Sei sulla pagina 1di 7

Duke University Press

Porter Institute for Poetics and Semiotics


Sign and System of Language: A Reassessment of Saussure's Doctrine
Author(s): Roman Jakobson and B. Hrushovski
Source: Poetics Today, Vol. 2, No. 1a, Roman Jakobson: Language and Poetry (Autumn, 1980), pp.
33-38
Published by: Duke University Press
Stable URL: http://www.jstor.org/stable/1772350
Accessed: 23-10-2015 10:18 UTC

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.

Duke University Press and Porter Institute for Poetics and Semiotics are collaborating with JSTOR to digitize, preserve
and extend access to Poetics Today.

http://www.jstor.org

This content downloaded from 216.165.126.139 on Fri, 23 Oct 2015 10:18:42 UTC
All use subject to JSTOR Terms and Conditions

SIGN AND SYSTEM OF LANGUAGE

A Reassessment
ofSaussure's
Doctrine*
ROMAN JAKOBSON

It is remarkablethat Suassure's Cours de linguistiquegendralewas frequently


mentionedin thissymposium,as if one wishedto establishwhathas changedin
thebasic assumptionof generallinguistics
overthe fifty
yearswhichseparateus
fromthe lecturesof the Genevan master.For the theoryof languageand for
as a wholeit was indeedhalfa centuryof cardinaltransformations.
It
linguistics
seemsto me thatour fruitful
discussionconveysa clearnotionas to whatin this
famousheritagerequiresfar-reaching
revisions,and whichpartsof Saussure's
in
the
version
edited
his
teaching
by
pupils - remainsvalid to thisday.
Of the two basic prinicplesof the Cours, les deux principesgendraux,as
Suassure labeled them, one may see today the firstbasic proposition I'arbitrairedu signe,the "arbitrariness"
of the languagesign - as an arbitrary
As
Beneviste
has
shown
it
in Acta LinguisticaI, fromthe
principle.
beautifully
of
view
of
a
synchronic
point
languagecommunity
usingthelanguagesigns,one
mustnot ascribeto theman arbitrary
nature.It is not at all arbitrary
butrather
to
for
"cheese"
in
and
to
in
cheese
obligatory sayfromage
French,
say
English.I
believethatone mayconcludefromthewholediscussionon "arbitrariness"
and
"unmotivated"signs,thatI'arbitrairewas a mostunfortunate
choice of a term.
This questionwas dealtwithmuchbetterbythePolish linguistM. Kruszewski,a
of Saussure(and highlyestimatedbythelatter),as earlyas in the
contemporary
betweentwobasic factors
beginningof the 1880s.Kruszewskimade a distinction
in the lifeof a language,two associations:by similarity
and by contiguity.
The
relationbetweena signansand a signatum,whichSaussurearbitrarily
described
as arbitrary,
is in realitya habitual,learnedcontiguity,
whichis obligatoryforall
membersof a givenlanguage community.But along withthis contiguitythe
la ressemblance,assertsitself.As it was mentionedhere,
principleof similarity,
and as Kruszewskialreadysaw it, thisprincipleplays an enormousrole in the
area of derivationsand in the area of word families,wheresimilarity
between
*Lecturein Erfurt,East Germany,2 Oct. 1959,at the Ist International
Symposium"Signand System
of Language,"publishedin Germanin R. Jakobson'sSelectedWritings,
Vol. I (The Hague: Mouton,
1971). AuthorizedtranslationfromGermanby B. Hrushovski.
@ Poetics Today, Vol. 2:la (1980), 33-38

This content downloaded from 216.165.126.139 on Fri, 23 Oct 2015 10:18:42 UTC
All use subject to JSTOR Terms and Conditions

34

ROMANJAKOBSON

wordsof one root is decisive,and whereit becomesimpossibleto speak about


In morphophonological
arbitrariness.
is
issues,thequestionof similarstructures
of primaryimportancewhenwe recognizethatthereexistcertainmodels,certain
structural
and selectionof phonemesin roots,and other
typesof thedistribution
of
or
of
derivation
suffixes
and conjugation.Finally,theissue of
types prefixes
soundsymbolism,on whichI shallnot further
dwellhere,remains,in spiteof all
in
an
voiced
the
and
skepticism
past, important fascinating
problemin thestudy
of language. And so are all questionsconcerningthe foundationof language
symbolsin image and indication(or, as CharlesSandersPierce,the pioneerof
thetheoryof signs,wouldhave said: theproblemof iconicor indexicalsymbols).
It seems to me that the second principlein Saussure's Cours, the so-called
linearite,mustalso be seenas a dangeroussimplification.
Actuallywe encounter
two-dimensional
unitsnotonlyon thelevelof thesignatum,as demonstrated
by
Ch. Bally,butalso in thefieldof thesignans.If we recognizethatthephonemeis
not the ultimateunitof language,but can be decomposedinto distinctivefeathatwe mayspeakin phonologytoo about two
tures,thenitbecomesself-evident
and of
dimensions,(as we have accordsin music),thedimensionsof successivity
This, however,mustlead to abandoninga numberof Saussure's
simultaneity.
theseson basic laws of languagestructure.
Thus, I believethattheterm"syntagrelationswethink
to syntagmatic
matic"is oftenmisleading,sincewhenreferring
in time;however,besidesthecombinationin temporalsuccession,
of successivity
we mustdeal also withcombinationof simultaneousfeatures.It wouldbe advisable in this respectto speak simplyabout combination,seen as contrastedby
another factor,namely,selection.Selection of units or of combinations,in
contrastto combinationperse, belongsto theparadigmaticlevelof language.It
is substitution,as distinguishedfromboth simultaneityand successivity.In
assertsitself.
selection,theprincipleof equivalence,or associationbysimilarity,
and
than
axes
rather
Whileobservingtheparadigmatic
successivity simultaneity,
I do not believethatwe abandon the domain of the objectiveand plungeinto
Linguisticresearchesof recentyearshave shownthatin thisarea an
subjectivity.
of components,exists.One encountershere
a hierarchy
objectivestratification,
theproblemof primaryand secondaryfunctions,
theproblemof predictability,
whichhas been outlinedbrilliantly
by Kurytowiczin thethirtiesand whichhas
transformations
been recentlydevelopedin Americain thetheoryof syntactical
- one of themosttopicalproblemsof linguisticanalysis.At thesame time,the
and
and indispensablequestionarises,as to therelationship
evenmoreimportant
betweenparadigmaticseriesand combinationalseries(chainsor
the difference
clusters).
idea of
as in all modernsciences,withthesignificant
We deal here,apparently,
variantson the
invariance.We speak about combinational,context-dependent
levelof sound as wellas on thelevelof grammar.But it would be impossibleto
speak about variantsas long as we have not clarifiedthe natureof the basic
invariant,the unitto whichall these variantsare related.The search for the
invariantsis now the most substantialproblemnot only in phonology,but in
grammaras well. When dealing withthe sign, the bilateralsignumas a link

This content downloaded from 216.165.126.139 on Fri, 23 Oct 2015 10:18:42 UTC
All use subject to JSTOR Terms and Conditions

SIGN AND SYSTEM OF LANGUAGE

35

betweenthesignansand signatum,how do we discoversuch invariantson one


hand in the domain of the signansand on the otherhand in the fieldof the
betweenthetwo, froma linguisticpointof view,
signatum?The basic difference
is that the signans must necessarilybe perceptiblewhereasthe signatumis
translatable.In bothcases theprincipleof equivalenceobtains.In thedomainof
the signans the relativeequivalencemust be externallyperceivable;it can be
ascertained,however,onlyin respectofthefunctionof thesesoundrelationsin a
featuresand, by meansof a specgivenlanguage.We recognizesuchdistinctive
trograph,we are able to translatethemfromthe acoustic fieldinto the visual
level.And likethesignans,thesignatumtoo mustbe studiedin a purelylinguistic
and objective manner.A purelylinguisticsemanticscan and must be constructed,ifwe agreewithPeircethatthebasic propertyof anyverbalsignlies in
its capability of being translatedinto another verbal sign, either a more
deployed,explicitsign,or, on the contrary,a more ellipticalsign,of the same
one. Thistranslatability
languagesystemor of a different
laysbarethatsemantic
invariantforwhichwe are searchingin thesignatum.In such a way it becomes
possible to submitsemanticproblemsof language to distributionalanalysis.
sentences,suchas "A roosteris a maleof a hen"belong
Metalinguistic
identifying
to thetextinventory
of theEnglishlanguagecommunity;
thereversibility
of both
"A
of
male
a
hen
a
is
rooster"
demonstrates
how
the
expressions
meaningof
wordsbecomesa real linguisticproblemthrougha distributive
analysisof such
commonmetalingualutterances.
Among the basic featuresof the Cours de linguistiquegednraleis the split
natureof linguistics:
theseparationof synchrony
fromdiachrony.The thorough
work done over severaldecades in both partial areas, as well as the refined
methodologydevelopedin this research,broughtabout a seriousdangerof a
flagrant gap between these two descriptions,and also the necessityof
of thecontrastbetweensynchrony
overcomingthisgap. Saussure'sidentification
and diachronywiththecontrastbetweenstatisticsand dynamicsturnedoutto be
is not at all static;changesare always
misleading.In actual realitysynchrony
Actual synchronyis dynamic.Static
emergingand are a part of synchrony.
is an abstraction,whichmaybe usefulto theinvestigation
of language
synchrony
for specific purposes; however,an exhaustivetrue-to-thefacts synchronic
considerthe dynamicsof language.
descriptionof languagemustconsistently
Both elements,the pointof originand the finalphase of any change,existfor
some time simultaneouslywithinone language community.They coexist as
variants.Whentakingthisimportantfactintoconsideration,we realize
stylistic
thattheimageof languageas a uniformand monolithicsystemis oversimplified.
Language is a systemof systems,an overall code which includes various
subcodes. These variagated language styles do not make an accidental,
mechanical aggregation,but rathera rule-governedhierarchyof subcodes.
Though we can tellwhichof the subcodesis the basic code, it is nevertheless
a
to excludethediscussionof theothersubcodes. If we
dangeroussimplification
considerlangueas a totalityof theconventionsof a language,thenwe mustbe
verycarefulnot to be researchingfictions.

This content downloaded from 216.165.126.139 on Fri, 23 Oct 2015 10:18:42 UTC
All use subject to JSTOR Terms and Conditions

36

ROMANJAKOBSON

I believethat today our chieftask should be to become realists,to build a


realisticstudyof languageand combatanyfictionalismin linguistics.We must
ask ourselves:what is the real linguisticconventionthat enables exchangeof
and serveseffectively
thevarioustasksof
speechin a givenlanguagecommunity
communication?Some linguistsask, whyshouldlinguisticsdifferfromphysics
in its methodology?Why could not the scholar of language impose his own
systemof symbols,his creativemodel, upon the investigatedmaterial,as is
commonin thenaturalsciences?Indeed,one observes,in manyrespects,an ever
more meaningfuland fruitfulcontact betweenthe natural sciencesand linas well. In
one mustkeep in mindthe specificdifferences
guistics;nevertheless
cardinal
the
theLondon school of mathematicalinformation
theory
difference
was clearlyrecognizedand theproblemof communicationwas separatedfrom
otheraspects of information.First of all, one must distinguishbetweentwo
classesof signs- indicesand symbols,as Piercecalledthem.Indices,whichthe
physicistextractsfromthe externalworld, are not reversible.He transforms
theseindicesgivenin natureinto his own systemof scientificsymbols.In the
science of language the situationis cardinallydifferent.The symbols exist
immediatelyin language. Instead of the scientist,who extractscertainindices
fromthe externalworld and reshapestheminto symbols,here an exhangeof
symbolsoccursbetweentheparticipantsof a communication.Here therolesof
addresserand addresseeare interchangeable.Hence the task of the scienceof
We are simplytryingto translateintometalanguage
languageis quite different.
For thenatural
thiscode, whichis objectivelygivenin thelanguagecommunity.
scientistsymbolsare a scientifictool, whereasforthe linguisttheyare beyond
that,and firstof all, the trueobject of his research.The physicistNiels Bohr
understoodperspicaciouslythisnaturalrealismof the linguist'sposition.
HavingmentionedNielsBohr,I wouldliketo recallhismethodologicaldictum
essentialboth forphysicsand linguistics.Namely,that,whenan observationis
exactlytherelationbetweentheobserverand
made, itis imperativeto determine
is
theobservedthing.A descriptionthatdoes not complywiththisrequirement
imprecisefromthepointof viewof today'sphysics,as itis fromtoday'slinguistics. It is our taskto clarifythevariouspositionsof scholarsvis-A-vis
language.
The so-calledcrypto-analytical
positionis thepointof viewof an observerwho
does not know the language code, and who could be comparedto a military
to decipheran enemy'sencodedmessage.He triesto
attempting
crypto-analyst,
break the foreigncode througha carefulanalysisof the text. In the studyof
unknownlanguages such devices may obviouslybringfruitfulresults.This,
however,is merelythefirststageof research,and itis byno meanstheonly,but
ratherone of many methodologies,a firstapproximation.Then the observer
attemptsto reach the second, more advanced stage, the stage of a pseudoHe does not anymoremove from
participantin thegivenlanguagecommunity.
thetextto the code, but ratherabsorbsthecode and triesto use the code fora
of themessage.
betterunderstanding
Such is theessentialassumptionof descriptive
linguistics.Butherea difference
not hypostatizethe code, but
We
must
considered.
which
is
rarely
emerges,

This content downloaded from 216.165.126.139 on Fri, 23 Oct 2015 10:18:42 UTC
All use subject to JSTOR Terms and Conditions

SIGN AND SYSTEM OF LANGUAGE

37

ratherenvisageit fromthe point of view of the speech exchange. One must


distinguishsharplybetweentwopositions,of the encoderand thedecoder,in
otherwords:betweentherole of theaddresserand thatof theaddressee.This
seemsto be a banality,but indeed,banalitiesare most oftendisregarded.The
whole mode of observinga messageis cardinallydifferent
forthe two participants in a speech event. The heareris lead throughthe distinctivefeatures,
through the phonemes he recognizes, to the grammaticalform and to
understandingthe meanings. In this process the probabilityfactorplays an
enormousrole. The probabilitiesof thetransitionhelpone to perceivea text,its
phonologyand thenitsgrammar;aftercertainunitsotherunitsfollow,endowed
with higheror lower probabilities,and many are a priorilyexcluded. The
perceiveris endowedwitha subliminalstatisticalset; homonymyis forhim an
essentialprocess.On theotherhand, forthe speaker,theorderof thelanguage
stagesis reversed.His road leads fromthe sentencethroughthe hierarchyof
immediateconstituents,and finallythroughthe morphologicalunits to the
sound formin whichtheyare manifested.Bothordersoccurequallyin language
exchange;theirmutualrelationslie, as Bohrwouldhave said, in theprincipleof
Both languageaspectsexistfortheencoderas well as forthe
complementarity.
decoder,but the directionthat is primaryfor one becomes secondaryfor the
other.For thespeakerqua speakerno homonymy
exists.For example,whenhe
pronouncesin English/sAN/he knowspreciselywhetherhe meanta son or a sun;
whereasthehearermustuse a different
methodof probabilityin orderto solve
thisquestion.Bothattitudes,productionand perception,haveequal claimsto be
describedbythelinguist.It wouldbe a mistaketo reducethistwo-sidedlanguage
realityto merelyone side. Both methodsof descriptionparticipateand have
equal rights.Using only one of the two withoutkeepingin mindwhetherone
representsthe positionof the speakeror the heareris like playingthe role of
Jourdain,who spoke prose withouthaving knownthat it is prose. The real
danger arises when one makes compromises between both positions,
to therulesof each side. For example,ifa linguistselectsencoding
contradictory
as the point of departureof his languagedescriptionand analysis,and hence
forgoesthe use of statisticsand theoryof probability,proceedswitha grammatical analysis of immediate constituents,and observes the primacy of
morphologyoverphonology,thenhe cannot - if he followsa logicaldirection
- excludemeaning.Meaningcan be excludedonly whenone worksfromthe
positionof the decoder,since forhim meaningemergesonly as a conclusion,
whereasforthe speakermeaningis primary.The speakerproceedsde verboad
vocem,whereasthehearerproceedsin theoppositedirection,as SaintAugustine
had alreadystressedin his deliberationson thetheoryof language.
Many thingswillbecomeclearerin liguisticdescriptionsand in thetheoryof
languagewhena clean demarcationis undertakenand theproperattentionpaid
to thedifferent
modesof observationof theencoderand decoder.The modesof
observation,however,are not exhaustedby those two kinds.One should also
take into account the considerableprocess of "recoding":in this case one
in thelightof anotherlanguage,or one styleof speechin
languageis interpreted

This content downloaded from 216.165.126.139 on Fri, 23 Oct 2015 10:18:42 UTC
All use subject to JSTOR Terms and Conditions

38

ROMANJAKOBSON

thelightof anotherone; one code or subcodeis translatedintoanothercode or


is one of themost
subcode. This is a mostilluminating
problem,sincetranslation
essentialand increasingly
importantlinguisticactivities,and themethodologyof
translation,as well as the consistentanalysisof translation,are placed on the
dailyorderof contemporary
pureand applied linguistics.

This content downloaded from 216.165.126.139 on Fri, 23 Oct 2015 10:18:42 UTC
All use subject to JSTOR Terms and Conditions

Potrebbero piacerti anche