aumoega 2. faaipoipoga 3. faafiafiaga 4. suesueina itu
57
Af ai o se tagata e tautala tu i luma o alii ma tulafale i totonu o se fale, o le faalemigao lea. E tosoina o ia i fafo e ni taulelea ma f asi ia te ia ma tuli ese, 'aua le toe nofo mai i le mea o potopoto ai le nuu. Afai foi o se tagata e inu tu i totonu o se fale, o le faalemigao lea. E faia lava pei ona taua i luga.
13. 0 se tagata e tu ma inu i totonu o le £ale o le
1. ona 2. faalemigao 3. fia inu 4. agamasesei
14. 0 se tagata o loo tauaveina se amoga i luma o se matai e rnafai ona
1. sasaina 2. f asiotia 3. f aif ai 4 . faatelenoaina
15. 0 le upu i le palakalaf a e uiga alualu
1. totolo 2. tamo'e 3. tuli 4. su'e
16. 0 le ulutala sili mo lenei palakalafa o le
1 . Amioga a le Tupulaga 2. Faalagiga o le Matai 3. Finauga ma le Matai 4. Amioga Faalefeagai
r i
58
0 se alo teine o se alii, ua ta'ua foi o le taupou. E f aapotopotoina i ai o teine ma tamaitai o le nuu e fai se fale aualuma. Ua iloga foi le mamalu o lea tamaitai i le toatele o lona galuteine. E i ai foi i le aualuma ni isi taulelea, o tane a isi o le aualuma, e nonofo tumau ai i le ao ma le po. 0 lo latou igoa o palepaleaulama. 0 le galuteine foi o le tamaitai ua iloga ona soaf afine pe toatolu pe toafa. 0 latou na e mulimuli pea i le tamaitai taupou i mea uma e femaliuai ai. A taumafa foi le tamaitai, o le afafine o le tulafale e auauna i ai. Afai foi e alu le nuu i se malaga, ua iloga pea le taupou ma lona aualuma. E teuteu lelei i ie toga. A talimalo foi le nuu, o le taupou foi e pita tele ana teuga.
17. 0 mea ai a le taupou e laulauina e
1. uso o le tulafale 2. tama o le nuu 3. tuagane o le taupou 4. afafine o le tulafale
18. 0 le taupou e le tele se
1. aualuma 2. tof i 3. galuega 4. ietoga
19. 0 le upu i le palakalafa e uiga faaaloalo
1. mamalu 2. teuga 3 . aualuma 4 . galuteine
20. 0 le igoa sili mo lenei palakalaf a 0 le
1. Uo a le Taupou 2. Galuega a le Taupou 3. Aiga o le Taupou 4. Aualuma a le Taupou
59
E ese'ese uiga o tagata e gaoi i maumaga o isi tagata. 0 le isi e gaoi talo, ae fai lelei tiapula ma tuu lelei e toe toto i le maumaga o lea tagata. E faasalaina o ia i sala o le gaoi pe lima sefulu ni talo ae valu sefulu ni afi. 0 le isi tagata ua gaoi e ia le maumaga, ma ave uma talo ma tiapula. E faasalaina o ia i le sala lava o le gaoi e tai tutusa ma le o muamua.
21. 0 ituaiga gaoiga ese'ese, o le fasalaga e
1. ese'ese 2. tutusa 3. manafa 4. tele lava
22. O se solitulafono matautia o se tagata ua avea aa ma
1. ave tiapula 2. laulaau o tiapula 3. faasalalau o tiapula 4. toe toto tiapula
23. 0 le upu i le palakalafa e uiga e le faapenei o le
1. tutusa 2. eseese 3. isi 4. masani
24. 0 le ulutala sili mo lenei palakalafa o le
1. Tagata Gaoi 2. 0 Ituaiga Talo 3. Gaoiga Talo 4. Tiapula ma Talo
60
Afai ua nofo mai le tamaitai i le manaia sa alu i ai lana aumoega, ona la o mai lea i le nuu o le manaia. Ona faasaga atu lea o le nuu o le manaia, ua latou faia se galuega, o le ati o se paepae e faailoga ai le taupou. Ona mafaufau ai lea o se isi tulafale i sana togafiti ina ia na maua se ie o le tamaitai. Afai ua manao o ia i le taua o i le tama o le tamaitai, pe a aumai ona toga, e alu loa le tulafale ma f ai atu i le tama o le manaia, "Alii e, ai e leZei ona awnai o Zou faZeuZu e ufi ai Ze paepae o Ze tamaitai." Ua fai atu le alii, " Ua ZeZei, awnai."
25. E ofoina atu e tulafale mo se ulugalii fou se
1. ulu 2. fiafiaga 3. fale 4. ietoga
26. 0 le upu sili e f aamatalaina ai lenei tulaf ale 0 le
1. matamau 2. matape'ape'a 3. f ealof ani 4. fa'a'au'au
27. 0 le upu i le palakalaf a e uiga f aamanatu
1. aumai 2. faasaga 3. mafaufau 4. faailoga
28. 0 le ulutala sili mo lenei vaega o le
1. Faaiuga o le Faaipoipoga 2. Mauaina o se Ietoga 3. Fauina se Fale 4. Fauina o le Paepae
61
0 aumaga o fanau tama uma a alii ma tulafale. Tiga ona matutua ma ua tata, e le mafai ona avea i latou ma tagata o le aumaga, seiloga ua avatu se puaa e fai ma ana usufono, ona faatoa avea lea ma tagata o le aumaga. Ua faatoa fai lana monotaga. Ua faatoa maua ai sona tufaaga i mea a taulelea pe a fai ni faigaai.
29. A e lei auai le tama i le aumaga, e tatau ona f aia nei mea vagana ai le mea e tasi
1. ia atoa tausaga 2. ia faaipoipo 3. ia ave se puaa 4. ia ta le pe'a
30. I lenei palakalafa, 0 tama 0 loo i le aumaga e
1. aulelei 2. talavou 3. malosi 4. atamai
31. 0 le upu i le palakalafa e uiga faataga
1. maua 2. totogi 3. faatoa 4. avea
32. 0 le ulutala lelei mo lenei vaega o le
1. Taulelea 2. Aumaga 3. Monotaga 4. Tulafale
62
Af ai ua tigaina se alii ina ua o mai se malaga a sona tuafafine poo se tamalii, a ua leai se manu e tai ai se sua i lena malaga, ona filifili ai lea o ia poo ai se tulafale poo see alii o i ai se manu tele lava, pe tasi pe lua. Ona ia alu lea i le al ii o i ai sana manu ma fai atu, 11 AZii e, ua ou Ze maufautua, o Zea ua ou·sau ai i Zau manu o Zoo taa i Zou paepae ina ia ou avea e fai ai Zou faaZaveZave. 11 Ua tali atu le alii, " Ua ZeZei ua e maZiu mai i Zou tigaina. 0 mai ni ta.uZeZea e pue atu Ze manu."
33. 0 le matai e iai malaga (asiasiga) faateitei mai
1. tamalii 2. uo 3. fafine 4. taulelea
34. E le tatau i se matai ona laulau i ni malo lauiloa se
1. ta lo 2. puaa 3 . pilikaki 4. palusami
35. 0 le upu i le palakalaf a e uiga oto
1. filifili 2. fesuiai 3. talimalo 4. fautua
·36. 0 le igoa le lei mo lenei palakalafa o le
1. Meaalofa a le Matai 2. Faafiafiaga a le Matai 3. Fesoasoani a le Matai 4. Faalavelave a le Matai
63
Faapef ea le pule a alii i tagata ma le laueleele o le nuu? E pule i ona lava tagata, e tautua i lona lava alii. 0 latou foi e ta'ua o ona tagata e pule i ai. Ao fanua, e le soona pule, aua e iloga fanua e autu i le suafa. 0 le ua nofo i le suafa, e pule aoao i fanua uma e autu i le igoa. Peitai afai ua tea ma le igoa ona tea foi lea ma fanua uma sa pule i ai. E gata lava se fanua e pule i ai, o se fanua na maua e ona lava lima.
37. 0 tagata i lalo o le puleaga a le matai e taua latou o ona
1. pologa 2. USO 3. auauna 4. uo
38. E ie maf ai e le matai ona
1. tuf atuf a fanua 2. faatau atu laueleele 3. f aataumai f anua 4. ·gaoi f anua
39. 0 le upu i le palakalaf a e uiga faatamala
1. soona 2. taua 3. pule 4. tautua
40. 0 le ulutala sili mo le palakalafa o le
1. Pulega a le Matai 2. Fanua a le Matai 3. Augatupu o le Matai 4. Aiga a le Matai
64
Mo le fasioti tagata, e taui le oti i le oti. E le faatali sei maua le na ia fasioti le tagata, ae tau lava o se e vave maua e le aiga o le tagata ua oti; poo le uso, poo le matai o le aiga, poo sona atalii, e sui ai le ua oti. A ua i ai se togafiti e mafai ai ona ola o le sala; o le ifoga e alu i le aiga poo le itumalo o le tagata ua oti. E ave le pagota ma faapulou i le ietoga, e ta'ua lea ie o le ie o le malo. E mafai lava ona ola ai le pagota; e saoloto foi o ia i mea uma e fealuai ai.
41. E faamaloina se fasioti tagata pe a fai na te
1. alu ese 2. fai se ifoga 3. tautino lona mama 4. tala faaoti
42. 0 le aiga o le ua lavea e mafai ona tuli vag·ana ai le o le fasioti tagata.
1. f aif eau 2. atalii 3 . tuagane 4 . matai
43. 0 le upu i le palakalaf a e uiga ufiufi
1. f aatali 2. f aasala 3. f aapulou 4. faatauga
ff4. 0 le manatu taua o le palakalaf a o le
1. Faasalaga 2. If oga 3 . Finauga 4. Tauimasui
65
Ua ta'ua o le faalemigao le tu i luma o alii. Afai o se e ai tu i luma o se alii pe tautala, e faasalaina lea. Afai foi e amo tu se toi i lumafale o se alii e fasia lea tagata. 0 se tagata alu malaga, a tau atu i se galuega o fai, afai e alu loa, o le faaletea. E fasia tagata faapea. A tau atu i se galuega e fai i le ala, ona faigaluega ai lea, o le faaaloalo sili lea. E faapea foi o se vaa e tau atu o toso, e tatau ona faasega e tosovaa ai. Ua sa foi ona pa'o se galuega o alu se malaga i le ala. A se malaga o vaa e sa ona ulitu lo latou vaa. Ua sa foi ona faaaoga se faamalu poo se pulou lauua i luma o alii, pe fafatu se tama i lumafale o alii.
45. 0 a ituaiga tagata e tu i luma o alii?
1. pisa 2. faalemigao 3. tauaso 4. talavou
46. A e fetaui ma se tagata o loo toso lona vaa, e tatau ona e
1. faapea tulou ma e pasi 2. taofi lau taumafa 3. tuu i lalo lau avega 4. tu ma fesoasoani
47. 0 le upu i le palakalafa e uiga vavao
1. leaga 2. soli 3. faasala 4. sa
48. 0 le manatu taua o le palakalafa o le
1. galuega mo le matai. 2. aai faatasi ma le matai. 3. faaaloaloga mo le matai. 4. femalagaai ma le matai.
66
Af ai ua f ef e le nuu o le alii na ia mulilua ma le tamaitai ina nei tupu se mea mataga, (misa) ona o uma ai lea o le nuu o le alii, o taulelea ma tulafale uma ma alii e fai le ifoga, faatasi ma le na mulilua. E manao le nuu o le alii, latou te o atu i le vaeluapo ma faatatau ia latou taunuu atu i le nuu o le alii, pe'a ta le lima, a o lei ala tagata. Latou te savavali ma le filemu. E leai se isi e tautala, e leai foi se 1s1 e tale. 0 mea e lavalava e le pagota o le ietoga. Latou te ave foi se ietoga e taua o le ie o le malo, e pulou ai le na agasala. Ona latou nonofo ai lea i le malae ma ifo o latou ulu. Latou te punonou pea seia oso le la.
49. E lavalavaina e le mulilua le
1. lau 2. siapo 3. lavalava 4. ietoga
50. 0 le mulilua ma lona nuu e tatau ona taunuu
51.
52.
i le nuu ua ia faatigaina, a e lei feala tagata ina ia
1. taunuu vave 2. Malo lo laititi 3 . f ai if oga 4. f aatali
0 le upu i le vaega e uiga f usu
1. f inau 2. mataga 3. mis a 4. f ef e
0 le ulutala sili mo lenei vaega 0 le
1. Malaga 2. Ifoga 3. Mulilua 4. Faasalaga
67
0 faailoga vaaia o le tofiga ma pule i le nuu. Ua iloga lava le ua tofia i ni isi faailoga. Afai o se tulafale sili, e lua mea e faailoga ai: o se fue afa tele, ai pe valu ni pauna, ma se tootoo pau, ai e valu futu le umi. A lauga le tulafale i se fono tele a le nuu poo.le itumalo, e tu ma le tootoo ma le fue afa. 0 le fue lea ma le tootoo e tuuf aasolo lava i le f anau ma e e au i le igoa. Ua faapea, a tausi i fue afa ma le tootoo, e maua ai le poto o le tama e lana fanau. Ua sili fanau a tulafale i le matapopoto ma le televavave i feau a alii. 0 latou foi e tufaina ava o alii.
53. 0 le talitonuga, a tausi e le atalii fue af a ma le tootoo, na te maua le a le tama.
1. tupe 2. f anua 3. po to 4. f aaaloalo
54. 0 le f ue ma le too too 0 taua.
1. me lei 2. f aailoga 3. laau 4. pule
55. 0 le upu i lenei vaega e uiga f aasoasoa
1. sueina 2. aumai 3. aveina 4. tuf aina
56. 0 le ulutala sili mo lenei vaega o le
1. Mea a le Tupu 2. Atalii o le Matai 3. Faailoga o le Pule 4 . Agatonu i le Fono
So masani le nuu o Amoa i le tapui faititili. Sa tapuai i latou i faititili. Afai o se tagata ua ia ave
68
se mea i le fanua ua uma ona tapuia i le tapui faititili, o se malaia lava o ia pe to'ia i se faititili pe to'ia o latou fanua poo latou fale i se faititili. Afai e to'ia se tagata i se faititili poo so latou fanua, ona iloa ai lea o le tagata lava lea ua na aia le tapui pe ave ese se mea i le fanua ua tapui i le tapui faititili.
57. 0 faatoaga o Amoa o loo puipuia e
1. faititili 2. matagi 3. uila 4. timu
58. 0 lenei atua e mafai ona taia tagata o loo
1. paie 2. gaoi 3. valea 4. ona
59. 0 le upu i le vaega e uiga taisia
1. ootia 2. toia 3. fofoina 4. tapuai
60. 0 le manatu taua o lenei vaega o le
1. atua o le nuu 2. tala o le nuu 3. faomea o le nuu 4. tapui o le nuu
BIBLIOGRAPHY
Aborn, M., Rubenstein, H., and Sterling, T. D. 1959. Sources of contextual constraint upon words in sentences. Journal of Experimental Psychology 'i]_, 3:171.
Asher, S. R., Hymel, S. and Wigfield, A. 1976. Children's comprehension of high-and-low interest material and a comparison of two cloze scoring methods. ERIC ED 134 939.
Baldauf, R. B. and Propst, I. K. 1979. tests for elementary ESL students. 6:683-690.
Matching cloze Reading Teacher 32,
Bormuth, John. 1962. Cloze tests as measures of readability and comprehension ability. (Unpublished doctoral dissertation) University of Indiana.
. 1964. Experimental applications of cloze tests. -----In J. Allen Figurel (Ed.) Improvement of reading through classroom practice. International Reading Association Conference Proceedings 9:303-306. Newark, Delaware: International Reading Association.
. 1965. Validities of grarmnatical and semantic ---c~l-a-s-sifications of cloze test scores. In J. Allen
Figurel (Ed.) Reading and inquiry. International Reading Association Conference Proceedings 10:283-286. Newark, Delaware: International Reading Association .
. 1967. Comparable cloze and multiple-choice -----comprehension test scores. Journal of Reading 1.Q, 2:291-299.
. 1968. The cloze readability procedure. ---E~l-em-entary English 45, 4:429-436.
. 1969. Factor validity of cloze tests as measures ___ o_f_r-eading comprehension ability. Reading Research
Quarterly 4, 3:358-365.
Chall, Jeanne S. 1958. Readability: Appraisal of research and application. Columbus: Onio State University.
Cleland, D. L. 1966. A construct of comprehension. In J. A. Figurel (Ed.) Reading and inquiry. International Reading Association Conference Proceedings 10:59-64. Newark, Delaware: International Reading Association.
Cohen, Jacob and Cohen, Patricia. 1975. Applied multiple regression/correlation analysis for the behavioral . sciences. Hillsdale, N. J.: Lawrence Erlbaum Associates.
70
Dale, R. and Chall, J. S. 1948. A formula for predicting readability. Educational Research Bulletin 27:37-54.
Davis, F. B. 1944. Fundamental factors of comprehension in reading. Psychometrika ~' 3:185-197.
1972. Psychometric research on comprehension in reading. Reading Research Quarterly z, 4:628-678.
Dixon, W. J. (Ed.) 1977. BMD Biomedical Computer Programs. Berkeley: University oT"California Press.
Entin, Eileen B. and Klare, George B. 1978. relationships of readability, cloze and scores on a reading comprehension test. Reading Behavior ..!.Q., 4:417-435.
Some intermul tip le- choice
Journal of
Fillenbaum, S., Jones, L. V. and Rapoport, A. 1963. The predictability of words and their grammatical classes as a function of rate of deletion from a speech transcript. Journal of Verbal Learning and Verbal Behavior l, 2:186-194.
Flesch, Rudolf F. 1951. How to test readability. New York: Harper.
Fry, Edward. 1968. A readability formula that saves time. Journal of Reading!.!., 7:512-516.
Gallant, Ruth. 1965. Use of cloze tests as measures of readability in the primary grades. In J. Allen Figurel (Ed.) Reading and inquiry. International Reading Association Conference Proceedings 10:286-287. Newark, Delaware: International Reading Association.
Garrett, Henry E. 1965. Testing for Teachers. 2nd ed. New York: American Book Company.
Gilliland, John. 1972. Readability. London: University of London Press.
Harris, David P. 1969. Testing English as a second language. New York: McGraw-Hill.
Hittleman, D. R. 1978. Developmental reading: a psycholinguistic perspective. Chicago: Rana-McNally.
Johnson, John R. achievement.
1979. Oral language and reading ERIC ED 177 636.
Kirk, R. E. 1968. Experimental design. Belmont, Ca: Brooks, Cole.
71
Klare, George R. 1963. The Measurement of readability. Ames: Iowa State University Press.
Landsheere, G. 1972. Application du cloze test de W. L. Taylor a la langue francais. Scientia Paedagogica Experimentalis ~' 2:207-256.
Lorge, Irving. 1959. The Lorge formula for estimating difficulty of reading materials. New York: Columbia University.
MacGinitie, Walter H. 1961. Contextual constraint in English prose paragraphs. Journal of Psychology 51, 1:121-130.
McKenna, Michael C. 1976. Synonymic versus verbatim scoring of the cloze procedure. Journal of Reading 20, 2: 1"41-143.
Meredith, Keith and Vaughn, Joseph. 1978. Stability of cloze scores across varying deletion patterns. In P. David Pearson and Jane Hansen (Eds.) Reading: Disciplined inquir! in process and practice. Twentyseventh yearbook o the NationaI"l'leading Conference. Clemson, S.C: National Reading Conference:l81-184.
Miller, G. R. and Coleman, E. B. 1967. A set of thirtysix prose passages calibrated for complexity. Journal of Verbal Learning and Verbal Behavior~' 6:851-854.
Miller, G. A. and Friedman, E. A. of mutilated English texts. J:38-55.
1957. The reconstruction Information and Control l,
Nie, Norman H., Hull, C. Hadlai, Jenkins, Jean G., Steinbrenner, Karin, and Bent, Dale H. 1975. SPSS: Statistical package for the social sciences. 2nd ed. New York: McGraw-Hirr:-
Oller, John. 1973. Cloze tests of second language proficiency and what they measure. Language Learning 23, 1:105-118.
Oller, John W., Bowen, J. Donald, Dien, Ton That, and Mason, Victor W. 1972. Cloze tests in English, Thai, and Vietnamese: native and non-native performance. Language Learning 22, 1:1-15.
O'Reilly, Robert P. and Streeter, Ronald E. 1977. Report on the development and validation of a system for measuring literal comprehension in a multiple-choice format: preliminary factor analytic results. Journal of Reading Behavior~' 1:45-69.
72
Osgood, C. E. 1962. The nature and measurement of meaning. Psychological Bulletin !!2_, 3:197-237.
Pickering, Michael. 1977. Some observations on cloze tests. ERIC ED 140 602.
Rankin, Earl F. 1957. An evaluation of the cloze procedure as a technique for measuring reading comprehension. (Unpublished doctoral dissertation) University of Michigan.
Rankin, E. F. 1965. The cloze procedure--a survey of research. In E. L. Thurston and C. E. Hafner (Eds.) The philosophical and sociological bases of reading. Fourteenth yearbook of the National Reading Conference. Milwaukee, Wisconsin. National Reading Conference: 133-150.
Ruddell, R. B. 1964. The effect of oral and written patterns of language structure on reading comprehension. Reading Teacher]&, 4:270-275.
Singer, H. and Rhodes, A. 1976. Learning from text: a review of theories and research at the high school level. Reflections and investigations on reading. Twenty-fifth yearbook()f the National Reading Conference. Clemson, S.C: National Reading Conference:22-51.
Smith-Burke, Margaret, Gingrich, Patricia S., and Eagleeye, Daniel. 1978. Differential effects of prior context, style, and deletion pattern on cloze comprehension. In P. David Pearson and Jane Hansen (Eds.) Reading: Disciplined inquirt in process and practice. Twenty·seventh yearbook o tlie Nationa~eading Conference. Clemson, S.C: National Reading Conference:l33-137.
Spache, George. 1953. A new readability formula for primary grade reading material. Elementary School Journal .2l_, 3:410-413.
1963. Toward better reading. Champaign, Ill: Garrard Press.
Stubbs, J. B. and Tucker, G. R. 1974. The cloze test as a measure of ESL proficiency for Arab students. Modern Language Journal 58, 6:239-241.
Stuebel, C. Samoa.
1976. Tala o le Vavau: Myths and Legends of Wellington, NeW-Zealand: Reed.
Sullivan, Rita J. 1976. A comparison of results obtained using the cloze procedure with readability levels using the Dale-Chall formula in selected university textbooks. ERIC ED 136 246.
Taylor, Wilson. 1953. Cloze procedure: a new tool for measuring readability. Journalism Quarterly lQ, 3:415-433 .
73
. 1957. Cloze readability scores as indices of ~~_,...in-.-d~iv-idual differences in comprehension and aptitude.
Journal of Applied Psychology~' 1:19-26 .
Tuckman, Bruce W. 1978. 2nd ed. New York:
Conducting education research. Harcourt, Brace Jovanovich, Inc.
Tuiman, J. Jaap. 1972. Speculation on cloze as a search procedure. In Frank P. Greene (Ed.). Investigations relatin~ to mature reading. Twenty-first yearbook of the National Reading Conference. Milwaukee, Wisconsin. National Reading Conference:74-84.
Weaver, Wendell W. 1965. Theoretical aspects of the cloze procedure. In Eric S. Thurston and Lawrence E. Hafner (Eds.). The philosophical and sociological bases of reading. Fourteenth yearbo~of the National Reading Conference. Milwaukee, Wisconsin. National Reading Conference:115-132.
Weintraub, Samuel. 1968. The cloze procedure. Reading Teacher 21, 6:567-571.
Yamada, Jun. 1979. Cloze procedure and the measurement of reading comprehension in disconnected sentences and continuous passages. The Science of Reading 23, 2:33-43.