Professional Documents
Culture Documents
of Proficiency
Author(s): Peter Prince
Source: The Modern Language Journal, Vol. 80, No. 4 (Winter, 1996), pp. 478-493
Published by: Wiley on behalf of the National Federation of Modern Language Teachers Associations
Stable URL: http://www.jstor.org/stable/329727 .
Accessed: 28/06/2014 07:53
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.
Wiley and National Federation of Modern Language Teachers Associations are collaborating with JSTOR to
digitize, preserve and extend access to The Modern Language Journal.
http://www.jstor.org
ONE ASPECT OF LANGUAGE LEARNING linked to its native language equivalent. Many
thatin the past receivedlittleattention,but now teacherswork on the assumption that however
has become a focus of much research, is the effectivethisformof paired associate learning
learningofvocabulary.Increasingly,it is consid- (henceforward termed translation learning)
ered that effectivecommunication relies less mightbe, it tendsto reduce possibilitiesforsec-
upon the masteryof grammaticalrules than on ond language (L2) autonomyby linkingwords
the possession of an adequate and appropriate too exclusivelyto firstlanguage (Li) equiva-
vocabulary (Vermeer,1992). The size of the vo- lents; there is the added disadvantage that
cabulary needed to achieve general communi- learnersare likelyto suppose or look forequiva-
cation skills in English has been estimated at lence of meaning even when the contexts in
5,000 words,withcritical importance being at- which those words are used in differentlan-
tached to the learning of the most frequent guages clearly diverge (McCarthy,1990). As a
2,000 to 3,000 wordsas quicklyas possible (Na- result,most course books and teacherspresent
tion, 1993). L2 contextualized material right from the
Within this area of study,one question that outset. Compelling evidence that such an ap-
has yetto receive any precise formof answer is proach can be of benefitas faras receptivevo-
whetherthe learningof a new word best occurs cabulary is concerned emerges froma studyby
when the word is met in contextor in a "paired Saragi, Nation, and Meisler (1978), in whichthe
associate" condition, with the word being reading of a novel led to successfulrecognition
of 76% of newwordstested.Similarconclusions
about incidentallearningofvocabularyare pre-
TheModernLanguageJournal,80, iv (1996) sented by Nagy,Herman, and Anderson (1985).
0026-7902/96/478-493 $1.50/0 It is to be noted, however,thatwhen it comes
?1996 TheModernLanguageJournal
to recall, much of the research conducted in
this field has failed to demonstrate any clear texts, namely the L1 translation equivalent of
advantage of learning in context over transla- the word to be learned and an L2 sentence in
tion learning (Nation, 1982). Indeed, results whichtheword to be learned appears. Thus, the
provided by Seibert (1930) indicate that learn- word "context"willbe used in a narrowsense to
ing words in pairs is consistentlymore effective referspecificallyto the lattercondition, as op-
than learning them in context.Addressingthe posed to translationlearning in which the con-
same question, Pickering(1982) found learning text (in the wider sense) is provided by a single
in contextslightlymore advantageous, but not L1 equivalent.
enough for the differenceto be considered in It is also importantto specifythe use to which
any wayconclusive. In theirreviewof the ques- the context is put: Words presented in transla-
tion, Carterand McCarthy(1988) conclude that tion pairs are typicallythe focus of intentional
"it is difficultto draw precise lines to suggest learning, whereas an unknown word seen em-
when a move from key-wordtechniques, or bedded in an L2 text may frequentlyreceive
translation in pairs, or from using a mono- only such attentionas preventsit fromimped-
lingual or bilingual dictionaryto context-based ing comprehension-that is, the reader might
inferentialstrategies,is best instituted"(p. 15). choose to ignore theword ifhe or she considers
They suggestthatthere are "no clearlymarked thatit does not affectgeneral understandingof
stages of transition" in the learning process, the text. Hulstijn (1992) notes that the chance
and that therefore "a mixture of approaches of readers rememberinga word afterencoun-
should be adopted" (p. 15). One reason forthis tering it in a text that is read for its contentis
recommendationof a mixtureof approaches is verysmall indeed. In the experimentdescribed
that both methods have advantages and disad- in thisarticle,the contextwas to be used specif-
vantages,whichwill be discussed in the conclu- icallyforthe learningof new vocabularyforthe
sion to the present paper. purposes of production.
Set against the uncertaintyhighlighted by Another element that appears to be relevant
Carterand McCarthy(1988), the sheer abilityof is the degree of proficiencyof the learner.Evi-
learners to assimilatelarge quantities of words dence provided by Kroll and Curley (1988) sug-
with their translationsis impressive,although gests that in the initial stages of learning,new
there are wide disparities between individual words are stronglylinked to their L1 equiva-
learners (Webb, 1962). This abilitymay corre- lents,and a shiftoccurs afterabout 30 months
spond to, and reinforce,a preferentialstrategy of study,such thata networkof linkswithinthe
that learners adopt in the earlystages of learn- L2 begins to become effective. The data of
ing and subsequently find difficultto relin- Kroll and Curley come froma picture-naming
quish (Sautermeister,1989). The obvious short- experimentin which response times in the L2
coming of such a strategyis that it favoursthe decreased when subjects had more than 30
creation of a single "trace"'1in memoryover the monthsof studybehind them.The authorscon-
multipletraces provided by contextualencoun- cluded thatsubjects had thereforeformeda di-
ters. Thus, although the informationmay be rectlinkbetween the L2 wordsand the underly-
well established in memory,it is not necessarily ing concept, and were no longer passing byway
available foruse in appropriate contexts(Wise- of theirL1 in order to name the picturein their
man & Tulving,1976). L2. Although the workof Kroll and Curleycon-
One difficulty in this area of research, cerned lexical access as opposed to vocabulary
pointed out by Nation (1982), is thatthereis no learning,it seems likelythat the patternsof ac-
standarddefinitionofwhatis meantbycontext. cess followthose established during the learn-
In its broadest sense, of course, contextmaybe ing phase; in otherwords,in the firstmonthsof
said to comprise all the perceived phenomena learning, L2 words are more effectively stored
that accompany the processingof a given stim- in memory when they are linked to their L1
ulus, including the physical surroundings in equivalents.
which learning takes place. In this sense, a na- Gekoski (1980) has described this phenome-
tive word that serves as the basis for the learning non by statingthat at lower levels of L2 profi-
of a foreign word is a major element of the con- ciency,learners use L1 mediation in order to
text. Other possible contexts are pictures or L2 translatetheir"thoughts"into the L2. Because
synonyms (R6hr, 1993). Although a comparison most of the words encountered in the initial
between these different types of context would stages are of a concrete, imageable nature, it
undoubtedly be of interest, the present paper maybe thatas soon as learnershave ascertained
concentrates on the two most widely used con- the meaning of the word,eitherbyaccessing an
imagen store (Paivio & Desrochers, 1980) or pointed out by Hulstijn (1992), learners fre-
with the help of the teacher's explicit use of a quentlymake incorrectguesses about meanings
picture,theyneglect the L2 contextin order to because L2 contextsdo not alwaysprovidesuffi-
attach the new word formto the L1 expression cientinformationto make a correctguess possi-
of the meaning. Thus, in effect,the learning ble even when a learnerhas been trainedto use
occurs in a translationcondition. all the semantic and syntacticclues available.
It should be noted, however,that although Second, trainingis profitableonlywhen put to
the Kroll and Curley (1988) data corroborate effectiveuse. Over and above the specific tech-
intuitionsabout vocabularylearning in the be- niques fornoticinglinguisticclues, R6hr points
ginningstage, the shiftawayfromexclusiveuse to the importance of fostering pleasure in
of the translationlink afterabout 30 monthsof learners when theysearch for the meanings of
studydoes not implythatlearning does not re- words. This is arguablythe most importantas-
main more effectivein the translation condi- pect of any language teacher's work, yet one
tion. There are tworeasons forthis.First,as has which,in France at least, receivesverylittleat-
alreadybeen mentioned,the studyof lexical ac- tention in teacher training programs (except
cess does not necessarilyinformus about the those directed at teaching children as opposed
most effectiveconditions for learning. Second, to adolescents). Unfortunately, as Krashen
Kroll and Curleyexamined access to individual (1987) remarks,most nonspecialist L2 learners
words,whereas a context typicallycomprises a are not dedicated linguists, and when faced
sentence,withall the syntacticconstraintsthat with a choice between a high-effortstrategy
it entails. It is possible thatthe extra processing such as inferencingand a low-effortshortcut
of syntax required to understand a sentence such as translationlearning, theywill tend to
preventslearners from efficientlylearning vo- choose the latter.
cabulary in context for a period considerably Althoughtraininglearnersto elaborate meta-
longer than 30 months. Indeed, Sautermeister cognitive strategies is of undeniable impor-
(1989) reports vocabulary learning behavior tance, the experiment described in this study
among university,nonspecialist learners of does not examine the effectsof such explicit
English, who are consistentlypresented with training, but concentrates instead on the
new words in context,but who are not satisfied information-processing strategiesdeveloped by
until theyhave found an L1 equivalent to assist learners who have followed their natural incli-
their learning. This behavior is similar to that nations in vocabularylearning. In otherwords,
described for beginners. the subject population under scrutinyconsists
In summary,it appears thatlearningvocabu- of nonspecialist learners of English who have
laryin contextis widelyperceived by the teach- received only minimal training in metacogni-
ing profession as desirable, but that students tive strategiesand whose performanceis of in-
eitheractivelyresistit, believingthe translation terestfor that very reason. How efficientare
condition to be superior,or fail to elaborate the such learners in guessing the meaning of un-
strategiesthatmightmake it possible. This does knownwords,and to what extentare theyable
not mean, however,thatcontextlearningis nec- to learn words presented in L2 sentence con-
essarily ineffectivebecause it can be argued texts?If theycontinue to show a preferencefor
that learners are unlikelyto adopt appropriate translationlearning, is this indeed a handicap
strategiesunless theyare given guidance. The when it comes to accessing theirknowledgefor
question of strategiesis thus germane to the use in differentcontexts?The answersto such
issue of vocabulary learning, as indeed it is to questions will hopefullybe useful in making
L2 learning in general (O'Malley & Chamot, teachers more keenly aware of the processing
1990). Proposals for improvinglearners' ability strategiescommonlyadopted in inferencing,so
to use contextualclues to ascertainmeaningsof thateitheralternativeor complementarystrate-
unknownwordshave been put forwardbyR6hr gies may be proposed.
(1993) in theformof a methodicaltrainingpro- The method adopted in the experimentde-
gram. Given the rightguidance, therefore,it is scribed in the presentstudywas to compare per-
likely that even beginning learners can use formance in vocabularylearning as a function
metacognitive strategies and metalinguistic of proficiencyof the learner,condition of learn-
knowledge to enhance their comprehension ing (translationvs. context),and condition of
skillswhen a textcontains unknownwords. De- recall (again, translationvs. context). The aim
spite the benefits of such training, however, was to assess the abilityof learners of different
thereare two limitsto its effectiveness. First,as levels (a) to exploit sentence context informa-
tion both in learning and recall and (b) to translation.Alternatively, given thatthe taskre-
transferwhat theyhave learned fromone con- quired retrieving the translationsof words that
text to another. The firstpoint concerns the were actually presented, a possible strategy
level of proficiencyat which an L2 textstartsto could have been to ignore the context alto-
provide a useful basis fromwhich to learn new gether and concentrateon the recall aspect of
words. It attemptsto shed lighton the question the task. Clearly,however,the weaker subjects
raised by Carter and McCarthy (1988) about were unable to put eitherof thesestrategiesinto
when it is beneficial to move fromone strategy effectand sufferedfromover-stimulation. The
to another.The second point is also important effortrequired to understand the text appar-
because effectivelearning for the purposes of ently prevented them from using strategies
production can be said to have taken place only available to them.
when the informationstoredas a resultof learn- In view of the preceding remarks,it was pre-
ing in one contextcan be retrievedand used in dicted thatin the presentexperiment,advanced
another.Presumably,teachersencourage learn- learnerswould make more efficientuse of con-
ers to learn new words directlyin L2 sentence text than weaker learners, both during the
contextsbecause theyfeel thatlearningnew vo- study phase and during recall, but that little
cabularyin the restrictivecontextof translation differencewould emerge as regardsthe transla-
linksis not conducive to the retrievaland use of tion condition. Althoughthisresultwould con-
thatvocabularyin the richercontextsof L2 sen- formboth to the intuitionsand observationsof
tences. However,such a belief, although intu- teachersand to the data providedbyCohen and
itivelyplausible, remains to be demonstrated Aphek (1980), it is importantto analysethe pos-
empirically. sible causes as a firststep towardsestablishinga
One limitation of the experimental studies model of the wayL2 learnerscompensate fora
conducted by Seibert (1930) and by Pickering lack of lexical knowledge when processing
(1982) was that the effectof proficiencyupon sentences.
learners' performancewas not examined. Pick- Cohen and Aphek (1980) did not examine
ering gives no indication of level other than to differentconditions of learning,only of recall;
say that the subjects were in secondary school. however,it is to be expected that the effortre-
Althoughthese subjects found contextlearning quired to process textwould make learningnew
no easier than translationlearning,we cannot wordsin an L2 contextconsiderablymore diffi-
know if this resultwould still be obtained with cult than learning with translations.Although
more advanced learners. manytypesof processingare possible in transla-
Cohen and Aphek (1980) examined the effect tion learning-such as using keyword tech-
of proficiencylevel in their study of Hebrew niques (Pressley,Levin, Hall, Miller, & Berry,
vocabularylearningby native English speakers. 1980; Roediger, 1980)-there is only one man-
They found that their more advanced learners datorystep,thatis, associatingthe L2 wordwith
displayed clearly superior performance in re- a familiar,easilyretrievableword in the L1.
call when the recall task involved L2 contexts. Learning in context, however,requires that
Specifically,subjects were required to provide the subject (a) process the sentenceto gain suf-
the translationsof certain words,presented ei- ficientunderstandingof it, (b) use thatunder-
ther in Hebrew or in English,withina Hebrew standing to inferthe meaning of the unknown
text. In a pure translation task, on the other word, and (c) associate the meaning with the
hand, no differenceappeared between the two formof the unknownword in such a waythata
groups of learners. Cohen and Aphek (1980) representation is formed that is available for
conclude that context only provides a useful futureuse. The thirdstep mayinvolveaccessing
basis once learners have reached a level where the translationequivalent in the L1, thus in ef-
theyare not "over-stimulated"byan L2 context. fect revertingto translationlearning, or form-
This explanation is plausible ifone considers ing a representationthatfitsdirectlyintoan L2
that subjects' controlled attentional resources network.These two processes are not mutually
are limited (Shiffrin& Schneider, 1977) and exclusive and may indeed occur in parallel.
thatthe comprehensionof a text,less automatic Whateverthe exact nature of the links thatare
in a weaker learner, diminishes the resources formed,it is clear that the effortrequired for
normallyavailable forthe recall of the word. In contextlearning is greater than that for learn-
theory,however,the recall task used allowed ing with translations,and such an effortmay
subjects enough time to proceed serially,first put a limiton the quantityof material thatcan
understandingthe textand then retrievingthe be learned in thisway.
the week, they received homework in which meanings of the missingwords. In order to en-
none of the experimentalwords appeared. sure thatthiswas also the case during the recall
During the studyphase, subjectswere divided phase of the experiment itself,subjects were
into two groups, correspondingto the two mo- asked to fillin witha word in French the blanks
dalities of the factorL, the learning condition. theyhad not filled in during the test. This in-
Half of themreceived a sheetwiththe 44 words structionwas onlygivenat the end of the exper-
and their translationsand were instructed to imentin order to preventsubjects fromresort-
learn them (TL condition). The other half re- ing to such a strategytoo rapidly.
ceived the list of 44 sentences that had been
used in the pretest,but thistimewiththe target
SUBJECTS
wordsappearing in place of theblanks. Subjects
were instructedto read the sentences,to guess Forty eight students, enrolled at the Phar-
the meaning of the unknown words, and to macy Faculty of the Universityof Montpellier,
learn them (CL condition). During this phase, took part in the experiment on a voluntary
subjects were not allowed to write. basis. All the subjects had been studying
The study phase was of roughly the same Englishforbetween 5 and 8 years;Englishwas a
lengthfor the two groups, in otherwords 20 to compulsorysubject in their universitystudies.
25 minutes,or about 30 seconds per word to be For the purposes of the experiment,theywere
learned. Thirtyseconds per word was the time divided into twogroups,weak and advanced, on
chosen by Pickering (1982), whose method was the basis of a placement test taken shortlybe-
differentin so faras the material to be learned fore the experiment.The testadministeredwas
was projected onto a screen, so that the time theTOEFL, on whichtheweak group scored an
each item was attended to could be controlled. average of 397 and the advanced group 480.
Despite being a morerigorousexperimentalpro- The placement test was their firstencounter
cedure, thismethodwas consideredproblematic withthe TOEFL. Subsequent testsrevealed that
in thatlearnerswere unable to use the strategies both groups were able to improve their scores
theynormallyadopt, such as formingassocia- substantiallyas soon as theybecame familiar
tions between items. The time constraintim- with the specificitiesof the examination. The
posed in thepresentexperimentwas nonetheless main skillnot testedbythe TOEFL, namelyoral
relaxed to allow certainsubjectsto complete,or expression, was also assessed separately on a
consider thattheyhad completed,the task. subsection of four studentsfrom each group,
Afterthe studyphase, whichtook place at the who were asked to tell a storyas depicted by a
beginning of a normal language class, subjects comic stripwithoutcaptions. On a number of
moved on to a differentactivityfor about 40 criteria, ranging from ease of expression (as
minutes.In the course of this activity,theyen- measured by the number of words produced
countered none of the 44 words theyhad stud- per minute) to grammaticalaccuracy,the gap
ied. Then the recall phase took place, lasting between the two groups was equally as wide as
about the same time as the studyphase. Again, that suggested by the TOEFL scores. Although
the slowersubjects were given a fewextra min- both groups had studied English for roughly
utes to complete the test. For this part of the the same amount of time, a questionnaire re-
experiment,each subject received a sheet con- vealed that the weaker group had rarelyfound
taining 22 words to translate (TR condition) the subject motivatingor participated actively
and 22 English sentencesthatincluded a blank in class. Their general L2 performance was
to be filled with one of the targetwords (CR characterisedby a lack of confidence, a fear of
condition). In the TR condition, half the sub- takingrisks,and a markedtendencyto revertto
jects received French words to translate into theirL1 wheneverpossible.
English, the other half English words to trans- As regards the vocabulary learning pro-
late into French. The sentencesused in the CR cedures previouslydeveloped by subjects, the
condition, differentfrom those seen in the highlycentralizedand tightlycontrollednature
studyphase, had been tested previouslyon a of the French educational systemmeans that
group of eight learnersof the same level as the these maybe regarded as broadlysimilarforall
subjects belonging to the weaker group in this learners;thatis to saythatvocabularyand gram-
experiment.In this test, learners indicated in mar are L2 contextualizedfromthe outset,but
French the word that was missing in the sen- specific trainingin metacognitivestrategiesis
tence, thus demonstrating that the contexts at presentsporadic or minimal.Withinthisgen-
presented did indeed allow them to access the eral framework,however, there is obviously
TABLE 1
Percentageof CorrectResponsesbyGroup,LearningCondition,and Recall Condition,Plus Percentage
of Total CorrectResponsesbyLearningConditionand Recall Condition
TL CL M
Weak Learners TR 85.81 44.52 65.17
(Group 1) CR 32.37 32.22 32.30
M 59.09 38.37 48.73
AdvancedLearners TR 74.54 56.84 65.69
(Group 2) CR 74.03 63.92 68.97
M 74.29 60.38 67.33
TR 80.18 50.68 65.43
Total CR 53.20 48.07 50.64
M 66.69 49.38 58.03
Note.TL = translationlearning;CL = contextlearning;TR = translationrecall;CR = contextrecall.
TABLE 2
Percentageof CorrectAnswersin TranslationRecall,as a Functionof Directionof Translation
Weak Learners AdvancedLearners
(Group 1) (Group 2)
L2-L1 L1-L2 L2-L1 L1-L2
TL 94.55 77.07 TL 73.82 75.27
CL 42.76 46.28 CL 62.90 50.78
Note.TL = translationlearning;CL = contextlearning.
use low-effort strategiesthatlead to rapid learn- the "cost" involved in arriving at an under-
ing but are ultimately ineffective in using standingof the sentence. There are likelyto be
broader skills. In view of this, it seems impor- two causes of this cost. First, learners might
tant,as R6hr (1993) stresses,to fosterthe pleas- process textsmore slowlybecause theirL2 net-
ure of exploring L2 lexical relationshipsat the works are not as richlydeveloped as those of
earliest possible stage of learning. more advanced learners. For example, in the
weaker learner's mental lexicon, the word
The Use ofContext "cut," although a well-known,high-frequency
word,does not have as manylinkswithotherL2
Accounting for the differencebetween the words as foran advanced learner,who mayhave
two groups in the experimentrequires a theory formed links with less common words,such as
of the taskunder consideration.One important "slice," "chop," "scissors," or "shortcut." The
element of context theoryis the semantic gap basic premise here is that the richer the net-
that is created by the presence of an unknown work in which a word is integrated,the more
word. Each gap may be said to have a certain automatically the word is accessed, both in
shape, the contours of which are fixed by the meaning and in form (Keatley, Spinks, & DE
clues provided by the other words in the sen- Gelder, 1994). Thus it has been consistently
tence. Take an example fromthe experimental shown thatword recognitionis slowerin an L2
material:"Let's make dinner together,shall we? than in one's L1 (e.g., Favreau & Segalowitz,
Ifyou peel the potatoes, I'll cut themand make 1983), and, although the differencemay only
the chips." The shape of the gap is determined consist in tenthsof a second, it is increased as
mainlyby the word "potato," which,when com- task load increases.
bined with the reader's previous knowledge of Second, the weaker learner will be slower to
chip-making,servesto create a mentalmodel of use syntacticinformation.Indeed, research by
the situation (Johnson-Laird,1983). Before cor- Barnett(1986) indicatesthata poor level of syn-
rectly ascribing a meaning to the word "peel," tactic knowledge diminishes reading profi-
however,the learnermusteliminateotherpossi- ciency to the point where understandingis not
bilities by noticing, for example, that the cut- improvedeven by increased lexical knowledge.
ting involved in chip-making is already ac- One implication of this could be that when
counted for,so peeling is probably a different readers stumble over syntactic analysis, they
activityfrom cutting. The most plausible re- cannot make the necessaryinferencesto arrive
maining candidate forthe meaning of "peel" is at the meaning of an unknownword. Thus, in
"to removethe skin." In uncontrivedsentences, our example, the words "let's" and "together"
of course, the contoursof the gap are rarelyso presuppose that the situation involves more
well defined,so in practice manysentencesmay than one person, and the words "you" and "I"
be required before the featuresthat make up in the second half specifythe number as two.
the semantic domain of a word are fullyascer- The word "them" must also be understood as
tained. Indeed, it could be objected thateven in referringto "potatoes." Even where a sentence
the sentences used in the present experiment, contains no potentiallyambiguous words, our
meanings could not be guessed unequivocally: weaker subjectsmaysimplyhave been less sensi-
It is also typicalto wash potatoes beforecooking tive to these anaphoric elements, thus ham-
them, so "peel" could be construed as a syn- pered in their abilityto guess the meaning of
onym for "wash." The fact that subjects were unknown words. It should be remembered,
able to fill in the blanks with the correct L1I however,thatthe materialin the presentexperi-
word during the pretestargues against thispos- mentwas testedon learnerswhose levelwas sim-
sibility, but theymaywell have feltless confident ilar to that of the experimental subjects, who
than the strongergroup that their guess was found no difficultyin providing the correct
indeed correct.This lack of confidence maybe missingword in theirL1, and thatat the end of
caused by the elaboration of a mental model the experiment,the subjects themselveswere
thatis less precise and in which the gap created also able to provide the appropriate L1 word.
by the unknownword is correspondinglyfuzzy. Thus, it cannot be stated that the weaker sub-
Whyshould thisbe so when the sentence con- jects were preventedfrommakingcorrectinfer-
tains onlyone unknownword?The explanation ences, only that theywere slower to do so. A
offeredhere is an elaboration of what Cohen broader claim thathas been made is thatwhere
and Aphek (1980) presumably mean by the syntacticprocessing is costly,L2 reading tends
"over-stimulation" of textualprocessing,namely to be more conceptuallyguided (Ulijn, 1980).
view, in a one-to-one correspondence to L1 1987; Nagy, Herman, & Anderson, 1985), thus
words. contributingto an increase in receptivevocab-
2. An inabilityto exploit the L2 lexicon effec- ulary,it is probable thatthe formationof a rep-
tivelyfor the purposes of production. resentationthatremains accessible forthe pur-
3. Relativelyeffortfulprocessing of L2 sen- poses of production requires more elaborate
tences due to slower word identificationand processing than that provided by inferencing.
syntacticprocessing. One formof processingwould be to pay atten-
These three characteristicsdo not preclude tion specificallyto the formof the word, both
the existence of a fourth,namely the develop- orthographic and phonological, by, for exam-
mentof automatic linkswithinthe L2 network, ple, repeating it aloud (Ellis & Beaton, 1993);
along semantic and associative lines, for the another would be to associate the new word
purposes of recognition. This fourth charac- consciously with words already known. A fur-
teristicis not in contradictionwith the third, ther stage would involve the retrievalof newly
which concerns the recognition of word forms learned words in a varietyof production tasks
as opposed to word relationships. in which the clues provided draw upon all the
In lightof these comments,what conclusions semantic and associative links that contribute
may be drawn for teaching practice? One gen- to the meaning of the word to be learned.
eral pointis thatthe developmentof metacogni- Where the aim is to master the most common
tive strategiesand attitudesis of prime impor- words in a language rapidly (Nation, 1993),
tance in this domain (O'Malley & Chamot, such an approach is especially importantbe-
1990; Oxford, 1990). Notably,initial expectan- cause it cannot be assumed thateven frequently
cies as to the effortinvolvedin language learn- encountered words are readilyavailable forex-
ing should be probed and discussed so thatthe pression.All thesestrategiesforlearningvocab-
pitfallsof low-effortstrategieslike translation ulary involveisolating the word fromthe con-
learning are well understood from the begin- text, so that context provides the means to
ning. In the absence of such discussion, some identifythe meaning of the new word and not
learnersare liable to focus on the development necessarilythe means to learn it. Although a
of rapid strategies for understanding, to the sufficiently imageable contextmaydo thatalso,
detrimentof the more effortfulprocesses in- the value of contextlies above all in its authen-
volved in production. These observations cor- ticity,the benefitsof which are of three differ-
roborate and extend the output hypothesisas ent sorts.
elaborated by Swain (1985). First,assessingthemeaningofa wordin context
Turning to the more specific question of obliges the learnerto develop strategies,such as
learningwordsin contextor withtranslations,it anticipatingand inferencing,whichbecome in-
is noteworthythatwhen the correcttranslation creasinglyprofitableas learning progressesbe-
link receives sufficient attention during the cause theyinstillan attitudeof self-reliancethat
studyphase, subjects' abilityto retrieveit is not is the hallmarkof proficiency.Second, system-
in doubt. The weakness lies less in the retrieval atically meeting new words in context under-
process itselfthan in the use to which the infor- lines the factthatwords are indeed used in dis-
mation is put once retrieved.In other words, course forpurposes of communication.Finally,
weaker subjects appear unwilling to envisage context provides an indication of the way the
the use of an L2 other than in a translation words are used. All these factorsmaybe said to
situation.Here again, the problemis one of atti- contribute to a learner's L2 autonomy and to
tude ratherthan of processing skill itself,indi- facilitatethe transferof knowledgethataccom-
cating a need to remedythe situationnot only panies it. They also point to the fact that the
by practice at developing the appropriate skills mental representationof a word's meaning de-
but also byworkingupon learners' perceptions velops with successive encounters in different
of long-termaims and of the ways the task at contexts,so it mayin principle alwaysundergo
hand contributesto them. modificationas new and finersemanticdistinc-
A further aspect, pointed out by Nation tions are added (Monsell, 1985).
(1982) and Hulstijn (1992), is that a distinction Despite the benefits of contextual presenta-
should be drawn between reading for compre- tion, it appears thateffectivelearning of words
hension and reading as a means to increase requires a stage in which the word is in fact
one's vocabulary.Althoughreading forcompre- isolated fromits contextand submittedto elab-
hension no doubt gives rise to incidental learn- orative processing. The extent to which links
ing (Herman, Anderson, Pearson, & Nagy, between concepts and L2 word forms can be
Ulijn, J. (1980). Foreign language reading research: Webb,W. B. (1962). The effectsof prolonged learning
Recent trendsand futureprojects.JournalofRe- on learningvocabulary.JournalofVerbal Learning
searchin Reading,3, 17-37. and Verbal 1, 173-182.
Behavior,
Vermeer,A. (1992). Exploring the second language Wiseman, S., & Tulving,E. (1976). Encoding specific-
learner lexicon. In L. Verhoeven& J. De Jong ity: Relations between recall superiority and
(Eds.), Theconstruct
oflanguageproficiency.
Applica- recognition failure.JournalofExperimental Psy-
tionsofpsychologicalmodelsto languageassessment chology:HumanLearningand Memory, 2, 349-361.
(pp. 147-162). Amsterdam/Philadelphia:John
Benjamins.
APPENDIX