RESEARCH ON READING A SECOND LANGUAGE

Evelyn Hatch*

Department of English, University of California,

Los Angeles
Abstract. The experimental research on reading a second language concerns three
questions: 1) Should initial reading he taught in the child's first language or in the
second language when the second is the language of instruction? 2) What, other
than "inadequate grasp of the language," accounts for slower reading speed and
lowered comprehension when reading a second language? 3) What are the most
effective methods for teaching reading in a second language? The literature
contrasting the reading behavior of native and non-native speakers of English in
each of these areas is reported and discussed.

Research on reading a second language has been directed by three very

practical questions: 1) Should initial reading be taught in the child's first language
or in the second language (if that is the language of instruction)? 2) What, other
than some such general shortcoming as "inadequate grasp of the language,"
accounts for slower reading speed and lowered comprehension when reading a
second language? 3) If causes can be identified, can the information be translated
into classroom activities that will improve the student's reading speed and
comprehension? If not, what procedures make sense in teaching reading in the
second language?
WHICH LANGUAGE FOR INITIAL READING?
A great deal of rhetoric has been directed at the first question, but
surprisingly little research has been reported. It should be easier to learn to read in a
language you know well than one you don't. If initial reading means learning the
phoneme-grapheme correspondences (the alphabetic principle), then it would be
helpful if the child knew the sounds of the language to start with. For example, if
he hears no difference between /r/ and /I/ in Englishand many Japanese students
don'tthen it seems he would have difficulty learning correspondences of letters to
sounds he cannot distinguish. If he hears no difference between /i/ and /I/and
* Request for reprints should be sent to the author, Department of English, University of
California, Los Angeles, Calif. 90024

Downloaded from jlr.sagepub.com by guest on May 2, 2016

54

Journal of Reading Behavior 1974 VI, 1

many Spanish-speaking children don't-it should make it more difficult to teach

him the letters for these sounds. Also, if his syntactic development is at the "no
want read" stage, it would be difficult for him to use his knowledge of the language
to predict the syntax of "Mother said, 'I see something pretty.' "
To verify this, Modiano (1968) conducted an experiment with phildren from
three Mexican Indian tribes. Children in the experimental group were taught
reading in the vernacular. When they had mastered the primers, they entered first
grade where all reading texts were in Spanish. The students were evaluated against
control groups both on formal tests and informally by their teachers. Teachers felt
the experimental group showed greater comprehension in reading the second
language (Spanish) than students who had been taught from the beginning in
Spanish. And the test data supported the teachers' judgments.
The Tarascan study (Barrera-Vasquez, 1953) showed that children introduced
to reading in their first language were better readers by the end of the second year
than students who had all their instruction in Spanish, their second language.
Osterberg (1961) found that children taught to read in their local dialect of
Swedish first and then transferred to standard Swedish were able to read standard
Swedish better than those who began reading in the standard dialect. Thonis
(1970), discussing the Marysville project, claimed success for children taught to
read in Spanish first and then transferred to reading English. She states that
children, once they have learned what reading is all about, do not have to learn to
read again in the second language; they only have to learn a new code.
These studies suggest that a wise policy would be to introduce reading in the
child's first language or dialect and then transfer to the second language. The claim
is that there is a strong transfer in skills; that reading a new code is not a difficult
new task. That is, once having learned that the letter a stands for /a/ in Spanish, it
will not be an insurmountable problem to learn that it can stand for /as/ or /a/ or
/e/ or //, /a/ or /o/ in English. While ESL (English as a Second Language) teachers
would consider such splits as extremely difficult in pronunciation, in vocabulary,
and even in syntax, the issue is ignored when it comes to teaching reading. The
transition, it is claimed, is not as traumatic as learning the correspondences first in
English.
There are, of course, findings to support the other side as well. The St.
Lambert study (Lambert et al, 1970), the Toronto study (Barik & Swain, 1972),
and the Culver City project (Cathcart, 1972; Cohen, 1973) seem to suggest that
learning to read in a new language is not as difficult as we might imagine. After two
years, both of the Canadian projects found that English-speaking children receiving
all school instruction in French do as well as the control groups in reading readiness
and reading achievement. They do report a lag in the first year. The Culver City
project shows the same results for English-speaking children instructed in Spanish.
There are dangers in trying to compare such studies. Children in some
programs were volunteered for second-language classes; others involved compulsory
schooling in the second language. In some programs children knew the second

Downloaded from jlr.sagepub.com by guest on May 2, 2016

Hatch

55

language well; in others they did not. In some, "reading" means readiness activities,
and in others it means reading test scores. The question, then, on using the first or
second language for initial iristruction is still open. Since it is a political issue,
emotional commitment to one side or the other is strong. As more and more
bilingual programs are evaluated, we should have more definite answers.
WHAT COMPLICATES READING A SECOND LANGUAGE?
The second question is addressed to students who have learned to read their
second language but show slower reading speed and lower comprehension scores in
the second language. MacNamara (1967) documented this in a series of studies with
twelve- and thirteen-year-old English-Irish bilinguals who had received most of their
school instruction in Irish, their second language. He found they read faster in their
first language than in their second and, in addition, they were able to solve
problems written in English more accurately than identical problems written in
Irish, even though tests showed they understood the vocabulary and syntax of the
Irish passages.
Is there anything specific that we can say about these differences in speed and
comprehension? The following research studies, involving either upper-elementary
subjects or university-level ESL students, were conducted in an attempt to discover
some of the more specific variables involved.
Tullius (1971) suggested that one reason ESL students take longer to read a
passage is that they make more eye fixations per line and they frequently regress to
check back on information when they do not understand what they read. To test
this, he conducted an eye movement study with university ESL students. He found,
to his surprise, that they did neither. Instead, the difference in their eye movements
when compared to that of monolingual students was in the duration of each
fixation. But what accounts for the longer fixation period is not determined.
One of the suggestions most frequently made (aside from "inadequate grasp
of the language") has been articulation difficultythat students have motor
problems with the less familiar language. But this has been discounted (Kolers,
1966) for silent reading of second languages. Serpell (1968), however, claimed that
pronunciation, if not articulation, was important in silent reading. He found Ss who
did not hear the distinction between r and / (Bantu speakers in this case) misread
and misinterpreted light as right, cloud as crowd, lip as rip, etc. Using his technique,
one study (Hatch, 1971) constructed a test for fourth-grade Spanish-speaking
children enrolled in two English-instruction schools. Multiple-choice items tapped
distinctions in English phonology not present in Spanish. For example, the letter a
is pronounced /a/ in Spanish. Would children, on seeing the word cat, mispronounce it as cot and then assign it the meaning of coti There is no /i, 1/ distinction
in Spanish. Would children misread answers to items such as: "Which of these are
parts of your body? hands, when, fit, eyes, shirt" or "Which of these are colors?
blue, happy, yellow, grin"! It was found, just as Serpell had, that children did

Downloaded from jlr.sagepub.com by guest on May 2, 2016

56

Journal of Reading Behavior 1974 VI, 1

misread and misinterpret words in this fashion. They thought/?? was feet and grin
was green. Of course, children don't normally read passages set up to catch this
type of problem. It's not likely that in a passage about tuna boats, the student will
misread "The ship docked at the harbor" as "The sheep docked at the harbor."
However, the results show that phonological interference does occur even in silent
reading.
MacNamara (1967) tried to separate the effect of three factors on reading
speed: articulation time, word recognition time, and time for processing syntax.
Using the information from several experiments, he estimated that his 5s took
3.900 seconds per 20 words longer to read in their second language than in their
first. Of this, he claimed 1.799 seconds was due to articulation and .785 was due to
syntax; the remaining 1.316 seconds he attributed to word recognition and other
factors. It seems strange that so little time was due to syntax. In one of his studies,
MacNamara had worked out an ingenious way to get at the effect of syntax (or at
least word order probabilities). Using Miller and Selfridge's (1950) procedure, he
constructed passage which closely approximated English syntax (e.g., road in the
country was insane especially in dreary rooms where they have some book to buy
for studying . . . ) and passages where word order was random (e.g., house reins
women bought scream especially much said cake love that school to a they in is
. . . ) . He also prepared passages for Irish, the second language. He hypothesized that
Ss would read the close approximations to English rapidly since the passages would
be somewhat predictable, and they would read the random order slowly. Secondly,
Ss reading the approximations to Irish would show no speed difference for the two
versions. That is, the syntax of any approximation to Irishwhether close or
random-would be "Greek" to them. His predictions held. The fact that Ss were
not able to use their knowledge of Irish syntax for prediction seems to show that
syntax is more important than his time allotment suggests.
Using the Corcoran (1966) procedure, Hatch, Polin, & Part (1970) approached the effect of syntax somewhat differently. If you turned to the beginning of this
paper and crossed out as quickly as possible every letter e on the first page, you, as
a native speaker of English, would miss a large number of e's. We found that ESL
students at the university were highly successful at the task, much more so than
native speakers of English. Once we started looking at where letters went unnoticed
for the two groups, an interesting pattern appeared. First, Anglos marked letters
when they appeared in content words but not in function words. The ESL students
marked them everywhere; they paid as much attention to letters in words which
show grammatical relationships (prepositions, articles, conjunctions, etc.) as they
did to content words. Secondly, native speakers crossed out letters which appeared
in stressed syllables. That is, asked to cross out the letter a, they marked the first a
but not the second in vocabulary; they marked the second but not the first in
apparently. ESL students showed no pattern as to stressed or unstressed syllables.
We are not sure what the results of this study mean. Goodman (1970),
Hochberg (1970), and others have suggested that the native speaker approaches

Downloaded from jlr.sagepub.com by guest on May 2, 2016

Hatch

57

reading like a "psycholinguistic guessing game." He scans a line and fixes at a point
to permit eye focus. He picks up graphic cues and makes a guess-a prediction
about what appears on the printed page. While his guess is partly based on graphic
cues, it is also subject to his knowledge about the language and what he has read up
to that point. If his guess makes semantic and syntactic sense, he continues to read.
If it doesn't, he rechecks and makes an amended guess. Neisser (1967) even suggests
that at each focus point, the reader must be able to recognize which features seen in
peripheral vision are uninformative in order to move to the next fixation point.
ESL students obviously are unable to make such predictions rapidly. They are
unable to recognize the uninformative features because to them all features are
informative. We assume they need function words in order to make their guesses
about what is being read. The study shows that native speakers of English use their
knowledge of English syntax and stress and that ESL students either do not have
that information or are unable to use it.
In a pilot study, Johnston (1972) found that university ESL students do not
consistently attend to graphic cues which signal stress and intonation information.
This allowed them to misinterpret phrases in a variety of ways. For example, after
reading a sentence from a passage on the San Diego Zoo about Monkey Mesa and
the Great Ape Grotto, students ignored the capital letters on Great Ape Grotto and
changed grotto from a noun to a verb; they thought the great ape was grottoing.
Having read about the Children's Zoo, where one can pet friendly little deer,
camels, and backyard animals, they responded that you could pet dear little camels.
5s were given unlimited time and line reference to check their answers. And they
were given information on the vocabulary items. For example, a grotto was
mentioned earlier in the passage as the place where apes live. Johnston is currently
reworking her study to look more closely at pronoun reference, word group errors,
and misreading of clause groups.
What information do all these studies give us to the second question? If
students 1) have difficulty with articulation, 2) do not recognize words, 3) cannot
predict syntax, 4) ignore graphic cues, they will read slowly and with lowered
comprehension. In other words, we are back at "inadequate grasp of the language."
We must teach the student phonology, vocabulary, and syntax-that is, we must
teach him to "grasp the language."

WHAT HAVE WE LEARNED FOR THE CLASSROOM?

This brings us to question three-classroom application of the research
findings. In initial reading, the research shows that phonological interference from
the first language does occur. Rather than struggle with teaching letters for sounds
that students may neither be able to discriminate nor produce, teachers have turned
to the basal (whole word) approach or to the "linguistic" approach (pattern
recognition). Pattern recognition (fat, sat, bat, cat, mat, pat, etc.) forms the basis of
the Miami Linguistic Readers, probably the most widely used set of materials for

Downloaded from jlr.sagepub.com by guest on May 2, 2016

58

Journal of Reading Behavior 1974 VI, 1

beginning ESL students. The approach seems to work fairly well */comprehension
is also stressed and // the method is supplemented with experience materials and
some phonics work.
The research also suggests that some reading improvement materials for
advanced ESL students are more valuable than others. Tullius' work shows that we
can discard those exercises that emphasize widening eye span. MacNamara's
experiments support word recognition exercises. However, we have found that
university ESL students are very good at recognition tasks such as the following
(Harris, 1966): sing/sink sins sang sing sign; cheese/crease choose chess cheats
cheese, where the student matches the test word with the identical word in the list.
They are also good at word-pair tasks such as: poor day poor pay S D; new chair
new chair S D, where they must check S if the pair are identical and D if they are
different. They do not excel when asked to find a synonym: shut/watch close sleep
need; speak/point talk hope see. Nor are they able to perform a same-different task
for synonyms: stop go S D speak talk S D with any speed. While word recognition
is important, one should not waste time on visual perception since this is a skill
students already possess at this level. Instead, once basic skills in reading are
acquired, we should relax our emphasis on recognition of small units and
concentrate on the larger process.
Good readers are able to create a grammatical structure out of the series of
images their eye fixations give them. Some people call this "reading by structure."
Amble and Muehl (1966) found that extensive practice in reading short phrases was
effective with Anglo fifth grade children. Palister (1968) has suggested we give ESL
students practice in reading by phrases. His system requires rewriting passages so
that each column is one phrase wide. This doesn't mean that students read a
column per eye fixation; instead it gives students information on what words
should be grouped together. In a sense, it teaches the syntax of the passage.
Good readers are able to predict what they will read on the basis of their
knowledge of English syntax and the previously read material in a passage.
Suggestions have been made to use cloze passages (where every fifth word is
omitted or where- selected content words are omitted) to force ESL students to
make predictions about what might be "seen" in the blank.
Another suggestion has been the "outlandish proposal" of Burling (1968), a
system used successfully to teach "reading German" and "reading French" to
graduate students. Assume you wished to learn enough French to pass the ETS
exam. The first step of the procedure is to read an English passage in which the
only change has been to put it into French word order. The second step might be to
insert French articles in front of all the nouns. The next step might be to change
English-French cognates to their French spelling. Gradually, English would be
changed into French. In a sense it is set up to teach syntax; it shows what is
grouped together (that the articles go with the nouns, for example). The method
has been tested and found successful, but it is still too outlandish for many ESL

Downloaded from jlr.sagepub.com by guest on May 2, 2016

Hatch

59

teachers to accept, particularly those who have been conditioned against anything
that looks like translation.
Simplified texts have also been suggested as a bridge to more difficult reading.
Rosensweig (1973) ran a pilot study on the effect of simplified vs. original texts
with university ESL students. His findings showed that students not only did better
on comprehension questions covering simplified versions but that they also enjoyed
the simplified short stories more than the originals. Simplified stories supposedly
are written with controlled vocabulary and simple syntax. The simple syntax
frequently turns out to be more difficult than imagined since the control is usually
in terms of sentence length only.
Some teachers feel that we should relegate all manipulative materials to a
secondary place. They feel primary attention must be paid to "active organization"
of what is read. Their concern is with training students to pick out relevant points
in their reading and to organize them for recall. The methods usually reflect the
classic study technique SQ4R (survey, question, read, recite, repeat, review). Many
ESL teachers feel that this does not attack the "real" language problems of their
students.
Finally, Kennedy (1973) has argued against most ESL reading materials on
the basis of content and student interest. She compared the content of ESL texts
and materials created by ESL students using the LEA (language experience
approach). With this approach, ESL students record anecdotes, short stories, or talk
about themselves and their personal problems. These are typed and distributed
either to individual students or to the whole class to be used as reading material.
The Cambridge Learning Center has collected two volumes of reading material using
this approach.
The research on reading a second language is both interesting and confusing.
I'm not certain that it would be less confusing if there were more of it. Yet, there
are many areas which need to be investigated. Answers for one age group or one
proficiency level are not the answers for another. Considering the number of
children who are enrolled in our schools for whom English is a second language and
the number of adults who are studying English in night schools, it is surprising how
little attention has been paid to their reading problems. The three questions
reported on here may not even be the important questions to ask. But once we have
found the important issues, we will be on our way, hopefully, toward finding ways
to help students who must do the majority of their reading in a second language.

REFERENCES
AMBLE, B. & MUEHL, S. Perceptual span training and reading achievement.
Journal of Educational Psychology, 1966, 57, 192-206.

Downloaded from jlr.sagepub.com by guest on May 2, 2016

60

Journal of Reading Behavior 1974 VI, 1

BARIK, H. C. & SWAIN, M. Bilingual education project, interim report on the

1972 testing programme. 1972, Modern Language Center, Toronto, Canada.
BARRERA-VASQUEZ, A. The Tarascan project in Mexico, UNESCO Monographs
on Fundamental Education, 1953, 8, 77-86.
BURLING, R. Some outlandish proposals for the teaching for foreign languages.
Language Learning, 1968, 18, 61-76.
CATHCART, R. L. Report on a group of Anglo children after one year of
immersion in instruction in Spanish. Unpublished master's thesis, University
of California, Los Angeles, 1972.
COHEN, A. D. The Culver City Spanish immersion program-the first two years,
UCLA Workpapers in Teaching English as a Second Language, 1973, 7,
65-74.
CORCORAN, D. W. An acoustic factor in letter cancellation. Nature, 1966, 210,
39-43.
GOODMAN, K. S. Reading: a psycholinguistic guessing game. In D. V. Gunderson
(Ed.) Language and reading, an interdisciplinary approach. Washington, D. C.:
Center for Applied Linguistics, 1970, 107-122.
HARRIS, D. P. Reading improvement exercises for students of English as a second
language. Englewood Cliffs, New Jersey: Prentice Hall, 1966.
HATCH, E. Interference problems in reading a second language. Paper presented at
the meeting of the California TESOL Association, San Diego, 1971.
HATCH, E., POLIN, P., & PART, S. Acoustic scanning or syntactic processing.
Paper presented at the meeting of the Western Psychological Association, San
Francisco, April, 1970.
HOCHBERG, J. Attention organization and consciousness. In D. I. Mostofsky (Ed.)
Attention: Contemporary theory and analysis. New York: AppletonCentury-Crofts, 1970, 99-124.
JOHNSTON, G. A. Some effects of acoustic input on reading comprehension.
Unpublished master's thesis, University of California, Los Angeles, 1972.
KENNEDY, K. A. A guide to using language experience with adults. Cambridge
Learning Center, Cambridge, Mass., 1973.
KOLERS, P. A. Reading and talking bilingually. American Journal of Psychology,
1966, 79, 357-376.
LAMBERT, W. E., JUST, M., & SEGALOWITZ, N. Some cognitive effects of
following the curricula of grades one and two in a foreign language. Language
Research Group, McGill University. Mimeograph, 1970.
MacNAMARA, J. Comparative studies of reading and problem solving in two
languages. Language Research Group, McGill University, 1967.
MacNAMARA, J., & KELLAGHAN, T. The teaching of reading in a second
language. In Reading instruction: An international forum, Proceedings of the
1st World Congress on Reading, Paris: UNESCO, 1966.
MILLER, G. A. & SELFRIDGE, J. A. Verbal context and the recall of meaningful
material. American Journal of Psychology, 1950, 63, 176-185.
MODIANO, N. National or mother language in beginning reading: a comparative
study. Research in the Teaching of English, 1968, 1, 32-43.
NEISSER, V. Cognitive psychology. New York: Appleton-Century-Crofts, 1967.

Downloaded from jlr.sagepub.com by guest on May 2, 2016

Hatch

61

OSTERBERG, T. Bilingualism and the first school languagean educational

problem illustrated by results from a Swedish dialect area. Umea, Sweden:
Vasterbottens Tryckeri AB, 1961.
PLAISTER, T. Reading instruction for college level foreign students, TESOL
Quarterly, 1968, 2, 164-168.
ROSENSWEIG, F. Literature: an experiment in simplified and original texts.
Unpublished paper, University of California, Los Angeles, 1973.
SERPELL, R. Selective attention and interference between first and second
languages. H.D.R.U. Reports, University of Zambia, 1968.
THONIS, E. Teaching reading to non-English speakers, New York: Macmillan
Company, 1970.
TULLIUS, J. Analysis of reading skills of non-native speakers of English.
Unpublished master's thesis, University of California, Los Angeles, 1971.

Downloaded from jlr.sagepub.com by guest on May 2, 2016