Download as pdf or txt
Download as pdf or txt
You are on page 1of 25

Teachers of English to Speakers of Other Languages, Inc.

(TESOL)

Does Second Language Instruction Make a Difference? A Review of Research


Author(s): Michael H. Long
Source: TESOL Quarterly, Vol. 17, No. 3 (Sep., 1983), pp. 359-382
Published by: Teachers of English to Speakers of Other Languages, Inc. (TESOL)
Stable URL: http://www.jstor.org/stable/3586253 .
Accessed: 22/06/2014 16:49

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Teachers of English to Speakers of Other Languages, Inc. (TESOL) is collaborating with JSTOR to digitize,
preserve and extend access to TESOL Quarterly.

http://www.jstor.org

This content downloaded from 185.2.32.21 on Sun, 22 Jun 2014 16:49:07 PM


All use subject to JSTOR Terms and Conditions
TESOL QUARTERLY, Vol. 17, No. 3, September 1983

Does SecondLanguage Instruction


Make a Difference?
A ReviewofResearch
MICHAEL H. LONG
ofHawaiiat Manoa
University

Does second language instruction promotesecond language acquisi-


tion? Some studiesconclude thatinstructiondoes not help (or even
that it is counter-productive);othersfindit beneficial.The picture
becomes clearerif two distinctionsare made. First,researchersmay
address one or both of two issues: the absolute effectof instruction,
on the one hand, and itsrelativeutility,on the other.Second, studies
need to be subclassifiedaccordingto whetheror notthecomparisons
they make involve controllingfor the total amount of instruction,
exposure,or instruction plus exposure-that is, forthetotalopportun-
ityto acquire the second language.
Observingthesedistinctions, a reviewof researchfindingsconcludes
thatthereis considerable (althoughnotoverwhelming)evidence that
instructionis beneficial 1) for children as well as adults, 2) for
beginning,intermediate,and advanced students,3) on integrativeas
well as discrete-pointtests, and 4) in acquisition-richas well as
acquisition-poorenvironments. These findingshave implicationsfor
theoriesof second language acquisition,such as Krashen'sMonitor
Theory,which make predictionsabout second language acquisition
with and withoutinstruction, and also for those involved in educa-
tionaladministration, programdesign,and classroomteaching.

INTRODUCTION
Of many important questions facing the language teaching profes-
sional, the most basic must be: Does instructionmake a difference?
Judging by the plethora of prescriptive papers at conferences and in
journals, one might presume that the answer was already known and
affirmative.In fact, few researchers have ever addressed the question,
and of four studies which have sought direct answers to it (by
comparing second language acquirers with and without instruction),
not one claims to have found evidence that instructionhelps.
The question may be approached through a variety of comparisons,
as listed in Table 1. Some refer to the absolute effect of instruction,
DOES INSTRUCTION MAKE A DIFFERENCE? 359

This content downloaded from 185.2.32.21 on Sun, 22 Jun 2014 16:49:07 PM


All use subject to JSTOR Terms and Conditions
othersto itsrelativeutility,
withthealternatives
beingeithersimple
exposure to the second language(SL) in use, a combinationof
or
instruction
plusexposure.Ananswerinvolvesshowinga causalrelation-
ship(or thelackofone) between,on theonehand,instruction (only),

TABLE1
Instruction, andSecondLanguage
Exposure, (SLA):
Acquisition
PossibleComparisons

Comparison issueAddressed
1. livs0i Effectofinstructioninpopulations withinstruction
only
2. EivsOi Effectofexposure inpopulationswithexposureonly
3. livsI Effectofamountofinstruction inpopulationswithinstruction
only
4. EivsE3 Effectof amountof exposurein populations withexposure
only
5. livsEi Relativeutility
ofinstruction
and exposure
6. livsEj(+ Ik) SLA process(sequenceofacquisition)
7. livsIi+ E3 SLA process(sequenceofacquisition)
8. EivsEi+ I; SLA process(sequenceofacquisition)
9. Ei+ OjvsEi+ Ij Effectofinstructionon populationswithexposure
10. li+ Ojvs li+ E Effectofexposureon populations withinstruction
11. vs + E onlyand thesametotalamount
of instruction
Relativeutility
I/ .I ofinstructionandexposure
12. Eivs [ I + E of exposureonlyandthesametotalamountof
Relativeutility
instruction
andexposure(Table2)
13. I + E I vs k +E i Relative utilityof differingamountsof instruction and
exposureinpopulations withthesamietotalamountofboth
(Table 3)
14. Ei + Il vsEi+ Ik Effectofamountofinstruction withthesame
on populations
amountofexposure(Table4)
15. li+ E1vsli + Ek Effectofamountofexposure on populationswiththesame
amountofinstruction (Table5)
16. Ii + Ekvs Ij + Em Effectsof amountof instruction
and of amountof exposure
onpopulations
(independently) withdiffering
amounts
of
bothinstruction
andexposure(Table6)
Note:I = instruction 0 = "filler"
activity
E = exposure i = sameamount

exposure (only),or instruction plus exposureand, on the other,the SL


process (e.g., sequence of acquisition),or rateor ultimateattainmentin
second language acquisition (SLA).x A definitiveanswer, therefore,
Thisarticlefocusesexclusively
on rateand ultimate
attainment in SLA. Fora comprehensive
literature
reviewandempirical studyofprocessissues,see Pica (1983).

360 TESOL QUARTERLY

This content downloaded from 185.2.32.21 on Sun, 22 Jun 2014 16:49:07 PM


All use subject to JSTOR Terms and Conditions
requires use of a true experimentaldesign, that is, (minimally)an
experimentaland a controlgroup,plus randomassignmentof subjects
to each.
Of thesixteenpossible comparisonsin Table 1, thosediscussedhere
are Types 12 through16. They are of importancefor two reasons.
First,theyprovidesupportingor disconfirmatory evidence fortheories
of SLA, such as Krashen'sMonitorTheory,2which make predictions
about SLA with and withoutinstruction.Second, they speak to the
efficacyof instructionand/or exposure in situationswhich can be
manipulatedby educationaladministrators,programdesigners,teach-
ers,and students.

CLASSIFICATIONAND EXAMINATIONOF STUDIES


EXPLORINGTHE EFFECTS OF INSTRUCTION
The RelativeUtility
ofExposureOnlyand the
SameTotalAmountofInstruction and Exposure
(comparison12)
Four studieshave made comparisonsof Type 12. All have dealt with
Englishas a second language (ESL) in the United States,three(Hale
and Budar 1970, Mason 1971, Upshur 1968) involvingadolescents
and/or adults, one (Fathman 1975) focusingon youngerand older
children(see Table 2)." Researchersin all fourstudiesclaimed to find
no advantage forinstruction plus exposureover exposure(only),Hale
2 Krashenclaimsthatformalinstruction
resultsin learning;thatis, it resultsin conscious
knowledge of"easy"rulesofa SL, suchas subject-verb agreement andwhento usea oran in
English.He further claimsthatthisknowledge can (notwill)be accessedbylearners whoare
monitor-users whenthey1) havetime,2) arefocusedonform(accuracy), and3) knowtherule.
Anunspeeded, discrete-pointtest,forexample,might meetalloftheseconditions. Whether the
learneris a childoran adult,mostofa SL, according to Krashen,
isacquired(viathecreative
construction process)through the processingof comprehensible inputreceivedin natural
communication. Theresult ofthisinformal exposure istheacquiredsystem, oracquisition,that
is,whatthelearner knowsabouta languageat theunconscious level.It is theacquiredsystem
thatdoes mostof theworkin normalSL use,thelearnedsystemactingonlyas a monitor,
planning andediting theoutputfromtheacquiredsystem ontherareoccasionswhenthethree
conditions foritsusearemet.Sinceconscious(meta-)linguistic knowledge impliesattainment
of the formaloperations stageof cognitivedevelopment, youngchildrencannotlearnor
monitor inthesetechnical senses,so theyshouldnotbe abletoprofit fromformal instruction
in
theseways.Similarly, neither children noradultsshouldbe able to benefit frominstruction at
"intermediate" proficiency levelsor beyond,foradvancesin proficiency at laterstagesvia
learningwouldinvolvemorecomplexrules,ruleswhichareeitherunknown (undescribed by
pedagogicgrammarians), unteachable, unlearnable,unusable,or severaloftheabove.Apart
fromteachinga fewrulesand a fewconversational routines(whichthelearnercan use to
obtaincomprehensible inputoutsidethe classroom),the main function of SL teaching,
therefore, is to providea sourceofcomprehensible input(foracquisition) to beginners,
who
oftencannotobtainthiselsewhere.(Forfurther detailsandrecentstatements oftheposition,
see Krashen 1980;1982,Chapter4.)
3Theroleofformal SL instruction inSLA wasa focusintheseandalltwelvestudiesreviewedin
thisarticle.Severalofthem,suchas Fathman(1975),involvedadditional issuesnotdiscussed
here,however.

DOES INSTRUCTION MAKE A DIFFERENCE? 361

This content downloaded from 185.2.32.21 on Sun, 22 Jun 2014 16:49:07 PM


All use subject to JSTOR Terms and Conditions
oa c3 0
9
~ a b
-1C
l ~ r
1; S0

9v -4 ChY n4?
%~r; 0tl a
a,~~1 o ~m:~ ~E

j ~ h

h
It
_3
C~ a~ a 3 3 Cf "
qj I-
O~ e
2 C)
o

C
"X~ ~v t
x
~ -cJ o ,~ ~l.c >
oc u" ~d CU
k c )3 3c )Q

i: rU yl Ei m U o4 u,o
~pa ~h X
?r,
E=
w,
;t; C ~ 'iib E: r*
bO
Qi c, $C u XC
cO

B
V) C cce m =3 c,
Ok 't;f C CA
~d~l eE WCL(
YU~ 0)
Q, C
O `L* v, Q)
cUUc 3 1 a)
v,a t,
3

~i Clj~ rj ~ Q1 rE ~I~ C4i


rr,
Uo,
r;czi M o

B
U

P1 M I t cJ I M

.~ ~

0o
-~ -~
a
iii t=
E
0
0 d

This content downloaded from 185.2.32.21 on Sun, 22 Jun 2014 16:49:07 PM


All use subject to JSTOR Terms and Conditions
andBudarclaimingthattheirexposure-only studentsinfactdidbetter.
Examination ofthestudiesbyUpshur,Mason,andFathmanconfirms
thattheirdata supporttheircommonconclusion:fortheirsubjects,
exposurewas as effective inpromoting SLA as thesametotalamount
ofinstructionplusexposure.As showninTable 2, however,therewas
some indicationthatinstruction helpedin all threestudies.The Hale
and Budarstudy,on theotherhand,seemsopen to otherinterpreta-
tionsthanthoseoftheoriginalresearchers.
Hale and Budarstudiedtwo groupsof children(grades7 through
12) inschoolsintheHonoluluarea.Therewas a totalof537students at
thebeginning ofthestudy,all ofwhomweretestedfortheamountof
Englishtheyacquiredafterone yearor twoyearsin one oftwokinds
ofprograms. Testscoresare providedfor329ofthe537.4Of the329,
70 were mainstreamed for one or two years.Their schoolswere
locatedin middleand upper-middle classneighborhoods and had an
averageratioof 110childrenwho werenativespeakersof Englishto
each child who was a non-nativespeaker.This group (Group A)
receivedno ESL instruction.
The remaining 259 studentsforwhomwe have testscoreswerein
varioussortsof "pull-out" ESL programsforone or twoyears.Their
schoolswereinlowsocio-economic areas,thekindwheremanynewly
arrivedimmigrants tendto settlein highdensities.In theschoolsin
question,therewas a ratioof 25 nativeEnglish-speaking childrento
everynon-nativechild. In otherwords,theretendedto be other
studentswithwhomnon-native speakerscouldconverseintheirnative
languagesin theseschools.Hale and Budarreportthat,in fact,many
childrenin thissecondgroup(GroupB) spokein theirmothertongue
whenevertheywereoutoftheclassroom, atrecess,duringlunchhour,
and at home.5
Notice,then,thatwhileHale and Budarintendedto comparetwo
groupsofchildren spreadoutthrough a varietyofHonoluluschoolsin
termsofwhether ornottheyreceivedanyESL instruction, theyinfact
compared two groups which differedin thisand otherways:
4Hale and Budar offerno explanationforthe missingdata on 208 (39%)of the subjects.
5 The fact thatstudentsin the Group B schools spoke more in theirnative languages than was
possible for childrenin the Group A schools (due to the highernumbersof non-English-
speakingchildrenin theGroup B schools) meansthatthechildrenin Group B were exposed to
less Englishthanthenon-nativesin Group A. It is not clear fromthe Hale and Budar studyjust
how many childrenin Group B thiswas trueof, however. If the majorityof childrenin the
instructionplus exposurecondition(Group B) received less informalexposureto Englishthan
children in the exposure-onlycondition (Group A), then this was not strictlya Type 12
comparison-the Group B childrenreceivingless totalinstruction plus exposurethantheGroup
A children'stotalexposure.If thiswas what happened, the exposure-onlychildren(Group A)
mighthave been expected to have an advantage, independentof the type of treatmentthey
received.

DOES INSTRUCTION MAKE A DIFFERENCE? 363

This content downloaded from 185.2.32.21 on Sun, 22 Jun 2014 16:49:07 PM


All use subject to JSTOR Terms and Conditions
GroupA (n=70) GroupB (n=259)
exposure
only plusexposure
instruction
SL exposure (no ESL) SL instructionplus exposure (ESL)
more total exposure to English less total exposure to English
(higherratio of native to non-native (lower ratio of native to non-native
speakers) speakers)
middle/upper-middle class workingclass
more positiveparentalattitudes? less positiveparentalattitudes?

Hence, any differencesobserved in the SL attainmentof the two


groupscould notbe attributedto any one of thesevariables,including
the provisionor withholdingof ESL instruction.
Hale and Budar compared childrenfromthetwo kindsof programs
on the basis of an aggregate of 1) scores on an (unvalidated) oral
interview(conductedby one of theresearchers),2) scoreson theDavis
test, and 3) general scholastic average. Comparisons were made of
childrenwho had been in theprogramsforone yearand fortwo years.
No inferential statisticswere employedin the analysis.On thebasis of
the raw scores foreach group,the investigatorsclaimed thatstudents
who were mainstreamed(Group A - no ESL) did betterat SLA. They
recommended to schools in Honolulu thatthey"maximizethe immi-
grantstudent'stotal in-schoolexposure to the English language and
culture,and minimize English language teachingin formalTESOL
classes" (1970:491).
The researchers'claim thatthe exposure-onlychildren(Group A)
did betterwithoutESL instruction is questionable.There are enough
flaws in the study (some of which, to theircredit,Hale and Budar
acknowledge in theirreport)to invalidateany claims theymightwish
to make based upon it. Most problematic,perhaps,are thesocial class
differencesbetween the two groups of schools. The relationship
between social class and educationalattainment is so well documented
thatchildrenin theworkingclass (instruction plus exposure)group (B)
mighthave been expected to do less well thantheirmiddle and upper-
middle class peers, regardlessof the treatmenttheyreceived. This is
especiallytruein thisstudywhen itis rememberedthatthedependent
variable was not just a measure of SL proficiencybut also contained
generalacademic achievementscores. Parentalattitudesto SL educa-
tion and use have also been found importantin Canadian studiesof
immersioneducation (Swain 1981) and may well have played a role
here.
Given these circumstances,one could argue thatinstruction had a
positiveeffectif therate of SLA was equal in both conditions.In fact,
as indicated in Table 2, chi-squaretestsrunon Hale and Budar's data
show thatthiswas preciselythecase fortheone-yeargroup.Although
the two-yearand overall scores forboth years combined did indeed
364 TESOL QUARTERLY

This content downloaded from 185.2.32.21 on Sun, 22 Jun 2014 16:49:07 PM


All use subject to JSTOR Terms and Conditions
favorthe exposure-only group,it appearsthatinstructionmaywell
havebeenbeneficialin theearlystages.Fromthefirstyearresults, at
least,therewas no evidencethatexposureonlywas advantageous-
quitethereverse-andso no basisforHale and Budar'srecommenda-
tion,at leastas faras thisgroupwas concerned.
In general,the resultsof the fourstudiesin Table 2 suggestno
difference betweenprogramsof exposureonlyand the same total
amountof instruction plus exposureforchildren,adolescents,and
adults.There are, however,severalhintsof possible benefitsfor
instruction, particularlyforstudentsof lower SL proficiency who,
becauseoflow SL proficiency (notjusttherelatively
smallernumbers
ofSL speakerswithwhomtheymaycomeintocontact),canoftenfind
itdifficulttosustainSL conversation and,thereby,obtaincomprehen-
sible input. For such students,as Krashen (1980) has argued, the SL
classroommay be the main, and so an especially valuable, source of
exposureto the targetlanguage.

The RelativeUtilityofDiffering
AmountsofInstruction and
Exposurein PopulationswiththeSame TotalAmountof Both
(comparison13)
Two studies have made comparisonsof Type 13. Both have dealt
withESL in the United States,Upshur (1968) focusingon adults,and
Fathman (1976) studyingelementaryand secondary school children
(see Table 3). Upshurfoundno significantdifferencebetweenthegain
scores of two groups of intermediateand advanced studentsaftera
seven-week period in which one group received one hour of ESL
instructionper day and one group two hours per day, and during
which period both groups also attended law classes and lived in an
English-speakingenvironmentat the Universityof Michigan.That is,
therewas no differencebetween two groups in which both received
the same total amountof instructionplus exposure,even thoughone
groupreceived more instruction.
Using an oral interview and the SLOPE test, Fathman (1976)
measured the ESL proficiencyof a total of 331 childrenin public
schools in the Washington,D.C. area at the beginningand end of a
year in which some received three,some five,and some ten hoursof
ESL instructionper week. During thisyear, all childrenwere main-
streamed for the rest of the school day. Fathman then computed
percent gain scores for the two groups by dividing the actual pre-
test/post-testgain that students made by the total possible gain.
(Actualgain= post-testscore minuspre-testscore. Total possible gain=
maximumpossible testscore minuspre-testscore.) Thus, a group of
studentswitha mean pre-testscore of 10 and a post-testscore of 40 on
a 100-pointoral interviewwould have an actual gain score of 30. The
DOES INSTRUCTION MAKE A DIFFERENCE? 365

This content downloaded from 185.2.32.21 on Sun, 22 Jun 2014 16:49:07 PM


All use subject to JSTOR Terms and Conditions
TABLE 3
RelativeUtilityofDiffering
Amountsof Instruction(I) and Exposure(E) in Population

Subjects&
Study Proficiency Treatment Duration Test Type
Level

1. Upshur(1968) adults 1 vs 2 hours 7 weeks DP 1. no signi


I A ESL per day
plus E (law
classes)

2. Fathman(1976) children 3, 5, or 10 1 year ESL I 1. largerg


(elementary hoursESL Note: theu
& secondary per week ever
school) plus E (school
B I A classes)

Note:B = beginning DP = discrete-point * interpre


I = intermediate I = integrative
A = advanced

T1-

This content downloaded from 185.2.32.21 on Sun, 22 Jun 2014 16:49:07 PM


All use subject to JSTOR Terms and Conditions
totalpossible gain forthe group would be 100 - 10 = 90. The percent
gain score would be 30 divided by 90 = 33%.A group with a mean
pre-testscore of 80 and a post-testscore of 90 would have an actual
gain of 10, a total possible of 100 - 80 = 20, and a percent gain
score of 10 divided by 20 = 50%.Thus, a gain of 30 points by a low
proficiencygroup would resultin a considerablysmallerpercentgain
thana gain of 10 pointsby an advanced group.Fathmanfoundthat,on
both tests,studentswith less instructionmade greaterpercent gain
scores thanthosewithmore instruction.
As the sample computationsshow, if studentswith lower pre-test
scores are compared with studentswith higherpre-testscores who
receiveless instruction (probablybecause it is feltthattheirhigher
startingproficiency meanstheyneedless),as was thecase inFathman's
study,students withless instructionare likelyto appearto do better.
Examination of theraw scoresin Table 2 of Fathman'sstudy(1976:
437)showsthatthegieatestabsolutegainson bothtestsweremadeby
thestudents withmoreESL instruction. The problemis theusualone
ofhowto interpret gainscores.Use ofeitherabsoluteor percentgain
scores(calculatedby dividingactualgainscoresby pre-test scores)
would producethe oppositeresulton herdata of thatreportedby
Fathman.Such analyseswould also be misleading, however,unless
one made the(unjustified) assumptionthatan absolutegainat lower
levelsofproficiency is comparabletothesameabsolutegainathigher
levels.(Upshurattempted todealwiththisprobleminhisstudy, which
alsoinvolveda comparison ofgroupswithdiffering initialproficiency,
byuse ofANCOVA.)
In summary,when the relativeutilityof differing amountsof
instructionand exposureinpopulations withthesametotalamountof
bothhas beenstudied,theresultsareambiguous.One study(Upshur
1968)has foundno effectformoreinstruction forintermediate and
advanced-level adultsovera short(seven-week) period. Anotherstudy
(Fathman1976)has reportedthesame resultforchildrenof various
proficiency levels over a longerperiod (one year). Findingsof the
latterstudyareambiguous, however,and couldas easilybe arguedto
showan effectfortheamountof instruction (thatis, moreis better
thanless),especiallyat lowerlevelsofSL proficiency.

TheEffect
ofAmount
ofInstruction
on Populations
withtheSameAmount
ofExposure
(comparison
14)
Two studies,
Krashen andSeliger(1976)andKrashen, and
Seliger,
Hartnett
(1974),havelookedattheeffect
ofamountofinstruction
on
withthesameamountof exposure(see Table 4). The
populations
DOES INSTRUCTION MAKE A DIFFERENCE? 367

This content downloaded from 185.2.32.21 on Sun, 22 Jun 2014 16:49:07 PM


All use subject to JSTOR Terms and Conditions
TABLE 4
EffectofAmountofInstruction(I) on Populationswiththe SameAmo

Subjects&
Study Proficiency Treatment Duration Test Type
Level

1. Krashen& adults dplEt (from I 1. in 9 pair


Seliger I A monthsto more I s
(1976) years?)

2. Krashen, adults dplE (long res- DP 1. in 8 pair


Seliger,& B I A idence for more I s
Hartnett most;I?)
(1974) 2. in 11 pa
but in w
member

Note: B
" beginning DP= discrete-point
I -=intermediate I = integrative
A = advanced
-]
H
dplE = differing periods of I and E
u,O 2 practice = years of residence x amount of English spoken outside class on a scale of I to 10

(3
t0
CI
Hp

This content downloaded from 185.2.32.21 on Sun, 22 Jun 2014 16:49:07 PM


All use subject to JSTOR Terms and Conditions
studies involved adult acquirers of ESL in the United States with
differinglevels of SL proficiencyand differingperiods of instruction
and exposure(dplE in Table 4). A wide rangeof periods of exposure,
particularly,was representedin thesamples,some studentshavinghad
several years of residence in the New York area. In both studies,the
procedure was to match pairs of students with equal periods of
exposure but differentperiods of instruction,and then to test the
hypothesis that more instructionwas beneficial by seeing if the
member (of each pair) who had had more instructionhad greater
proficiency."Exposure" was quantifiedin thesestudiesby calculating
a "practice" score for each subject. This was done by having the
studentsreporthow much English(the SL) theyspoke (on a scale of 1
to 10) to nativespeakersof English,to speakersof theirown language,
and to otherforeignerswho were not native speakers of either.The
sum of thesethreescores,a "talking"score,was thenmultipliedby the
lengthof time in the United States to give the "practice" score. The
hypothesiswas sustained on both occasions-more instructionpre-
dicted higher SL proficiency.In addition, Krashen, Seliger, and
Hartnettlooked at theproficiencyof eleven pairsof studentsmatched
forexposure(heredefinedby lengthof residencein theUnitedStates)
in which the member with less instructionhad had more "practice."
Again, studentswith more instructionscored higherthan those with
less instructionin a significantly
highernumberof pairs (see Table 4).
There is clearlya consistentpatternin these results:in populations
withthesame amountof exposure,moreinstruction predictshigherSL
proficiency.Further,Krashen,Seliger, and Hartnett'sResult 2 (see
Table 4) suggeststhatmore instruction can even compensate forless
exposure (as measured by the amountof "practice"). On the basis of
these studies alone, however, one cannot be certain that it is more
instructionper se that is having an effect.Comparisons of Type 14
involve subjects with more instructionbut also with more overall
opportunityto acquire the SL (by virtueof the greateramount of
instruction).Krashen(1982) has argued thatthe greaterproficiencyin
the groups withmore instruction is due to the additionalopportunity
for acquisition (unconscious learningvia the creative construction
process) in the classroom setting,here functioningas a source of
comprehensibleinputforthe SL learner.
Comparisonsof Type 14 alone cannotresolvethisissue.If Krashenis
correct,however,one would predictthesame resultin studiesof Type
15 where,among studentswithequal amountsof instruction, some had
had more exposureand, hence, a greatertotal opportunityto acquire
the SL. But a findingthat those with more exposure had higher
proficiencywould not prove Krashen'sclaim forthe same reasonthat
studiesof Type 14 cannot disproveit: each comparisoncould simply

DOES INSTRUCTION MAKE A DIFFERENCE? 369

This content downloaded from 185.2.32.21 on Sun, 22 Jun 2014 16:49:07 PM


All use subject to JSTOR Terms and Conditions
be showingthatsubjectswithgreatertotalopportunityto acquire a SL
do better.Suppose, however, thatstudies of Type 15 were found to
show no advantage for students with more exposure but equal
That would contradictKrashen'sclaim,unlesssome alterna-
instruction.
tive explanationwere available. It would also suggestthatthe more
obvious interpretation
of the Krashenand Seliger (1976) and Krashen,
Seliger, and Hartnett(1974) results,namely that more instruction
predictshigherproficiencydue to a genuineeffectforinstruction, is
indeed thecorrectone. The studiesof Type 15,to whichwe now turn,
are thereforeimportantnot onlyin theirown right,but because of the
lighttheycan shed on studiesof Type 14.

The EffectofAmountofExposureon Populations


withthe SameAmountofInstruction
(comparison15)
There have been threestudiesof Type 15. All involved adult ESL
acquirers in the United States and acquirers at various proficiency
levels (see Table 5). Krashenand Seliger (1976) conducted a similar
studyto thosedescribedunderType 14 above, withthesame matched
pairs designand thesame measuresof exposure("practice"and length
of residence). This time, however, they matched for amount of
instructionand compared members of each pair with more or less
exposure. Krashen, Seliger, and Hartnett(1974) also followed this
procedure in a thirdcomparisonin the studydescribed earlier.In all
threecases, not more,but fewersubjects withmore exposurescored
higheron the proficiencymeasures (the differencebeing statistically
non-significant in each case).
In the onlyothercomparisonof Type 15, Martin(1980) reportsthe
resultsof a studyin which 166 adults (83 pairs matched on Michigan
pre-testscores) received 22.5 hoursof ESL instruction
per week for14
weeks, duringwhich time half of the studentslived with othernon-
nativespeakersin universitydormitories(less exposure) and the other
halfwithAmericanfamiliesin a "homestay"program(moreexposure).
The post-testconsistedof scores on the TOEFL and class grades in
grammar,reading,composition,and speaking.Resultsof an ANOVA
showed thatthehomestay(moreexposure)studentsscored significant-
ly higheron theTOEFL (p < .05) and on all theothertests(p < .001).
Martin'sresults,then,appear to conflictwith those of Krashen and
Seliger (1976) and Krashen,Seliger,and Hartnett(1974).
Atleast two factorsdistinguishMartin'sstudyfromtheothertwo of
its type. First,the studentsin the homestayprogramvolunteeredto
participate.As Martinnotes,theymay, therefore,have differedfrom
the dormitorygroup in ways otherthanthe amountof exposurethey

370 TESOL QUARTERLY

This content downloaded from 185.2.32.21 on Sun, 22 Jun 2014 16:49:07 PM


All use subject to JSTOR Terms and Conditions
t3
O
FT1 TABLE 5
OrZ EffectofAmountofExposure(E) on Populationswiththe SameAmo

z Subjects&
Study Proficiency Treatment Duration Test Type
I pvl

1. Krashen& adults dplE (from I 1. in 14 p


Seliger I A monthsto more p
(1976) years?)
2. in 12 p
E (res
CTI
zri
?3 2. Krashen, adults dplE (long DP 1. in 21 p
Seliger,& B I A withlong residence practi
'rl Hartnett residence period in Note:sam
(1974) period in mostcases; than 1 yea
mostcases I?)

3. Martin adults 22.5 hours 14 weeks DP I I. ANOV


(1980) I A ESL a week were h
forall; p < .00
dormitory reside
residence
withother
foreignstu.
dentsor home-
stayprogram
withAmerican
families

Note: B = beginning DP= discrete-point dplE = dif


I = intermediate I = integrative
A= advanced

This content downloaded from 185.2.32.21 on Sun, 22 Jun 2014 16:49:07 PM


All use subject to JSTOR Terms and Conditions
received. They may, forexample,have been more motivatedor have
had differentreasons for wantingto learn. In any case, theirhigher
scores cannot be attributedwith certaintyto an effectfor greater
exposure.Second,thebriefreportavailableofMartin's studymakesit
seemthathersubjectswerereceivingtheirfirstintensive exposureto
English,whereasmanyof thesubjectsin theothertwo studieswere
long-time residentsin the New Yorkarea. Perhapsthethree-month
periodof the Martinstudyis enoughforsome acquisitionto occur
through exposureinintermediate and advancedstudents,butnotlong
enoughfora ceilingontheeffect ofexposuretobe reached,attainment
of whichmeansthattheadvantagesof exposureforacquisitionwill
havebenefited students and beyondwhichlittleorno further benefit
willaccrue.Severalyearsof SL exposurefora groupofsubjectsmay
benefitall ofthemequally,and so masktheeffectofthatexposureon
studentswithmoreor less thanthemaximumusefulperiod(one to
threeyears?).Further data on theamountofinstruction and exposure
inall threestudiesis needed.
Meanwhile,whateverthe reasonsfor the variableresultsacross
studiesof Type 15, theresultsthemselves have implicationsforthe
Type 14 studies.The factthatthreenullfindings were obtainedby
Krashenand hisassociates(Table 5) foramountof exposuresuggests
stronglythattheeffectforamountofinstruction instudiesofType 14
is genuineand is notsimplytheresultofa greateroverallopportunity
foracquisitionin and out of theclassroom.If true,thiswouldhave
implicationsforvariousaspectsofMonitor Theory,as willbe discussed
later.For thoseinvolvedinlanguageteaching, thereinterpretation
of
the Type 15 studieswould suggestthatmore instruction can be
beneficial,even forstudentswithconsiderableSL exposurein their
language-learning histories.
StudiesofTypes14and 15,takentogether,
wouldsuggestthatmoreinstruction can also be morebeneficialthan
moreexposureforsubjectsofthiskind.

TheEffectsofAmount ofInstruction
andofAmount
ofExposure(Independently) on Populationswith
Amounts
Differing ofBothInstruction
and Exposure
(comparison 16)
Four studieshave made a Type 16 comparison(see Table 6). All
fourhavefounda statistically
significant between
positiverelationship
amountof instruction and test scores; threehave found such a
foramountof exposure.The strength
relationship of therelationship
withinstruction was strongerthanthatwithexposurein two of the
studiesfinding
both(BriBre 1978;Krashen, andUsprich
Jones,Zelinski,
1978),andweakerinone (Carroll1967).Thenullfinding foramountof
372 TESOL QUARTERLY

This content downloaded from 185.2.32.21 on Sun, 22 Jun 2014 16:49:07 PM


All use subject to JSTOR Terms and Conditions
TABLE6

ofAmount
Effects ofInstruction ofExposure(E) (Inde
(1) andofAmount
withDiffering ofBothInstruction
Amounts andExpo

c-I
Hb
&
Subjects
z Study Proficiency Treatment Duration TestType
Level

'TI 1. Krashen, adults ESL I and dplE; x E DP I 1. positi


Jones, I A residence = 4.05years, all test
Zelinski, inUSA i I = 2.28 2. positi
& Usprich years all tes
(1978)

C') 2. Briere children SpF/SL& 1 yearplus DP 1. positi


(1978) B limitedE ofI, & listen
by parent dpE 2. positi
SL use listen
3. Carroll adults FL I, & E dplE I 1. signif
(1967) B I A by"year listen
abroad"& (Frenc
somepar- 2. positi
entaluse ingtes
(36
p < .01
4. Chlihara
& adults EFL & dplE; RE DP I 1. positi
Oiler B I A shortU.S. = I month all fou
(1978) visits i I = 8.41 2. no sign
years scores
(Japan) Note:ver
Note:B = beginning DP = discrete-point S
I = intermediate I = integrative dpI
A = advanced

This content downloaded from 185.2.32.21 on Sun, 22 Jun 2014 16:49:07 PM


All use subject to JSTOR Terms and Conditions
exposurein a studyby Chihara and Oiler (1978) is presumablydue to
the factthatthe amountof exposurefortheirsample of 123Japanese
EFL students(an average of about one month,judgingby the mean
and standarddeviationreportedin theirTable 3, 1978: 60) is simply
insufficient
formuchacquisitionto occur.
Withoutstudiesof Type 15, it would again be difficultto interpret
thesefindings.Each independentcomparisoncould eitherbe showing
an effectfor more instructionor more exposure, or for more total
opportunityfor SLA throughmore total instructionplus exposure.
Giventhenullfindingsformoreexposurein populationswiththesame
amount of instruction(Type 15), however, one is more inclined to
interpretthecurrentpatternas further evidence of a genuineeffectfor
amountof instruction across all fivestudies.

SUMMARYAND DISCUSSION
Table 7 summarizes the twelve studies discussed above. If the
interpretation proposed for them is basically correct,it would seem
thatthereis considerableevidence to indicatethatSL instruction does
make a difference.There are six studiesthatsupportthisconclusion,
two ambiguous cases (both of whichmightin factbe argued to show
thatinstructionhelps), and threewhich have null findings,although
each again containssome hint(s)of an advantage forinstruction. (The
studyby Martindoes not speak to thisissue since all subjectsreceived
the same amount of instruction.)Further,as shown by the sub-
classificationsof the studiesin Table 7, the effectforinstruction
holds
1) for childrenas well as adults, 2) for intermediateand advanced
students,not just beginners,3) on integrativeas well as discrete-point
tests,and 4) in acquisition-richas well as acquisition-poorenviron-
ments.The effectforinstruction is also strongerthanthatforexposure
in fivecases.
Unlesssome alternativeexplanationexists,thesefourresultsseem to
runcontraryto thepredictionsof MonitorTheory.6(1) is notpredicted
because childrenshould not be able to learn,in Krashen'ssense of the
term; they supposedly lack the cognitive maturitywith which to
develop consciousknowledge of rulesof theSL and/orto apply them
by monitoring.(2) is not predictedforsomewhatmore complex rea-
sons. Instructionis supposed to resultin learning,definedby Krashen
as conscious knowledge of rules of the SL. This kind of knowl-
edge (and its subsequentuse via monitoring)is held onlyto be possi-
ble with a few "easy" grammarrules,such as thirdperson -s or the
6 If correct,theyare, of course,problematicforsome otherSLA theories,too. They are reviewed
in termsof MonitorTheorybecause itis one of thefewsufficiently
developed to make explicit
claims about therole of instruction.

374 TESOL QUARTERLY

This content downloaded from 185.2.32.21 on Sun, 22 Jun 2014 16:49:07 PM


All use subject to JSTOR Terms and Conditions
TABLE 7
RelationshipsBetweenInstruction(I), Exposure(E), and Second La

Study SLAType Subjects Proficiency Acquisition


(B, I or A) Environment

Studiesshowingthatinstruction
bepgs
trl
1. Carroll(1967) FLL in USA & adults B I A mixed
z
tO
SLA abroad
2. Chihara& Oller (1978) EFL (Japan) adults B I A poor
C) 3. Briitre(1978) SpSL (Mexico) children B mixed
4. Krashen,Seliger, ESL in USA adults B I A rich
& Hartnett(1974)
5. Krashen& Seliger ESL in USA adults I A rich
(1976)
6. Krashen,Jones,Zelinski, ESL in USA adults B I A rich
& Usprich(1978)

"3 cases
Ambiguous
0' 7. Hale & Budar(1970) ESL in USA adolescents B I A rich
8. Fathman(1976) ESL in USA children B I A rich

Studies
showingthatinstruction
doesnothelp
9. Upshur(1968; Experiment1) ESL in USA adults I A rich
10. Mason (1971) ESL in USA adults I A rich
11. Fathman(1975) ESL in USA children B I A rich

Additional
studyshowingthatexposurebels
12. Martin(1980) ESL in USA adults I A mixed

Note: B = beginning DP= discrete-point


I= intermediate I = integrative
A = advanced

This content downloaded from 185.2.32.21 on Sun, 22 Jun 2014 16:49:07 PM


All use subject to JSTOR Terms and Conditions
a/an distinction (see Seliger1979). Such rulesare not sufficient for
instruction tohavemade a difference at theintermediate oradvanced
levels,as seemsto havebeenthecase inseveralstudies(see Table 7).7
Theyarenoteventhekindof"grammar" taughtat theselevelsinmost
ESL programs. (3) is notpredicted for similarreasons.Learning, inthe
formofconsciousknowledgeofsuch"easy"rules,is supposedlyonly
availableon discrete-point tests,whichstimulate a focuson formand
bring the Monitor into play ifthere is time for it to operate.Atleast
fourstudiesin Table 7 show an effectforinstruction (fromwhich
learning on test
supposedlyarises) integrative performance. (4) is not
predictedbecause,according to Monitor Theory, instructionis sup-
to
posed helponly in the of
earlystages SLA, and even thenchiefly as a
sourceof comprehensible for
input acquisition. It is onlyof use to
moreadvancedstudents iftheylackalternative sourcesofcomprehen-
sibleinputoutsidetheclassroom,thatis, iftheylive in "acquisition-
poor"environments. Severalstudiesin Table 7 appearto be showing
an effect forinstructionamongstudents atlaterstagesofSLA andwith
of for in
plenty opportunity exposure "acquisition-rich" environments.
Thereare twopotentialalternative explanations for thesefindings,
eitherof which,if sustained,would make themcompatiblewith
MonitorTheory.First,as statedabove,Krashenhas claimedthatthe
classroomwillserveas a sourceof comprehensible inputforacquisi-
tioninanotherwise acquisition-poor environment, is,oneinwhich
that
littleor no comprehensible inputis availableoutsidetheclassroom.
Exampleswould includeforeignlanguagelearning(as withEFL in
Japan,in Chiharaand Oiler 1978,and Spanishin predominantly
indigenous-language-speaking partsofruralMexico,inBribre1978),as
wellas SLA bylearners livingina "linguistic ghetto" inthewidertarget
languageenvironment, such as the predominantly Spanish-speaking
partsofEastLos Angeles(as mayhavebeenthecase forsomesubjects
inthestudiesbyKrashenandhisassociatesinNew York,andagain,for
some childrenin the studyby Bribre).This explanationdoes not
appearto accountforthefindings in at leastfourstudiesin Table 7,
7 It is difficult has reallybeen shownto help intermediateand advanced
to be surethatinstruction
students,as it appears to have done in fivestudiesin Table 7 (Carroll 1967; Chiharaand Oiler
1978; Krashen,Seliger, and Hartnett1974; Krashen and Seliger 1976; and Krashen,Jones,
Zelinski,and Usprich1978). The entriesforstudents'proficiencylevels in each case have been
based on the original investigators'descriptions,such as "students of a wide range of
proficiency"or "all proficiencylevels." However, in some cases, theselabels are relative,not
absolute, and may have involved studentswho, for example, were "advanced" in the ESL
programfromwhich theywere drawn but not veryproficientin any absolute sense. At least
one study, Carroll (1967), however, does constitutea clear counter-exampleto Krashen's
prediction,forthe average FSI-equated oral proficiencyratingforstudentsin thatstudywas
2+. A 2+ on theFSI scale means thatthespeakercan handle routinesocial demands and limited
work requirements,includingcasual conversationsabout currenteventsand one's work and
family,and can understandmostconversationson non-technical subjects.This is clearlynotthe
descriptionof a beginner,and many subjects in Carroll's study were of still higher SL
proficiencythanthis,giventhat2+ was theaverage forthe SL groups.

376 TESOL QUARTERLY

This content downloaded from 185.2.32.21 on Sun, 22 Jun 2014 16:49:07 PM


All use subject to JSTOR Terms and Conditions
however,whereinstruction was foundto have an effectin acquisition-
richenvironments, unlessnearlyall thesubjectsfailedto encounterthe
SL in several years of residence in countriesin which the target
language was spoken. Certainly,none of thestudiesgives any indica-
tionthatthiswas the case.
The second potential explanation lies in the nature of the input
obtained by subjects in some of the studies. MonitorTheory claims
that acquisition will occur when comprehensibleinput is available
which containsinput at "i + 1", that is, which includes forms(lexis,
morphology,syntacticconstructions)that are one stage beyond the
learner'scurrentstage of interlanguagedevelopment. Krashen (per-
sonal communication) suggests that the subjects in some studies,
particularlythoseby Krashenand hisassociatesin New York,obtained
comprehensibleinput throughliving in the United States, but not
necessarilyinput containing"i + 1." Some, for example, may have
worked in occupations in which they received input which they
understoodbecause of itslimitedrange,highdegree of predictability,
and formulaic,routinenature. Gas stationattendants,for example,
mightunderstandeverythingcustomerssaid to themwithouthearing
anything"new" which was also comprehensible("i +1"). This would
indeed account forthe data. It is, however,a post hoc explanation(as
Krashenis aware) and surelya likelyone in only a few cases. How
many subjects in these studies had such limitedexposure to English
(and nothingmore) duringfairlylong periods of residence,and how
trueis it thatsuch occupations are accomplished withsuch a limited
rangeof input?
While theremay indeed be "somethingwrong withthe studies,"in
the sense that they were not measuring what they seemed to be
measuring,anotherexplanationis thatthereis somethingwrongwith
the theory.A fullexplorationof thispossibilityis beyond thescope of
thisarticle,but a few suggestionsfollow.
The firstmodificationof MonitorTheorywhich would enable it to
account for the data on instructionwould consist of redefiningthe
constructlearning,which seems currentlyto be too narrow.Learning
must involve somethingmore than conscious knowledge of "easy"
grammar rules. As Krashen claims, children may not be able to
develop suchrulesuntiltheonsetof formaloperations,but theyclearly
do develop otherkindsof metalinguistic awareness (startingat around
two yearsof age; fora review, see Clark 1978) and, as we have seen,
theyseem to benefitfromformalSL instruction.The same abilityis
presumably still available to adults. Further,as argued earlier, if
conscious knowledge of the SL only involved knowledge of a few
"easy" rules, there would be no way of explaining the benefit of
instructionto intermediateand advanced adultacquirers,suchas those

DOES INSTRUCTION MAKE A DIFFERENCE? 377

This content downloaded from 185.2.32.21 on Sun, 22 Jun 2014 16:49:07 PM


All use subject to JSTOR Terms and Conditions
studiedby Brown (1980), forwhom "a few easy grammarrules"are no
longerwhat matters.
Perhapslearninginvolvesthe experience(obtained throughinstruc-
tion) of treatinglanguage as object and the concomitantabilitiesthis
brings,includingnotonlytheabilityto monitorwith"easy" ruleswhen
conditionspermit,as Krashenclaims,butalso theabilityto improveSL
performancein language-likebehaviorin general.Language testsof all
kindsprobably encourage the use of thisability.Note thatthe studies
reviewed hereshow subjectswithexperienceof thiskinddoing better
thansubjectswithoutit (or withlessof it) on tests,whetherdiscrete-
pointor integrative. Thisis a resultwhichwouldbe predicted,given
Farhady'sre-analysis of the data on discrete-point and integrative
measures(Farhady1979). Farhadyprovidesevidenceto the effect
that,whatevertheirrelativemeritsas diagnosticand placement
instruments, discrete-point and integrative testsare equallyvalid as
measuresofSL ability.
The dataoninstruction wouldalsosuggestthattheMonitor Theory's
claimsas to whenlearningcan be used (i.e.,theMonitorhypothesis)
also need to be extendedto include (at least) all "language-like"
behavior,notjustdiscrete-point tests.Initially,
thisextension inscope
oftheMonitor hypothesisappearsto createa problemfortheNatural
Orderhypothesis. This,however,turnsoutto be a falsealarm.
The NaturalOrderhypothesis is usedin MonitorTheoryto capture
the generalization, based on numerousstudiesin the 1970s (see
Krashen1977fora review),thatthereis a striking similarityin the
accuracy order fora of
group Englishgrammatical morphemesin
child and adult SL learnersfroma varietyof firstlanguageback-
groundsacrossa variety oftasksandmodalities. (Monitor Theorygoes
on tousethenaturalorderas evidenceoftheworkings ofacquisition.)
The naturalorderhasbeen foundtobe systematically disturbed inthe
SL performance ofsubjectson taskswhichcan be monitored, suchas
unspeeded,discrete-point, paper-and-pencil grammartests."Easy"
grammatical items,suchas thirdperson-s andregularpast,tendtorise
inaccuracy(andinrank)on suchtests.Thisphenomenon is takentobe
evidenceof subjects'use, via monitoring, of whattheyknow con-
sciouslyofSL grammar.
In orderto accountfortheeffectofinstruction on childrenand on
intermediate and advancedadultlearners(noneof whomshouldbe
able tolearninKrashen's sense),andonintegrative as wellas discrete-
pointtests,itis nowclaimedthatmonitoring mustbe possibleon tests
in general,regardlessof the learner'sage or proficiencylevel. If so,
however,how are we to explain the naturalorder observed on most
integrativetests,such as dictation,cloze, and freecomposition?
Note that the Monitor hypothesis already allows not only for

378 TESOL QUARTERLY

This content downloaded from 185.2.32.21 on Sun, 22 Jun 2014 16:49:07 PM


All use subject to JSTOR Terms and Conditions
disturbed morpheme accuracy orders but also for small general
improvementsin accuracy on tests in general. Further,the same
morphemestudies,upon which much of MonitorTheory's claim for
the limitedvalue of instructionrests,may have seriouslyunderesti-
mated instruction/learning. the studiesrecognized
First,statistically,
only the large disturbancesneeded to obtain a non-significantSpear-
man rankordercorrelationcoefficient.Second, theyconsideredonlya
very limited subset of SL grammar.It should be recalled that the
"morphemestudies"of the 1970swere originallymotivatedby a search
foruniversalsin thesequence of SLA, not forthe effectof instruction
(how they are now interpretedin Monitor Theory), for which a
disturbedaccuracyrankorderis a verylimitedand oblique test.
In summary,as indicated above, the proposed redefinitionof
learningwould affectthe acquisition/learning distinction(by upgrad-
ing the importanceof learningand, thereby,of instruction)and also
the Monitorhypothesis.It would not,however,challengetheAcquisi-
tion/Learninghypothesis.Failure to broaden the concept of learning,
on the otherhand, would mean changingthe Acquisition/Learning
hypothesis. Iflearning retained itscurrently narrowdefinition, itwould
be necessary topositthatlearning canbecomeacquisition, a possibility
thatMonitorTheoryrules out (Krashenand Scarcella 1978). The
changewouldbe necessary inordertoaccountfortheapparenteffect
ofinstruction on SL acquirersat theintermediate and advancedlevels
forwhominstruction is associatedwithproficiency even afterthe
passage of timeforexposure,and so foracquisition(see Table 7,
studies4, 5, and6). A re-evaluation oftheimportance oflearning(and
so ofinstruction), in otherwords,wouldobviatetheneed fora more
fundamental changeinMonitor Theory,onewhichwouldbe necessary
iflearning preserved itsnarrowdefinition andinstruction itssupposed-
lyrather insignificant rolebeyondthebeginning level.8
Whatevertheplace of instruction in a theoryof SLA, thestudies
reviewedherehave implications forlanguageteachingprofessionals.
Put rathercrudely,instruction is good foryou, regardlessof your
proficiency level,of the widerlinguistic environment in whichyou
receive it, and of the type of testyou are going to performon.
Instruction appearsto be especiallyusefulin theearlystagesof SLA
8 In fact,thereseem to be severalreasons,in additionto the data on the effectof instruction, for
positingthat(redefined)learningcan become, or aid, acquisition-convergentvalidationfor
theidea. Some are empiricallymotivated(see, forexample,Peters1980,Schneiderman1982),
and some theoretically. As an example of a theoreticalmotivation,positinga "cross-over"from
learningto acquisitionwould reconcilestudiesshowingan advantageforinstruction in rateand
ultimateattainment in SLA withthefactthatsome adultbeginnersseem capable of acquiringa
native-likecommand of SL syntaxwithoutinstruction. If learningcontinuedto be held not to
aid acquisition,it would be necessaryto posit two learnertypes (those who do/do not need
instructionto go all the way). Such a distinctionwould make the Theorymore powerful,and
seems to lack any compellingindependentmotivation.

DOES INSTRUCTION MAKE A DIFFERENCE? 379

This content downloaded from 185.2.32.21 on Sun, 22 Jun 2014 16:49:07 PM


All use subject to JSTOR Terms and Conditions
and/or in acquisition-poorenvironments, but neitherof these condi-
tionsis necessaryforitseffectsto show up. Further,thereis some slight
evidence thatlargerproportionsof instructionare helpfulin cases of
instruction and exposure,but the evidence is onlyslight.Lastly,while
the positive effectsfor instructionin the studyby Bribre(1978) are
probablydue at least in partto theinstructionhavingtakenplace in an
acquisition-poorenvironment, there is some indication from other
studies(Fathman 1976,Hale and Budar 1970) thatinstruction can help
childrenand adolescents as well as adults, with the benefitsagain
being strongestat beginninglevels and in acquisition-poorenviron-
ments,but possiblynot limitedto these.
For SLA theoryand SL educators alike, on the basis of currently
available studies,an answerto thequestion"Does SL instruction make
a difference?"is a not-so-tentative"Yes." However, even if,as I hope,
the data on instruction have been correctlyinterpretedhere,theyare
obviouslynot as clear-cutor as "positive"as mostTESOL profession-
als would like. There is obviouslya genuineneed forfurther research
addressingat least fourquestions:
1. Does SL instruction make a difference?
2. Does typeof instruction make a difference?
3. Does typeof learnermake a difference?
4. Does typeof instruction interactwithtypeof learner?
Answers to these questions are vital, not just for the credibilityof
TESOL as a profession,but also because theywill affectthe lives of
countless individuals, children and adults, for whom a SL is the
gatewayto educationand to economic and social survival.
U

ACKNOWLEDGMENTS
This is a revised versionof a paper presentedat the TESOL Research Committee's
session on the state of the art at the 16th Annual TESOL Convention,Honolulu,
Hawaii, May, 1982. I thankCathy Doughtyand JudithMatsunobufortheirhelp with
the literaturesearch. While writingthe paper, I received usefulcommentsfromAnn
Fathman,Steve Krashen,Tere Pica, Charlie Sato, and Herb Seliger.Special thanksgo
(again) to Steve Krashen and to Barry Taylor, each of whom provided detailed
feedback on the originalversion. Needless to say, not all of these people agree with
everything I have said here. Final responsibility
forany errorsof factor interpretation
.remainsmy own.

380 TESOL QUARTERLY

This content downloaded from 185.2.32.21 on Sun, 22 Jun 2014 16:49:07 PM


All use subject to JSTOR Terms and Conditions
THIE AUTHOR
MichaelLong,Assistant Professorin theDepartment of ESL at the Universityof
Hawaii,has co-editedseveralcollections and has publishedextensivelyin TESOL
as wellas in a widevarietyofprofessional
publications journals.Beforegoingtothe
Universityof Hawaii in 1982,he taughtESL/EFL and trainedteachersin Europe,
NorthAmerica, andLatinAmericaandspenttwoyearsonthefaculty oftheGraduate
Schoolof Educationat theUniversity of Pennsylvania.
He is currentlyco-editor
of
ResearchReportsforStudiesinSecondLanguageAcquisition andis a memberofthe
EditorialAdvisory
BoardoftheTESOL Quarterly.

REFERENCES
Bribre,Eugine J.1978.Variables affectingnativeMexican children'slearning
Spanishas a second language. Language Learning28 (1):159-174.
Brown, James D. 1980. Newly placed studentsversus continuingstudents:
comparingproficiency.In On TESOL '80, JanetC. Fisher,MarkA. Clarke,
and JacquelynSchachter(Eds.), 111-119.Washington,D.C.: TESOL.
Carroll, John B. 1967. Foreign language proficiencylevels attained by
language majors near graduationfromcollege. Foreign Language Annals
1:131-151.
Chihara, Tetsuro, and John W. Oller, Jr. 1978. Attitudesand attained
proficiencyin EFL: a sociolinguisticstudy of adult Japanese speakers.
Language Learning28 (1):55-68.
Clark, Eve V. 1978. Awareness of language: some evidence from what
childrensay and do. In The child'sconceptionof language,A. Sinclair,R. J.
Jarvella,and W. J. M. Levelt (Eds.), 17-43.Berlin:Springer-Verlag.
Farhady,Hossein. 1979. The disjunctivefallacybetween discrete-pointand
integrativetests.TESOL Quarterly13(3): 347-357.
Fathman, Ann. 1975. The relationshipbetween age and second language
productiveability.Language Learning25(2):245-253.
Fathman,Ann. 1976. Variables affectingthesuccessfullearningof Englishas
a second language. TESOL Quarterly10(4):433-441.
Hale, Thomas, and Eva Budar. 1970. Are TESOL classes the only answer?
Modern Language Journal54:487-492.
Krashen,Stephen. 1977. Some issues relatingto the monitormodel. In On
TESOL '77, H. Douglas Brown,Carlos A. Yorio,and RuthCrymes(Eds.),
144-158.Washington,D.C.: TESOL.
Krashen,Stephen. 1980. The inputhypothesis.In Currentissues in bilingual
education,JamesE. Alatis (Ed.), 168-180.Washington,D.C.: Georgetown
UniversityPress.
Krashen,Stephen. 1982. Principlesand practice in second language acquisi-
tion.Oxford:PergamonPress.
Krashen,Stephen,C. Jones,S. Zelinski,and C. Usprich.1978.How important
is instruction?
EnglishLanguage TeachingJournal32(4):257-261.
Krashen,Stephen, and Robin Scarcella. 1978. On routinesand patternsin
languageacquisitionand performance.Language Learning28(2):283-300.
Krashen,Stephen, and Herbert W. Seliger. 1976. The role of formaland

DOES INSTRUCTION MAKE A DIFFERENCE? 381

This content downloaded from 185.2.32.21 on Sun, 22 Jun 2014 16:49:07 PM


All use subject to JSTOR Terms and Conditions
informallinguisticenvironments in adultsecond languagelearning.Interna-
tionalJournalof Psycholinguistics 3-4(5):15-20.
Krashen,Stephen,HerbertW. Seliger,and D. Hartnett.1974. Two studiesin
adult second language learning.KritikonLitterarum3:220-228.
Martin,Gail M. St. 1980. English language acquisition:the effectsof living
withan Americanfamily.TESOL Quarterly14(3):388-390.
Mason, Charles. 1971. The relevance of intensivetrainingin English as a
foreignlanguage foruniversity students.Language Learning21(2):197-204.
Peters,A. M. 1980.Unitsof acquisition.University of Hawaii WorkingPapers
in Linguistics.
Pica, Teresa. 1983. An investigationof adult second language acquisitionin
threelanguage contexts.Language Learning33: in press.
Schneiderman,E. 1982. The modifiedstage hypothesis:a criticalevaluation.
Paper presented at the Fourth Los Angeles Second Language Research
Forum,Universityof California,Los Angeles,April,1982.
Seliger, Herbert W. 1979. On the natureand functionof language rules in
languageteaching.TESOL Quarterly13(3):359-369.
Swain, Merrill.1981. Immersioneducation: applicabilityfor nonvernacular
teachingto vernacularspeakers. Studies in Second Language Acquisition
4(1):1-17.
Upshur, John A. 1968. Four experimentson the relation between foreign
language teachingand learning.Language Learning18(1 & 2):111-124.

382 TESOL QUARTERLY

This content downloaded from 185.2.32.21 on Sun, 22 Jun 2014 16:49:07 PM


All use subject to JSTOR Terms and Conditions

You might also like