The Thorndike-Lorge Reading Test for Grades 7 to 9
by Irving Lorge - 1945
A description of the Thorndike-Lorge Reading Test. The Thorndike-Lorge Test is planned as a general test of silent reading comprehension. It includes all the important factors in silent reading with reasonable weight for each factor.
READING varies widely. Its goal may .be information, action, inspiration, or entertainment. Its operations range from mere perception of words and skimming of a page to appreciation of subtle aesthetic effects and search for answers to difficult questions. By the end of grade six, children will "have learned to read" by recognizing in the printed symbol some of the concepts and organization that they ordinarily get in comprehending the spoken word. In a sense, when they enter the junior high school, children will be more or less efficient in the mechanics of reading, but there will still be room for tremendous improvement in reading comprehension.
Children in the junior high school need to extend their ability in vocabulary, in the metaphoric use of language, in the comprehension of large units of thought, in the understanding of the less usual literary constructions, and in synthesizing their reading with previous experiences, both direct and vicarious. All these important abilities, moreover, must be so learned that children can use them independently for pleasure and for knowledge. By the time pupils are ready to enter the senior high school their vocabularies should include some 25,000 or more word meanings. It must be recognized that some of the more frequent words in English, like "fall" and "run," have many meanings. As a matter of fact, each of these two common words has over two hundred different meanings listed in the Oxford English Dictionary. It cannot be expected that each meaning will be taught separately. On the other hand, children and adults must learn to appreciate the significance of the meanings of words from their context. Much of the reading of adults implies comprehending what is read on the basis of the clues within the paragraph or passage.
Some of these meanings undoubtedly reflect figurative and metaphoric use of words and phrases. Frequently the literal meanings of words will act as obstacles to the understanding of passages unless the child has been made aware of the fact that words and phrases can be used in numerous senses aside from the more conventional meanings. Many of these figurative uses he will never hear, but he will see them in print. The junior high school grades, therefore, must stress the value of contextual and figurative understandings, so that the child will be able to read independently for pleasure, for information, for appreciation, and for extension of understandings.
English sentence constructions, moreover, will vary from the direct, "I like that book" to such constructions as, "Who knows but that the king may be here," and "Her being a witch is quite conceivable." Other English constructions, such as, "eager to go as I was" or "The wood was not so thick but that she could see" or "two feet long," or "fighting five days," present problems in comprehension. The less usual English word and sentence constructions often are significant obstacles to understanding. In the junior high school grades, children must learn to recognize and comprehend such constructions. Too often the range of those available to children has been restricted.
Finally, by the time the pupil has reached grade ten, he should be able to handle larger and larger thought units. This should involve holding in mind several elements in the organization of the passage, the fusing of these elements with previous knowledge and experience, and the ability to go beyond the paragraph in terms of inferences, judgments, and hypotheses. The teaching of reading in the junior high school, therefore, implies teaching the pupil to give reasonable weights to the various facts, relations, and modes of expression to increase his understanding and appreciation of the passage. Certainly, such a process is related to the abilities usually measured by intelligence tests, but just as certainly the process of relating, valuing, and judging can be taught to a degree.
Each of these abilitiesunderstanding meanings from context, understanding and appreciating figurative and metaphoric use of words, understanding the functional use and force of various English constructions, understanding large units of thought, and inferring beyond the passageshould be, and frequently is, among the objectives of junior high school reading. But these reading abilities are not discrete; they are interacting abilities which should produce a total effect of improved ability to comprehend the written word rapidly and independently.
There could be a dozen valid tests of reading, each different from the others in its relative emphases upon a reader's purposes, attitudes, intellectual abilities, backgrounds, and operations. The Thorndike-Lorge Test1 is planned as a general test of silent reading comprehension. It includes all the important factors in silent reading with reasonable weight for each factor. Among the curricular objectives of silent reading there would certainly be included the ability to keep in mind the printed text; the understanding of the text in terms of answering questions specifically answered by it and in terms of implications and extensions from the text; and the ability to attend to, and understand, the text with reasonable speed. The factors specifically incorporated in the Thorndike-Lorge Reading Test are:
1. Knowledge, of the meanings of words and phrases in context.
2. Knowledge of English constructions, including idioms.2
3. Ability to infer the meanings of words from their component parts, and their resemblances to known words and their contexts.
4. Ability to keep in mind the parts of a sentence, paragraphs, or passages, and to use them to comprehend the whole.
5. Ability to organize and use the meanings of sentences, paragraphs, and passages so as to fulfill the informational, aesthetic, and extensional purposes of reading.
6. Ability to read fast enough to facilitate factors 4 and 5 and to prevent undue expenditure of time in reading.
As a total test requiring forty minutes of a pupil's time, the test measures the usual curricular objectives of reading for the junior high school gradesability to get meaning from the printed page, to use the understandings to answer questions, and to organize and use information for the solution of problems.
In view of the factors in the test, it is necessary to emphasize that no separate test of vocabulary is included. It is believed that the functional understanding of vocabulary in context is a better indication of reading ability than scores on separate vocabulary tests.
There is no doubt that the range of vocabulary is important in estimating the informational and interest backgrounds of pupils, and that such specific vocabulary tests are related to general reading measures. The test, however, stresses the functional approach to the appraisal of word meanings and semantics.
Furthermore, the test measures the understanding and appreciation of figurative expressions, and of literary as well as informational reading matter. In the course of the work in developing the Teacher's Word Book of 30,000 Words3 and the English Semantic Count,4 it was observed that the reading material of adults in such magazines as the Saturday Evening Post, Ladies' Home Journal, Woman's Home Companion, Reader's Digest, and True Story frequently included proverbs, aphorisms, and significant literary quotations. A major aim of the teaching of reading is to give children the ability to understand and to get the literary values of the apt epithet, imagery, and metaphor as well as the factual values of history, science, or technology.
Ability to read with understanding and appreciation depends largely upon intelligence; and any adequate measure of ability to read will be more or less a measure of intelligence as it operates with language. The test, however, avoids sentences or paragraphs designed to be intellectual puzzles, tasks which require the pupil to use intellect in opposition to habits ordinarily successful in reading highly abstract matter or scientific formulae.
VALIDITY OF THE TEST
The experimental edition of each of the five forms of the test was given to about 200 pupils, altogether approximately 1,000 pupils, in a junior high school, together with the Iowa Silent Reading Tests (New Edition), Elementary Test (1939); New Stanford Reading Test, Form Y (1929); Stanford Reading Test, Advanced, Form D (1940); and the New York State Reading Progress Test for Grades 7 through 12, Form A (1940). The criterion for estimating the validity of each item of each of the five forms was the average of the standard scores in each of the five tests administered. For each item, the biserial correlation coefficient was computed. If an item did not have a validity coefficient significantly different from zero, it was rewritten or a new item was substituted. Fortunately, on the average, only five items in each form had to be changed or substituted.
With the cooperation of the Comprehensive Testing Program of the New York City Work Projects Administration,5 the revised experimental edition of five forms was administered to a representative sampling of pupils in grades 7 to 9 in junior high schools of New York City. The sample included a representative proportion of adjustment classes of the less able, of regular classes for the normal range of ability, and. of rapid advance classes of the intellectually gifted. Each form, together with the four tests enumerated above, was given to about 1,600 pupils. The comprehensive Testing Project made available the grade equivalents for the Iowa, the New Stanford, the Stanford, and the New York State Reading Progress tests.6 Recognizing the fact that some of the scores are not really distributed, the correlations among the various criterion tests correlated, on the average, .80. Each of the five forms of the Thorndike-Lorge Reading Test correlated with each of the criterion tests, on the average, .80. It is reasonable to infer that the Thorndike-Lorge Test measures essentially what is measured by other reading tests, or the community of reading ability measured by each of the four criterion tests.
For a sample of 2 3 2 pupils in grades 7 to 9 in one of the junior high schools, data were obtained, however, on the subtest raw scores of the New York State Reading Progress Test. These raw scores were correlated with the scores on the Thorndike-Lorge Reading Test, Form 1. The raw scores on the New York State Reading Test by subtests refer to the following:
"1. A test of detailed understanding of paragraphs; consisting of paragraphs of 100 to 150 words, each followed by three multiple-choice questions.
"2. A test of the appreciation of the central idea of a paragraph; consisting of 100-word paragraphs each followed by a multiple-choice question on the best title for the paragraph.
"3. A test of general vocabulary; consisting of synonyms items in a multiple-choice form."7
The intercorrelation of the scores on the Thorndike-Lorge Reading Test with those of the subtests of the New York State Reading Progress Test was .77 with subtest 1, .70 with subtest 2, and .75 with subtest 3. Among the sub-tests of the Reading Progress test the intercorrelations were .71 between sub-tests 1 and 2, .79 between subtests 1 and 3, and .71 between subtests 2 and 3. The intercorrelation between the properly weighted total score of the Reading Progress test and the Thorndike-Lorge Test was .82. This suggests that the Thorndike-Lorge adequately measures detailed understanding of paragraphs and appreciation of central thought and word meanings, and that separate measures of each of these objectives correlate among themselves as much as a global test of reading objectives correlates with each of them. Global is used in the sense of a total score measuring the resultant of the interaction of all reading abilities.
Against the criterion of the average standard score for grade equivalents based on the four criterion tests, the item validities were again checked in the new samples. Each item was positively and significantly related to the criterion.
On the basis, therefore, of the item analysis and of the applications with other tests of reading, it may be asserted that the Thorndike-Lorge Reading Test gives a valid estimate of global reading ability in grades 7 to 9.
To four samples of approximately 200 pupils each, in grades 7, 8, and 9, two different forms of the Thorndike-Lorge Reading Test were administered. The intercorrelations ranged from .86 to .92. The estimated reliability of a single form of the Thorndike-Lorge Reading Test is approximately .90. By the Kuder-Richardson formula,8 utilizing the difficulty of the item and its biserial correlation with the criterion, the reliability is somewhat higherranging from .92 to .95, with an average of .94. It may be asserted, therefore, that the test is discriminating enough to measure individual differences in global reading ability.
DESCRIPTION OF TEST
The items of the test involve the matching of statements such as, "You bring the sleet, blazing fire, and Christmas treat" with words, e.g., angels, December, lighthouse, and sky; the answering of specific questions concerning detail based on a short paragraph; the recognition of the equivalence of literary, aphoristic sentences such as, "The cheapest automobile makes the most racket" with other statements, e. g., "Do not overvalue yourself," "Off with the old, on with the new," "Cur dogs bark loud," and "Actions speak louder than words"; and the understanding of longer passages of from 200 to 300 words in terms of questions of general import, inference beyond the passage, organization of thought, and specific detail.
The test requires comprehension of the material read, with adequate emphasis upon speed. The credits for speed are so arranged that the relative weights of quality of comprehension to speed are approximately 12 to 1. No credit attaches to speedy misreading.
Administering and scoring the test requires no special training. The teacher orients the pupils to the task, and after forty minutes collects the papers. The scoring is quickly accomplished through the use of a strip-key.
SCORES AND INTERPRETATION
The score on the Thorndike-Lorge Reading Test is the number of right responses. There is no penalty for omitted or incorrect answers. For each score, a sample table gives the following facts:
1. The grade equivalent, i.e., the grade of pupils in American junior high schools in 1943, for which that score is "normal." In each instance, the grade equivalents are based on samples of 1,600 pupils, and have been checked against the grade equivalents of reputable reading tests.
2. The reading age equivalent, i.e., the age of pupils in grades 7, 8, and 9 of American schools in the year 1943 for which that score is "normal." Again, in each instance, the reading age equivalents are based on samples of 1,600 pupils, and have been checked against the reading age equivalents of reputable reading tests.
It must be pointed out, however, that, in local situations, the grade equivalents and the reading age equivalents should be estimated for the community. Variations in school practice with reference to philosophy of promotion may affect the meaningfulness of norms for local use. The developing tendency toward "100 per cent promotion" or "promotion according to maturity" has the effect of making grade scores and reading scores more and more an artificial device. At best, the grade score and the reading age equivalent can be used for comparing the status of students at different times. The reference to 1943 implies that the norms may not be adequate for 1950 or later.
Furthermore, it is important to recognize that the curve of the development of reading ability flattens out very rapidly after the fifteenth or sixteenth birthday. In this respect, the curve of reading growth parallels the curve of measured mental growth. It is probable that, after age sixteen, the curve of reading ability is almost horizontal.
The Thorndike-Lorge Reading Test measures the important curricular objectives of vocabulary in context, understanding of idiom, ability to answer questions of detail, of general significance, of inference, of organization for purposes of information, and of appreciation, with a proper emphasis upon speed. The test provides a reliable measure of individual differences and pupil growth.
The validity of the test has been doubly assured by item analysis against an adequate criterion and by the affiliation of the test with other reputable tests of reading ability. It achieves its reliability and validity without the use of a separate vocabulary test. The emphasis throughout the test is upon functioning reading comprehension necessary to the understanding of the written and printed materials at the junior high school level.
The items of the test measure the recognized factors in reading comprehensionfunctional word knowledge, reading as reasoning involving relationships within and among paragraphs as well as inferences beyond the passage, sensing the author's intent, following directions, obtaining factual detailed information, getting the general import of the passage, ability to understand figurative, literary, and metaphorical language, and the ability to deal with a large variety of English constructions.
1 Published in two forms by Bureau of Publications, Teachers College, Columbia University.
2 Edward L. Thorndike, Annie L. Evans, Laura H. V. Kennon, and Edith I. Newcomb, "An Inventory of English Constructions with Measures of Their Importance." Teachers College Record, 28: 580-610, February, 1927.
3 Edward L. Thorndike and Irving Lorge, The Teacher's Word Book of 30,000 Words. Bureau of Publications, Teachers College, Columbia University, New York, 1944.
4 Irving Lorge and Edward L. Thorndike, A Semantic Count of English Words. The Institute of Educational Research, Teachers College, Columbia University, New York. (Hectographed)
5 Dr. Thorndike and the author wish to express appreciation for the cooperation of the New York City Work Projects Administration, 665-97-3-55 "Comprehensive Testing Program."
6 The grade equivalents thus supplied had, however, a fundamental weakness. If the published grade equivalents did not show equivalents for very high or very low scores, the grade equivalent was reported as the highest or lowest tabled equivalent. Thus, on the New York State Reading Progress Test, scores higher than 124 were reported as 14.1 and scores lower than 50 were reported as 5.8. Similarly, on the New Stanford test the highest grade score was 10.0; on the Stanford, 11.0; and on the Iowa, 12.0.
7 University of the State of New York, The State Education Department. A Manual for Use and Interpretation of the New York State Reading Progress Test for Grades 7 through 12 (Form A). The University of the State of New York Press, Albany, N. Y. 1940.
8 G. L. Kuder and M. W. Richardson, "The Theory of the Estimation of Test Reliability." Psychometrika, 2: 151-60, September, 1937. Formula 8 was used.