WHAT READING DOES

FOR THE MIND

BY ANNE E. CUNNINGHAM AND KEITH E. STANOVICH

R EADING HAS cognitive consequences that extend

beyond its immediate task of lifting meaning from
a particular passage. Furthermore, these consequences
warding early reading experiences that lead to less in-
volvement in reading-related activities. Lack of expo-
sure and practice on the part of the less-skilled reader
are reciprocal and exponential in nature. Accumulated delays the development of automaticity and speed at
over timespiraling either upward or downward the word recognition level. Slow, capacity-draining
they carry profound implications for the development word recognition processes require cognitive re-
of a wide range of cognitive capabilities. sources that should be allocated to comprehension.
Concern about the reciprocal influences of reading Thus, reading for meaning is hindered; unrewarding
achievement has been elucidated through discussions reading experiences multiply; and practice is avoided
of so-called Matthew effects in academic achieve- or merely tolerated without real cognitive involve-
ment (Stanovich, 1986;Walberg & Tsai, 1983).The term ment.
Matthew effects is taken from the Biblical passage The disparity in the reading experiences of children
that describes a rich-get-richer and poor-get-poorer of varying skill may have many other consequences for
phenomenon.Applying this concept to reading, we see their future reading and cognitive development.As skill
that very early in the reading process poor readers, develops and word recognition becomes less resource
who experience greater difficulty in breaking the demanding and more automatic, more general lan-
spelling-to-sound code, begin to be exposed to much guage skills, such as vocabulary, background knowl-
less text than their more skilled peers (Allington, 1984; edge, familiarity with complex syntactic structures,
Biemiller, 1977-1978). Further exacerbating the prob- etc., become the limiting factor on reading ability
lem is the fact that less-skilled readers often find them- (Chall, 1983; Sticht, 1979). But the sheer volume of
selves in materials that are too difficult for them reading done by the better reader has the potential to
(Allington, 1977, 1983, 1984; Gambrell,Wilson, & Gantt, provide an advantage even here ifas our research
1981). The combination of deficient decoding skills, suggestsreading a lot serves to develop these very
lack of practice, and difficult materials results in unre- skills and knowledge bases (Cunningham & Stanovich,
1997; Echols, West, Stanovich, & Zehr, 1996; Stanovich
Anne E. Cunningham is visiting associate professor & Cunningham, 1992, 1993). From the standpoint of a
in cognition and development in the graduate school reciprocal model of reading development, this means
of education at the University of California, Berke- that many cognitive differences observed between
ley. Her research examines the cognitive and motiva- readers of differing skill may in fact be consequences
tional processes that underlie reading ability and the of differential practice that itself resulted from early
cognitive consequences of reading skill and engage- differences in the speed of initial reading acquisition.
ment. Keith E. Stanovich is professor of applied psy- The increased reading experiences of children who
chology at the University of Toronto/Ontario Insti- master the spelling-to-sound code early thus might
tute for Studies in Education. His recent awards in- have important positive feedback effects that are de-
clude the Sylvia Scribner Award from the American nied the slowly progressing reader. In our research, we
Educational Research Association and the Oscar S. have begun to explore these reciprocal effects by ex-
Causey Award from the National Research Confer- amining the role that reading volume plays in shaping
ence for his distinguished and substantial contribu- the mind and will share many of our findings in this ar-
tions to literacy research. ticle.
We should say at the outset that the complexity of
This research was supported by a Spencer Founda- some of the work we will describe in this article was
tion Small Grant to Anne E. Cunningham and grant necessitated in large part by the fact that it is difficult
No. 410-95-0315 from the Social Sciences and Hu- to tease apart the unique contribution that reading vol-
manities Research Council of Canada to Keith E. ume affords. One of the difficulties is that levels of
Stanovich. reading volume are correlated with many other cogni-

1 AMERICAN EDUCATOR/AMERICAN FEDERATION OF TEACHERS SPRING/SUMMER 1998

tive and behavioral characteristics. Avid readers tend adult speech in two contexts vary-
to be different from nonreaders on a wide variety of ing in formality. The words used in
cognitive skills, behavioral habits, and background vari- the different contexts were analyzed
ables (Guthrie, Schafer, & Hutchinson, 1991; Kaestle, according to a standard frequency
1991; Zill & Winglee, 1990). Attributing any particular count of English (Carroll, Davies, &
outcome to reading volume is thus extremely difficult. Richman, 1971). This frequency
count ranks the 86,741 different
word forms in English according to
Theoretical Reasons To Expect their frequency of occurrence in a
Positive Cognitive large corpus of written English. So,
for example, the word the is ranked
Consequences from number 1, the 10th most frequent
word is it, the word know is ranked
Reading Volume 100, the word pass is ranked 1,000,
In certain very important cognitive domains, there the word vibrate is 5,000th in fre-
are strong theoretical reasons to expect a positive and quency, the word shrimp is 9,000th in
unique effect of avid reading. Vocabulary development frequency, and the word amplifier is
provides a case in point. Most theorists are agreed that 16,000th in frequency.The first column,
the bulk of vocabulary growth during a childs lifetime labeled Rank of Median Word, is simply
occurs indirectly through language exposure rather the frequency rank of the average word
than through direct teaching (Miller & Gildea, 1987; (after a small correction) in each of the
Nagy & Anderson, 1984; Nagy, Herman, & Anderson, categories. So, for example, the average
1985; Sternberg, 1985, 1987). Furthermore, many re- word in childrens books was ranked
searchers are convinced that reading volume, rather 627th most frequent in the Carroll et
than oral language, is the prime contributor to indi- al. word count; the average word in
vidual differences in childrens vocabularies (Hayes, popular magazines was ranked 1,399th
1988; Hayes & Ahrens, 1988; Nagy & Anderson, most frequent; and the average word
1984; Nagy & Herman, 1987; Stanovich, 1986). in the abstracts of scientific articles
The theoretical reasons for believing that reading had, not surprisingly, a very low rank
volume is a particularly effective way of ex- (4,389).
panding a childs vocabulary derive from What is immediately apparent is
the differences in the statistical distribu- how lexically impoverished is most
tions of words that have been found be- speech, as compared to written lan-
tween print and oral language. Some of guage. With the exception of the spe-
these differences are illustrated in Table 1, cial situation of courtroom testimony,
which displays the results of some of the re- the average frequency of the words in
search of Hayes and Ahrens (1988), who all of the samples of oral speech is quite
have analyzed the distributions of words low, hovering in the 400-600 range of ranks.
used in various contexts. The relative rarity of the words in childrens books is,
The table illustrates the three different in fact, greater than that in all of the adult conversa-
categories of language that were analyzed: tion, except for the courtroom testimony. Indeed, the
written language sampled from genres as words used in childrens books are considerably
difficult as scientific articles and as sim- rarer than those in the speech on prime-time adult
ple as preschool books; words spoken on television. The categories of adult reading matter
television shows of various types; and contain words that are two or three times rarer than

SPRING/SUMMER 1998 AMERICAN EDUCATOR/AMERICAN FEDERATION OF TEACHERS 2

 more rare words in them than does adult prime-time
Table 1 television and the conversation of college graduates.
Selected Statistics for Major Sources of Spoken and Popular magazines have roughly three times as many
Written Language (Sample Means) opportunities for new word learning as does prime-
time television and adult conversation.Assurances by
Rank of Rare Words some educators that What they read and write may
Median Word per 1000 make people smarter, but so will any activity that en-
I. Printed texts gages the mind, including interesting conversation
Abstracts of scientific articles 4389 128.0 (Smith, 1989) are overstated, at least when applied to
Newspapers 1690 68.3 the domain of vocabulary learning. The data in Table
Popular magazines 1399 65.7 1 indicate that conversation is not a substitute for
Adult books 1058 52.7 reading.
Comic books 867 53.5
Childrens books 627 30.9
It is sometimes argued or implied that the type of
Preschool books 578 16.3 words present in print but not represented in
speech are unnecessary wordsjargon, academic
II. Television texts doublespeak, elitist terms of social advantage, or
Popular prime-time adult shows 490 22.7 words used to maintain the status of the users but
Popular prime-time childrens shows 543 20.2
Cartoon shows 598 30.8
that serve no real functional purpose. A considera-
Mr. Rogers and Sesame Street 413 2.0 tion of the frequency distributions of written and
spoken words reveals this argument to be patently
III. Adult speech false. Table 2 presents a list of words that do not
Expert witness testimony 1008 28.4 occur at all in two large corpora of oral language
College graduates to friends, spouses 496 17.3
(Berger, 1977; Brown, 1984), but that have apprecia-
Adapted from Hayes and Ahrens (1988). ble frequencies in a written frequency count (Fran-
cis & Kucera, 1982). The words participation, lux-
those heard on television. ury, maneuver, provoke, reluctantly, relinquish,
These relative differences in word rarity have di- portray, equate, hormone, exposure, display, in-
rect implications for vocabulary development. If variably, dominance, literal, legitimate, and infi-
most vocabulary is acquired outside of formal teach- nite are not unnecessary appendages, concocted to
ing, then the only opportunities to acquire new exclude those who are unfamiliar with them. They
words occur when an individual is exposed to a are words that are necessary to make critical dis-
word in written or oral language that is outside his tinctions in the physical and social world in which
current vocabulary.That this will happen vastly more we live. Without such lexical tools, one will be se-
often while reading than while talking or watching verely disadvantaged in attaining ones goals in an
television is illustrated in the second column of Table advanced society such as ours. As Olson (1986)
1. The column lists how many rare words per 1000 notes:
are contained in each of the categories. A rare word
It is easy to show that sensitivity to the subtleties of
is defined as one with a rank lower than 10,000; language are crucial to some undertakings.A person who
roughly a word that is outside the vocabulary of a does not clearly see the difference between an expression
fourth to sixth grader. For vocabulary growth to of intention and a promise or between a mistake and an
occur after the middle grades, children must be ex- accident, or between a falsehood and a lie, should avoid a
posed to words that are rare by this definition.Again, legal career or, for that matter, a theological one.
it is print that provides many more such word-learn- The large differences in lexical richness between
ing opportunities. Childrens books have 50 percent speech and print are a major source of individual dif-
ferences in vocabulary development. These differ-
ences are created by the large variability among chil-
Table 2 dren in exposure to literacy. Table 3 presents the data
Examples of words that do not appear in two large from a study of the out-of-school time use by fifth
corpora of oral language (Berger, 1977; Brown, 1984) but graders conducted by Anderson, Wilson, and Fielding
that have appreciable frequencies in written texts (1988). From diaries that the children filled out daily
(Carroll, Davies & Richman, 1971; over several months time, the investigators estimated
Francis & Kucera, 1982): how many minutes per day that individuals were en-
gaged in reading and other activities while not in
display literal school. The table indicates that the child at the 50th
dominance legitimate percentile in amount of independent reading was
dominant luxury reading approximately 4.6 minutes per day, or about a
exposure maneuver half an hour per week, over six times as much as the
equate participation child at the 20th percentile in amount of reading time
equation portray (less than a minute daily). Or, to take another example,
gravity provoke the child at the 80th percentile in amount of indepen-
hormone relinquish dent reading time (14.2 minutes) was reading over
infinite reluctantly twenty times as much as the child at the 20th per-
invariably centile.
Anderson et al. (1988) estimated the childrens read-

3 AMERICAN EDUCATOR/AMERICAN FEDERATION OF TEACHERS SPRING/SUMMER 1998

ing rates and used these, in conjunction with the In all, we developed two measures of adults reading
amount of reading in minutes per day, to extrapolate a volume and one for childrens reading volume. Briefly,
figure for the number of words that the children at var- the childrens measure, named the Title Recognition
ious percentiles were reading.These figures, presented Test (TRT), requires children to pick out the titles of
in the far right of the table, illustrate the enormous dif- popular childrens books from a list of titles that in-
ferences in word exposure that are generated by chil- cludes equal numbers of made-up titles.This task is easy
drens differential proclivities toward reading. For ex- to administer to large numbers of children, it does not
ample, the average child at the 90th percentile reads al- make large cognitive demands, and its results are reli-
most two million words per year outside of school, ableit is not possible for children to distort their re-
more than 200 times more words than the child at the sponses toward what they perceive as socially desirable
10th percentile, who reads just 8,000 words outside of answers. Because the number of wrong answers can be
school during a year. To put it another way, the entire counted against correct ones, it is possible to remove
years out-of-school reading for the child at the 10th the effects of guessing from the results (see Cunning-
percentile amounts to just two days reading for the ham & Stanovich, 1990; 1991; and Stanovich and West,
child at the 90th percentile! These dramatic differ- 1989 for a full description of these instruments and a
ences, combined with the lexical richness of print, act discussion of the logic behind them). The adults mea-
to create large vocabulary differences among children. sures, named the Author Recognition and Magazine
Recognition Test, have the same task requirements and
are described fully in Stanovich and West (1989).
Examining the Consequences A score on the Title Recognition Test, of course, is
of Differential Degrees not an absolute measure of childrens reading volume
and previous literacy experiences, but it does provide
of Reading Volume us with an index of the relative differences in reading
volume.This index enables us to ask what effects read-
It is one thing to speculate on how these differences
in reading volume may result in specific cognitive con- ing volume (rather than general reading comprehen-
sequences in domains like vocabulary; it is another to sion and word decoding ability) has on intelligence, vo-
demonstrate that these effects are occurring. In our re- cabulary, spelling, and childrens general knowledge. In
search, we have sought empirical evidence for the spe- short, it enables us to ask the question, does reading
cific effects of reading volume, effects that do not sim- in and of itselfshape the quality of our mind?
ply result from the higher cognitive abilities and skills The titles appearing on the TRT were selected from a
of the more avid reader. Although there are consider- sample of book titles generated in pilot investigations by
able differences in amount of reading volume in groups of children ranging in age from second grade
school, it is likely that differences in out-of-school read- through high school. In selecting the items that appear
ing volume are an even more potent source of the on any one version of the TRT, an attempt was made to
rich-get-richer and poor-get-poorer achievement pat- choose titles that were not prominent parts of classroom
terns. Therefore, we have sought to examine the reading activities in these particular schools. Because we
unique contribution that independent or out-of-school wanted the TRT to probe out-of-school rather than
reading makes toward reading ability, aspects of verbal school-directed reading, an attempt was made to choose
intelligence, and general knowledge about the world. titles that were not used in the school curriculum.
As part of this research program, our research group In our technical reports on this work, we have used a
has pioneered the use of a measure of reading volume powerful statistical technique known as hierarchical
that has some unique advantages in investigations of multiple regression to solve the interpretive problem
this kind (Cunningham and Stanovich, 1990; Stanovich that avid readers excel in most domains of verbal learn-
and West, 1989). ing and that, therefore, our measures of reading volume
might be spuriously correlated to a host of abilities
(Cunningham & Stanovich, 1990, 1991; Stanovich &
Table 3 Cunningham, 1992, 1993; Stanovich & West, 1989). We
Variation in Amount of Independent Reading have found that even when performance is statistically
equated for reading comprehension and general ability,
Independent Reading reading volume is still a very powerful predictor of vo-
% Minutes Per Day Words Read Per Year
cabulary and knowledge differences. Thus, we believe
98 65.0 4,358,000 that reading volume is not simply an indirect indicator
90 21.1 1,823,000 of ability; it is actually a potentially separable, indepen-
80 14.2 1,146,000 dent source of cognitive differences.
70 9.6 622,000
60 6.5 432,000
50 4.6 282,000 Reading Volume as
40 3.2 200,000 a Contributor to Growth
30 1.3 106,000
20 0.7 21,000 in Verbal Skills
10 0.1 8,000 In several studies, we have attempted to link chil-
2 0.0 0 drens reading volume to specific cognitive outcomes
Adapted from Anderson,Wilson, and Fielding (1988). after controlling for relevant general abilities such as
IQ. In a study of fourth-, fifth-, and sixth-grade children,
SPRING/SUMMER 1998 AMERICAN EDUCATOR/AMERICAN FEDERATION OF TEACHERS 4
we examined whether reading volume accounts for nature of these analyses is illustrated in a longitudinal
differences in vocabulary development once controls study that we have conducted (Cipielewski &
for both general intelligence and specific verbal abili- Stanovich, 1992). We addressed the question of
ties were invoked (Cunningham & Stanovich, 1991). whether reading volume can predict individual differ-
We employed multiple measures of vocabulary and ences in growth in reading comprehension from third
controlled for the effects of age and intelligence. We grade to fifth grade.We found that reading volume pre-
also controlled for the effect of another ability that dicted variance in fifth-grade reading comprehension
may be more closely linked to vocabulary acquisition ability after third-grade reading comprehension scores
mechanisms: decoding ability. Decoding skill might me- had been removed.Thus, in removing the contribution
diate a relationship between reading volume and a of reading comprehension in our adult studies, we are
variable like vocabulary size in numerous ways. High undoubtedly removing some of the variance in vari-
levels of decoding skill, certainly a contributor to ables such as vocabulary and general knowledge that is
greater reading volume, might provide relatively com- rightfully attributed to reading volume.
plete contexts for figuring out the meaning of words
during reading. Thus, reading volume and vocabulary
might be linked via their connection to decoding abil- Reading Volume and
ity: Good decoders read a lot and have the best con- Declarative Knowledge
text available for inferring new words. This potential
linkage was accounted for by statistically controlling In other studies, we have focused even more directly
for decoding ability prior to investigating reading vol- on content knowledge by addressing the issue of
ume. But we found that even after accounting for gen- Where Does Knowledge Come From?. Stanovich and
eral intelligence and decoding ability, reading volume Cunningham (1993) examined general ability, reading
contributed significantly and independently to vocabu- volume, and exposure to other media sources as deter-
lary knowledge in fourth-, fifth-, and sixth-grade chil- minants of individual differences in content knowl-
dren. edge. This study contained a particularly stringent test
These findings demonstrate that reading volume, al- of the role of reading volume and individual differ-
though clearly a consequence of developed reading ences in knowledge acquisition among 268 college stu-
ability, is itself a significant contributor to the develop- dents. We administered five different measures of gen-
ment of other aspects of verbal intelligence. Such rich- eral knowledge to the students. Then we stacked the
get-richer (and of course their converse, poor-get- deck against reading volume once again by statistically
poorer) effects are becoming of increasing concern in entering four measures of general ability before look-
the educational community (Adams, 1990; Chall, 1989) ing at the contribution of reading volume: high school
and are playing an increasingly prominent role in theo- grade-point average, performance on an intelligence
ries of individual differences in reading ability and test, an SAT-type mathematics test, and an adult reading
growth (Anderson, et al., 1988; Chall, Jacobs, & Bald- comprehension test. This set of tasks surely exhausts
win, 1990; Hayes, 1988; Hayes & Ahrens, 1988; Juel, the variance attributable to any general ability con-
1988, 1994; Stanovich 1986, 1989, 1993). struct; and, as one would expect, we found that general
In a study we conducted involving college students, ability accounted for a substantial proportion of vari-
we employed an even more stringent test of whether ance in the composite measure of general knowledge.
reading volume is a unique predictor of verbal skill Next we entered a composite measure of exposure to
(Stanovich & Cunningham, 1992). In this study we ex- television, but it did not account for any additional
amined many of the same variables as in our study of variance. However, a composite index of reading vol-
fourth- to sixth-grade students. However, we decided ume accounted for a substantial 37.1 percent of the
to stack the deck against reading volume by first re- variance when entered after the four ability mea-
moving any contribution of reading ability and sures and television exposure.
general intelligence. By structuring the analyses This pattern was replicated in each of the five
in this way, we did not mean to imply that read- measures of general knowledge we employed,
ing volume is not a determinant of reading com- including a homemade instrument we called
prehension ability. Indeed, we argue that there the Practical Knowledge Test. This task was
are grounds for believing that reading volume designed to address the criticism that our
facilitates growth in comprehension ability. other measures of general knowledge were
However, we wanted to construct the most too academicthat they tapped knowl-
conservative analysis possible by deliberately edge that was too esoteric or elitist and
allowing the comprehension measure to steal that was not useful in daily life. We didnt
some variance that is rightfully attributed to think this was true; many items on these
the measure of reading volume.The results of measures were mundane and concrete
our study again attest to the potency of read- questions such as, In what part of the
ing volume. We found that reading volume body does the infection called pneumo-
made a significant contribution to multiple nia occur? Nevertheless, in the Practical
measures of vocabulary, general knowledge, Knowledge Test, we made an effort to
spelling, and verbal fluency even after reading devise questions that were directly rel-
comprehension ability and nonverbal ability evant to daily living in a technological
had been partialed out. society in the late twentieth century;
One way of demonstrating the conservative for example, What does the carbure-

5 AMERICAN EDUCATOR/AMERICAN FEDERATION OF TEACHERS SPRING/SUMMER 1998

tor in an automobile do? If a substance is carcinogenic, associated with higher scores on the question, but tele-
it means that it is ______? After the Federal Reserve vision exposure was associated with lower scores.
Board raises the prime lending rate, the interest that you Scores among the group high in reading volume and
will be asked to pay on a car loan will generally in- low in television exposure were highest, and the low-
crease/decrease/stay the same? What vitamin is highly est scores were achieved by those high in television
concentrated in citrus fruits? When a stock exchange is exposure and low in reading volume. Our analyses
in a bear market, what is happening? and so forth. confirmed that these relationships were not due to dif-
The results indicated that the more avid readers in ferences in general ability.
our studyregardless of their general abilitiesknew Similarly, we have analyzed a variety of other mis-
more about how a carburetor worked, were more conceptions in a number of other different do-
likely to know who their United States senators were, mainsincluding knowledge of World War II, the
more likely to know how many teaspoons are equiva- worlds languages, and the components of the federal
lent to one tablespoon, were more likely to know what budgetand all of them replicate the pattern shown
a stroke was, and what a closed shop in a factory was, for this question. The cognitive anatomy of misinfor-
etc. One would be hard pressed to deny that at least mation appears to be one of too little exposure to
some of this knowledge is relevant to living in the print (or reading) and over-reliance on television for
United States in the late 20th century. information about the world. Although television
In other questions asked of these same students, we viewing can have positive associations with knowl-
attempted to probe areas that we thought might be edge when the viewing is confined to public televi-
characterized by misinformation. We then attempted sion, news, and/or documentar y material (Hall,
to trace the cognitive anatomy of this misinforma- Chiarello, & Edmondson, 1996; West & Stanovich,
tion. One such question concerned the sizes of the 1991; West et al., 1993), familiarity with the prime-
worlds major religions and was designed to assess time television material that defines mass viewing in
awareness of the multicultural nature of the modern North America is most often negatively associated
world.The question was phrased as follows:The 1986 with knowledge acquisition.
Encyclopedia Britannica estimates that there are ap- In another study, Stanovich, West, & Harrison
proximately nine hundred million people in the world (1995) examined a much older population in order
(not just the United States) who identify themselves as to investigate the extent to which age-related growth
Christians. How many people in the world (not just in knowledge can be accounted for by differences in
the United States) do you think identify themselves as reading volume. Although much research effort has
______? Space was then provided on the form for the been expended on describing cumulative growth in
subjects to make estimates of the number of Moslems, crystallized intelligence (e.g., acquired knowledge
Jews, Buddhists, Hindus, etc. such as vocabulary and general information), we
We will focus here on the estimates of Moslem and know little about the experiences that relate to
Jewish people because of our a priori hypothesis that knowledge growth in older individuals. For example,
availability effects caused by televised coverage of Is- educational experience (years in school) is a predic-
rael in the U.S. had skewed the perception of this ratio. tor of intellectual functioning in older individuals
While our samples median estimate of the number of (e.g., Schwar tzman, Gold, Andres, Arbuckle, &
Jewish people (20 million) was quite close to the actual Chaikelson, 1987). It is assumed that education
figure of 18 million according to the 1990 Universal Al- (which is received early in life) in part determines
manac, the number of estimated Moslemsa mean of the extent and quality of many intellectual activities
10 millionwas startlingly low (817 million is the esti- later in life. And it is presumably this intellectual ac-
mate in the Universal Almanac). For each participant tivity as one ages that is so crucial to the preserva-
in our study, we calculated the ratio of the Moslem to tion of cognitive capacities. Thus, while considerable
Jewish estimates to see how many students were aware development of cognitive skills and abilities can re-
of the fact that the number of Moslems is an order of sult from formal educational experiences, it is the
magnitude larger (the actual estimated ratio is approxi- lifetime use of these skills that is assumed to have
mately 33:1 according to the World Almanac; 45:1 ac- the beneficial effect.
cording to the Universal Almanac). The median ratio In this study, Stanovich, et al. (1995) examined the
in our sample was 0.5.That is, 69.3 percent of our sam- performance of college students and senior citizens
ple thought that there were more Jewish people in the on general knowledge, vocabulary, working memory,
world than Moslems. syllogistic reasoning, and several measures of reading
This level of inaccuracy is startling given that ap- volume. The older individuals outperformed the col-
proximately 40 percent of our sample of 268 students lege students on the measures of general knowledge
were attending one of the most selective public insti- and vocabulary, but did significantly less well than
tutions of higher education in the United States (the the college subjects on the working memory and syl-
University of California, Berkeley). We have explored logistic reasoning tasks. This dissociation between
the correlates of this particular misconception in a va- fluid intelligence (all-purpose general problem-solv-
riety of ways. We looked at the performance on this ing capacity) and crystallized intelligence (general
question as a function of students level of reading vol- knowledge and vocabulary) is a standard finding in
ume and television watching. We observed a clear ef- the literature (Baltes, 1987; Horn & Hofer, 1992; Salt-
fect of reading volume on the scores on the question house, 1988). However, a series of analyses indicated
and a significant effect of television viewing, but the ef- that when measures of reading volume were used as
fects were in opposite directions! Reading volume was control variables, the positive relationships between

SPRING/SUMMER 1998 AMERICAN EDUCATOR/AMERICAN FEDERATION OF TEACHERS 6

age and vocabulary and age and declarative knowl- eleventh-grade reading comprehension ability had
edge were eliminated (in contrast, the negative rela- been partialed out! In contrast, we observed that first-
tionships between age and fluid abilities were largely grade intelligence measures do not uniquely predict
unchanged). Thus, the results of this study are consis- eleventh-grade reading volume in the same way. Thus,
tent with the conjecture thatin the domain of ver- this study showed us that an early start in reading is
bal abilitiesreading a lot can even help to compen- important in predicting a lifetime of literacy experi-
sate for the normally deleterious effects of aging! enceand this is true regardless of the level of read-
(See also, Smith, 1996.) ing comprehension ability that the individual eventu-
ally attains.
This is a stunning finding because it means that stu-
How Do We Become dents who get off to a fast start in reading are more
Avid Readers? likely to read more over the years, and, furthermore,
this very act of reading can help children compensate
Moving back again to the other end of the age spec- for modest levels of cognitive ability by building their
trum, we switch focus to the question: Given that life- vocabularly and general knowledge. In other words,
long reading habits are such strong predictors of ver- ability is not the only variable that counts in the devel-
bal cognitive growth, what is it that predicts these opment of intellectual functioning. Those who read a
habits? Weve been looking at reading volume as a pre- lot will enhance their verbal intelligence; that is, read-
dictor of reading comprehension and cognitive ability, ing will make them smarter.
but what predicts reading volume or avid reading?
It is generally agreed that comprehension ability and
reading volume are in a reciprocal relationship. In an The Reciprocal Effects of
attempt to tease apart this reciprocal relationship, we
explored the linkages between childrens first-grade Reading Volume
reading and cognitive abilities and eleventh-grade out- We can begin to sketch a view of the reciprocal in-
comes in a unique ten-year longitudinal study (Cun- fluences of early reading acquisition and reading vol-
ningham and Stanovich, 1997). Most of our earlier stud- ume as determinants of later reading comprehension
ies involved assessing contemporaneous relations, but and other cognitive abilities. Early success at reading
in this study, we examined the performance of a sam- acquisition is one of the keys that unlocks a lifetime of
ple of students who had been tested as first graders reading habits. The subsequent exercise of this habit
(see Stanovich, Cunningham, and Feeman, 1984). serves to further develop reading comprehension abil-
About one half of these students were available ten ity in an interlocking positive feedback logic (Juel, Grif-
years later for testing as eleventh graders. At this time, fith, & Gough, 1986; Juel, 1988; Snow, Barnes, Chandler,
we administered a set of reading comprehension, cog- Goodman, & Hemphill, 1991; Stanovich, 1986, 1993).Al-
nitive ability, vocabulary, and general knowledge tasks, though it is difficult to tease apart, we have attempted
as well as several measures of reading volume. Addi- to trace the increasing divergence in childrens reading
tionally, some standardized test scores from the inter- ability, as well as other cognitive outcomes, by examin-
vening period were available. We were therefore able ing both sides of the important role of reciprocal cau-
to examine what variables in the first grade predicted sation. Our longitudinal study has permitted us to ob-
these cognitive outcomes in the eleventh grade.We in- serve these effects, whereby children who get out of
terpreted the reading volume measures administered the gate quicklywho crack the spelling-to-sound
in the eleventh grade as cumulative indicators of vari- code early onappear to enter into a positive feed-
ance in reading volume that had taken place many back loop. One of the benefits of these reciprocating
years earlier.Thus, we viewed the measures as in some effects may be a level of participation in literacy activi-
sense retrospective indicators tapping the cumulative ties that leads to a lifetime habit of reading and thus
experiences and habits of the students some distance sets the stage for future opportunitiesopportunities
in time before actual assessment. As a result, we were not enjoyed by children who enter into this feedback
able to examine how far this retrospective feature loop more slowly.
could be stretched. A positive dimension of our research is that all of
We addressed the question of whether the speed of our studies have demonstrated that reading yields sig-
initial reading acquisition in the first grade could pre- nificant dividends for everyonenot just for the
dict later tendencies to engage in reading activities smart kids or the more able readers. Even the child
even after differences in general cognitive abilities with limited reading and comprehension skills will
were controlled, as some models of Matthew effects in build vocabulary and cognitive structures through
educational achievement would predict (Chall, Jacobs, reading.
& Baldwin, 1990; Juel, 1994; Stanovich, 1986).We statis- We can thus elicit two crucial messages from our re-
tically removed the contribution of eleventh-grade search findings. First, it is difficult to overstate the im-
reading comprehension ability, in order to remove the portance of getting children off to an early successful
direct association between reading volume and cur- start in reading. We must ensure that students decod-
rent reading ability. Then we examined the contribu- ing and word recognition abilities are progressing
tion of three standardized measures of first grade read- solidly. Those who read well are likely to read more,
ing ability (decoding, word recognition, and compre- thus setting an upward spiral into motion.
hension) and observed that all three measures pre- Second, we should provide all children, regardless of
dicted eleventh-grade reading volume even after their achievement levels, with as many reading experi-

7 AMERICAN EDUCATOR/AMERICAN FEDERATION OF TEACHERS SPRING/SUMMER 1998

ences as possible. Indeed, this becomes doubly impera- Horn, J. L., & Hofer, S. (1992). Major abilities and development in the
tive for precisely those children whose verbal abilities adult period. In R. J. Sternberg & C.A. Berg (Eds.), Intellectual devel-
opment (pp. 44-99). Cambridge: Cambridge University Press.
are most in need of bolstering, for it is the very act of Juel, C. (1988). Learning to read and write:A longitudinal study of fifty-
reading that can build those capacities.An encouraging four children from first through fourth grade. Journal of Educa-
message for teachers of low-achieving students is im- tional Psychology, 80, 437-447.
plicit here. We often despair of changing our students Juel, C. (1994). Learning to read and write in one elementary
school. New York: Springer-Verlag.
abilities, but there is at least one partially malleable Juel, C., Griffith, P. L., & Gough, P. B. (1986). Acquisition of literacy: A
habit that will itself develop abilitiesreading! l longitudinal study of children in first and second grade. Journal of
Educational Psychology, 78, 243-255.
REFERENCES Kaestle, C. F. (1991). Literacy in the United States. New Haven, CT.:
Yale Press.
Adams, M. J. (1990). Beginning to read: Thinking and learning about
print. Cambridge, MA: MIT Press. Miller, G. A., & Gildea, P. M. (1987). How children learn words. Scien-
Allington, R. L. (1977). If they dont read much, how they ever gonna tific American, 257(3), 94-99.
get good? Journal of Reading, 21, 57-61. Nagy, W. E., & Anderson, R. C. (1984). How many words are there in
Allington, R. L. (1983).The reading instruction provided readers of dif- printed school English? Reading Research Quarterly, 19, 304-330.
fering reading abilities. The Elementary School Journal, 83, 548- Nagy, W. E., Herman, P. A. (1987). Breadth and depth of vocabulary
559. knowledge: Implications for acquisition and instruction. In M. G.
Allington, R. L. (1984). Content coverage and contextual reading in McKeown & M. E. Curtis (Eds.), The nature of vocabulary acquisi-
reading groups. Journal of Reading Behavior, 16, 85-96. tion (pp. 19-35). Hillsdale, NJ: Erlbaum.
Anderson, R. C.,Wilson, P.T., & Fielding, L. G. (1988). Growth in reading Nagy,W. E., Herman, P.A. & Anderson, R. C. (1985). Learning words from
and how children spend their time outside of school. Reading Re- context. Reading Research Quarterly, 20, 233-253.
search Quarterly, 23, 285-303. Olson, D. R. (1986). Intelligence and literacy: The relationships be-
Baltes, P. B. (1987).Theoretical propositions of life-span developmental tween intelligence and the technologies of representation and com-
psychology: On the dynamics between growth and decline. Devel- munication. In R. J. Sternberg & R. K.Wagner (Eds.), Practical intel-
opmental Psychology, 23, 611-626. ligence (pp. 338-360). Cambridge: Cambridge University Press.
Berger, K. W. (1977). The most common 100,000 words used in con- Salthouse,T.A. (1988). Resource-reduction interpretations of cognitive
versations. Kent, Ohio: Herald Publishing House. aging. Developmental Review, 8, 238-272.
Biemiller,A. (1977-1978). Relationships between oral reading rates for let- Schwartzman, A., Gold, D., Andres, D., Arbuckle, T., & Chaikelson, J.
ters, words, and simple text in the development of reading achieve- (1987). Stability of intelligence:A 40-year follow-up. Canadian Jour-
ment. Reading Research Quarterly, 13, 223-253. nal of Psychology, 41, 244-256.
Brown, G. D. (1984). A frequency count of 190,000 words in the Lon- Smith, F. (1989). Overselling literacy. Phi Delta Kappan, 70, 5, 354.
don-Lund Corpus of English Conversation. Behavior Research Smith, M.C. (1996). Differences in adults reading practices and liter-
Methods, Instruments, & Computers, 16 502-532. acy proficiencies. Reading Research Quarterly, 31, 196-219.
Carroll, J. B., Davies, P., & Richman, B. (1971). Word frequency book. Snow, C. E., Barnes, W. S., Chandler, J., Goodman, I., & Hemphill, L.
Boston: Houghton Mifflin. (1991). Unfulfilled expectations: Home and school influences on
Chall, J. S. (1983). Stages of reading development. New York: McGraw- literacy. Cambridge, MA: Harvard University Press.
Hill. Stanovich, K. E. (1986). Matthew effects in reading: Some conse-
Chall. J.S. (1989) Learning to read:The great debate twenty years later. quences of individual differences in the acquisition of literacy.
A response to Debunking the great phonics myth. Phi Delta Kap- Reading Research Quarterly, 21, 360-407.
pan. 71, 521-538. Stanovich, K. E. (1989). Has the learning disabilities field lost its intelli-
Chall, J. S., Jacobs, V., & Baldwin, L. (1990). The reading crisis: Why gence? Journal of Learning Disabilities, 22, 487-492.
poor children fall behind. Cambridge, MA: Harvard University Stanovich, K. E. (1993). Does reading make you smarter? Literacy and
Press. the development of verbal intelligence. In H. Reese (Ed.), Advances
Cipielewski, J., & Stanovich, K. E. (1992). Predicting growth in reading in child development and behavior, 24, 133-180. San Diego, CA:
ability from childrens exposure to print. Journal of Experimental Academic Press.
Child Psychology, 54, 74-89. Stanovich, K. E., & Cunningham, A. E. (1992). Studying the conse-
Cunningham,A. E., & Stanovich, K. E. (1990).Assessing print exposure quences of literacy within a literate society:The cognitive correlates
and orthographic processing skill in children: A quick measure of of print exposure. Memory & Cognition, 20, 51-68.
reading experience. Journal of Educational Psychology, 82, 733- Stanovich, K. E., & Cunningham, A. E. (1993). Where does knowledge
740. come from? Specific associations between print exposure and in-
Cunningham, A. E., & Stanovich, K. E. (1991). Tracking the unique ef-
formation acquisition. Journal of Educational Psychology, 85, 211-
fects of print exposure in children: Associations with vocabulary,
229.
general knowledge, and spelling. Journal of Educational Psychol-
Stanovich, K. E., Cunningham, A. E., & Feeman, D. J. (1984). Intelli-
ogy, 83, 264-274.
Cunningham, A.E., & Stanovich, K. E. (1997). Early reading acquisition gence, cognitive skills, and early reading progress. Reading Re-
and its relation to reading experience and ability 10 years later. De- search Quarterly, 19, 278-303.
velopmental Psychology, 33, 6, 934-945. Stanovich, K. E., & West, R. F. (1989). Exposure to print and ortho-
Echols, L. D.,West, R. F., Stanovich, K. E., & Zehr, K. S. (1996). Using chil- graphic processing. Reading Research Quarterly, 24, 402-433.
drens literacy activities to predict growth in verbal cognitive skills: Stanovich, K. E., & West, R. F., & Harrison, M. (1995). Knowledge
A longitudinal investigation. Journal of Educational Psychology, growth and maintenance across the life span: The role of print ex-
88, 296-304. posure. Developmental Psychology, 31, 811-826.
Francis, W. N., & Kucera, H. (1982). Frequency analysis of English Sternberg, R. J. (1985). Beyond IQ: A triarchic theory of human intel-
usage: Lexicon and grammar. Boston: Houghton Mifflin. ligence. Cambridge: Cambridge University Press.
Gambrell, L. B.,Wilson, R. M., & Gantt,W. N. (1981). Classroom observa- Sternberg, R. J. (1987). Most vocabulary is learned from context. In M. G.
tions of task-attending behaviors of good and poor readers. Journal McKeown & M. E. Curtis (Eds.), The nature of vocabulary acquisi-
of Educational Research, 74, 400-404. tion (pp. 89-105). Hillsdale, NJ: Erlbaum Associates.
Guthrie, J.T., Schafer,W. D. & Hutchinson, S. R. (1991). Relation of docu- Sticht,T. (1979). Applications of the audread model to reading evalua-
ment literacy and prose literacy to occupational and societal char- tion and instruction. In L. B. Resnick & P.Weaver (Eds.), Theory and
acteristics of young black and white adults. Reading Research practice of early reading (pp. 209-226). Hillsdale, NJ: Erlbaum Asso-
Quarterly, 26, 30-48. ciates.
Hayes, D. P. (1988). Speaking and writing: Distinct patterns of word Walberg, H. J., & Tsai, S. (1983). Matthew effects in education. Ameri-
choice. Journal of Memory and Language, 27, 572-585. can Educational Research Journal, 20, 359-373.
Hayes, D. P., & Ahrens, M. (1988). Vocabulary simplification for chil- West, R. F., & Stanovich, K. E. (1991).The incidental acquisition of infor-
dren:A special case of motherese. Journal of Child Language, 15, mation from reading. Psychological Science, 2, 325-330.
395-410. West, R. F., Stanovich, K. E., & Mitchell, H. R. (1993). Reading in the real
Hall, V. C., Chiarello, K., & Edmondson, B. (1996). Deciding where world and its correlates. Reading Research Quarterly, 28, 34-50.
knowledge comes from depends on where you look. Journal of Ed- Zill, N., & Winglee, W. (1990). Who reads literature? Cabin John, MD:
ucational Psychology, 88, 305-313. Seven Locks Press.
SPRING/SUMMER 1998 AMERICAN EDUCATOR/AMERICAN FEDERATION OF TEACHERS 8