Transcript

On Explaining LanguageEric H. LennebergScience, New Series, Vol. 164, No. 3880. (May 9, 1969), pp. 635-643.Stable URL:http://links.jstor.org/sici?sici=0036-8075%2819690509%293%3A164%3A3880%3C635%3AOEL%3E2.0.CO%3B2-JScience is currently published by American Association for the Advancement of Science.Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available athttp://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtainedprior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content inthe JSTOR archive only for your personal, non-commercial use.Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained athttp://www.jstor.org/journals/aaas.html.Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printedpage of such transmission.The JSTOR Archive is a trusted digital repository providing for long-term preservation and access to leading academicjournals and scholarly literature from around the world. The Archive is supported by libraries, scholarly societies, publishers,and foundations. It is an initiative of JSTOR, a not-for-profit organization with a mission to help the scholarly community takeadvantage of advances in technology. For more information regarding JSTOR, please contact [email protected]://www.jstor.orgSat Jul 21 17:28:17 20079May1969,Volume164,Number3880 On Explaining Language Thedevelopmentoflanguageinchildrencanbestbe understoodinthecontextofdevelopmentalbiology. Manyexplanationshave beenoffered fornlanyaspectsoflanguage;thereis littleagreement,however,onhowto explainvariousproblemsorevenon what there is to be explained. Ofcourse, explanationsdifferwiththepersonal inclinationsandinterestsoftheinvesti- gator.Myinterestsareinmanasa biologicalspecies,andIbelievethat thestudyoflanguageisrelevantto theseinterests becauselanguagehasthe followingsixcharacteristics.( i ) Itis aformofbehaviorpresentinallcul- turesoftheworld.(ii)I nallcultures itsonsetisagecorrelated.(iii)There isonlyoneacquisitionstrategy-itis thesameforallbabieseverywherein theworld.(iv)Itisbasedintrinsically uponthesameformaloperatingchar- acteristicswhateveritsoutwardform (1) .(v)Throughoutman'srecorded historytheseoperatingcharacteristics havebeenconstant.(vi)Itisaform ofbehaviorthatmaybeimpairedspe- cificallybycircumscribedbrainlesions whichmayleaveothermentaland motorskillsrelativelyunaffected. Anyformofhumanbehaviorthat hasallofthesesixcharacteristicsmay likewisebeassumedtohavearather specificbiologicalfoundation.This,of course,doesnotmeanthatlanguage cannotbestudiedfrom differentpoints ofview;itcan,forexample,beinves- Theauthori?professorofpsychologyand neurobiologyat CornellUniversity,Ithaca,New York. 9MAY1969 Eric H. Lenneberg tigatedforitsculturalorsocialvaria- tions,itscapacitytoreflectindividual differences,oritsapplications.The purposeofthisarticle,however,isto discuss the aspects oflanguage to which biologicalconceptsareappliedmost appropriately(2).Further,myconcern is withthedevelopmentoflanguagein children-notwithitsorigininthe species. Predictability ofLanguage Development Alittleboystartswashinghishands beforedinnernosoonerthanwhenhis parents decide that training in cleanliness shouldbegin.However,childrenbegin tospeaknosoonerandnolaterthan whentheyreachagivenstageofphys- icalmaturation(Table1). Thereare individualvariationsindevelopment, particularlywithrespecttoagecorrela- tion.It isinterestingthatlanguage developnlentcorrelatesbetterwithmo- tordevelopmentthanitdoeswith chronologicalage.Ifwetakethesetwo variables(motorandlanguage develop- ment)andmakeordinalscalesoutof thestagesshowninTable1andthen usethemforacorrelationmatrix.the resultisaremarkablysmalldegreeof scatter. Since motor developmentisone ofthemostimportantindicesofmatu- ration,itis not unreasonableto propose thatlanguagedevelopment,too,is relatedtophysicalgrowthandde-velopment.Thisimpressionisfur-thercorroboratedbyexaminationof retardedchildren.Heretheagecorre-lationisverypoor,whereasthecorre- lationbetweenmotorandlanguage developmentcontinuestobehigh(3). Nevertheless,thereisevidencethatthe statisticalrelationbetweenmotorand language development is notdueto any immediate,causalrelation;peripheral motordisabilitiescanoccurthatdo not delaylanguageacquisition. Justasitispossibletocorrelatethe variablelanguagedevelopmentwiththe variableschronologicalageormotor development,itispossibletorelateit tothephysicalindicationsofbrain maturation,suchasthegrossweightof thebrain,neurodensityinthecerebral cortex,orthe changingweightpropor- tionsofgivensubstancesineithergray orwhitematter.Onalmostallcounts, languagebeginswhensuchmatura-tionalindiceshaveattainedatleast65 percentoftheirmaturevalues,(In-versely,languageacquisitionbecomes moredifficultwhenthephysicalmatu- rationofthebrainiscomplete.)These correlationsdonotprovecausalcon-nections,althoughtheysuggestsome interestingquestionsforfurtherre-search. EffectofCertainVariations inSocialEnvironment Inmostofthestudiesonthistopic the languagedevelopmentofchildrenin orphanagesorsociallydeprivedhouse- holdshasbeencomparedwiththatof childreninso-callednormal,middle-classenvironments.Statisticallysignifi- cantdifferencesareusuallyreported, whichissometimestakenasademon-strationthatlanguagedevelopmentis contingentonspecificlanguagetrain-ing. That certainaspects oftheenviron- mentareabsolutelyessentialforlan- guagedevelopmentistrndeniable,but itisimportanttodistinguishbetween whatthechildrenactually do,and what theycando. Table1. Correlationofmotor Motormilestones Sitsusinghandsforsupport;uni- lateralreaching Stands;walkswhenheldbyone hand Prehensionandreleasefullyde-veloped;gaitpropulsive;creeps downstairsbackward Runs(withfalls);walksstairs withonefootforwardonly Jumpswithbothfeet;standson onefootforIsecond;builds towerofsixcubes Tiptoes3yards(2.7meters); walksstairswithalternating feet;jumps0.9meter Jumpsoverrope;hopsonone foot;walksonline andlanguagedevelopment(3, pp.128-130). Languagemilestones Cooingsoundschangetobabblingbyintroductionofconsonantal sounds Syllabicreduplication:signsofunderstandingsomewords;applies somesoundsregularlytosignifypersonsor objects,thatis,the firstwords Repertoireof3to50wordsnotjoinedinphrases;trainsofsounds andintonationpatternsresemblingdiscourse;goodprogressin understanding MorethanSO words;two-wordphrasesmostcommon;morein-terestinverbalcommunication;nomorebabbling Everydaynewwords;utterancesofthreeandmorewords;seems tounderstandalmosteverythingsaidtohim;stillmanygram-maticaldeviations Vocabularyofsome1000words;about80percentintelligibility; grammarofutterancescloseapproximationtocolloquialadult; syntacicmistakesfewerinvariety,systematic,predictable Languagewellestablished;grammaticalanomaliesrestrictedeither tounusualconstructionsortothemoreliterateaspectsofdis-course Thereisnothingparticularlysur-prisingorrevealinginthedemonstra- tionthatlanguagedeficitsoccurin childrenwhohearnolanguage,very littlelanguage,oronlythediscourseof uneducatedpersons.Butwhatinterests usistheunderlyingcapacityforlan- guage.Thisisnotaspuriousquestion; forinstance,somechildrenhavethe capacityfor language butdo notuseit, eitherbecauseofperipheralhandicaps suchascongenitaldeafnessorbecause ofpsychiatricdisturbancessuchas 8 0 I OI CE S 8 0 childhoodschizophrenia;otherchildren may not speak because they do not have asufficientcapacityforlanguage,on accountofcertainseverelyretarding diseases. Thereisasimpletechniqueforas-certainingthedegree ofdevelopnlentof thecapacityforspeechandlanguage. Insteadofassessingitbymeansofan inventoryofthevocabulary,thegram- maticalcomplexityoftheutterances, theclarityofpronunciation,andthe like,andcomputingascorederived fromseveralsubtestsofthiskind,itis preferabletodescribethechildren's abilityintermsofafewbroadand generaldevelopn~entalstages,suchas thoseshowninTableI.Testswhich are essentiallyinventories ofvocabuiary andsyntacticconstructionsare likelyt o reflectsimplythedeficienciesofthe environment;theyobscurethechild's potentialitiesandcapabilities. Ihaveusedtheschemadescribedto comparethespeechdevelopmentof childreninmanydifferentsocieties, some ofthem much more primitive than our own.In noneofthesestudiescould 1findevidenceofvariationindevelop- mentalrate,despitetheenormousdif- ferencesinsocialenvironment. Ihavealsohadanopportunityto studytheeffectofadramaticallydif- ferentspeechenvironmentuponthe developmentofvocalizationsduringthe first3monthsoflife(4).Itisvery commoninourculturefor congenitally deafindividualstomarryoneanother, creatinghouseholdsinwhichallvocal soundsaredecidedlydifferentfrom thosenormallyheardandinwhichthe soundsofbabiescannotbeattendedto directly.Six deafmothersandtenhear- ingmotherswereasked,duringtheir lastmonthofpregnancy,t oparticipate in our study.The babies werevisitedat homewhen they wereno morethan10 daysoldandwereseenbiweeklythere- afterforatleast3months.Eachvisit consistedof3hours ofobservationand 24hoursofmechanicalrecordingof allsounds madeand heardbythe baby. Datawereanalyzedquantitativelyand qualitatively.Figure1showsthatal- thoughtheenvironmentwasquantita-tively quite different in the experimental SCIENCE.VOL.164VOICES IN BACKGROUND T VAND RADI OCLOSE TO SUBJECTP C , 025 5, Si' 5 , C25-F6 0 rI53+.DELF40t GPLRENTS-COOING rmRHvTHMfic CRYl NC80FUSSI NG70 -NSI6 06 0 -ll; 7 0 53 k NS 50 -4040-- rl3 030 3 0 -2 02 02 0 -1010 10I0 - 0 0 -0) z 0 Ca fl! a U0 >fn I-2: 2 Z * O " ' S $ $ * m ' f ' Z p , z , q* r n l r z z 3 b , f f sP , I I l I , I I l l ! ,I I ! , , , , 0 " o m g " 2 h h $ 0 " 3 $ " 1 3 2 $0 ~ ~ ~ ~AMOUNT OFNOI SE PER DAY (PERCENT OF OCCURRENCES DURING DAY' S SAMPLING PERIOD1Fig.1. Frequencydistributions ofvariousnoises.Thebasiccountingunitisindividual recordingdays. andthecontrolgroups,thefrequency distributionsofvariousbabynoisesdid notdiffersignificantly;asseeninFig. 2, thedevelopmentalhistoriesofcooing noisesarealsoremarkablyalikeinthe twogroups.Figure3demonstratesthat thebabiesofdeafparentstendt ofuss an equal amount, eventhough the hear- ingparentsaremuchmorelikelyt o comet othechildwhenitfusses.Thus theearliestdevelopmentofhuman soundsappearstoberelativelyinde-pendentoftheamount,nature,or timingofthesoundsmadebyparents. Ihaveobservedthistypeofchild-rearingthroughlaterstages,aswell. Thehearingchildrenofdeafparents eventuallylearntwolanguagesand soundsystems:thoseoftheirdeafpar- entsandthoseoftherestofthecom- munity.Insome instances,communica- tionbetweenchildrenandparentsis predominantlybygestures.Innocase haveIfoundanyadverseeffectsupon thelanguagedevelopmentofstandard Englishinthesechildren.Althoughthe mothersmadesoundsdifferentfrom thechildren's,andalthoughthechil-dren'svocalizationshadnosignificant effectuponattainingwhattheywanted duringearlyinfancy,languageinthese childreninvariablybeganattheusual timeandwentthroughthesamestages asisnormallyencountered. Alsoofinterestmaybethefollowing observationsonfairly retardedchildren growingup instateinstitutionsthatare badlyunderstaffed.Duringthedaythe childrenplayinlarge,barerooms,at- tendedbyonlyoneperson,oftenan olderretardatewhoherselflacksa perfectcommandoflanguage.The children'sonlyentertainmentispro-videdbyalargetelevisionset,playing alldayatfullstrength.Althoughmost oftheseretardedchildrenhaveonly primitivebeginningsoflanguage,there are always someamongthem whoman- age,evenundertheseextremelyde-privedcircumstances,topickupan amazingdegreeoflanguageskill.Ap-parentlytheylearnlanguagepartly throughthetelevisionprograms,whose levelisoftenquiteadequateforthem! Fromtheseinstancesweseethat languagecapacityfollowsitsownnat-ural history.The child can availhimself ofthis capacity iftheenvironmentpro- videsaminimumofstimulationand opportunity.Hisengagementinlan-guageactivitycanbelimitedbyhis environmentalcircumstances,butthe underlyingcapacityisnoteasilyar-rested.Impoverishedenvironmentsare notconducivetogoodlanguagedevel- X = DEAF PARENTS0 =CONTROLl o t-AGEREAD:"^^ t he end of the nth week") Fig.2.Eachbaby'sdaywasdividedinto6-minute periods;thepresenceorabsenceof cooingwasnotedforeachperiod;thisyieldedapercentageforeachbaby'sday;days ofall babies wereordered bytheirages,andtheaverage wastaken for alldays ofiden- tical age. Nonaverageddatawere publishedin(4). opment,butgoodlanguagedevelop-tions. Nevertheless,there are indications mentis notcontingent on specifictrain-thatinheritanceisatleastpartiallyre-ingmeasures(5); awidevarietyofsponsible fordeviationsinverbalskills, ratherhaphazardfactorsseemstobeasinthe familial occurrenceofadeficit sufficient.termedcongenitallanguagedisability (2,chapter6).Studies,withcomplete pedigrees,havebeenpublishedonthe EffectofVariationsoccurrenceanddistributionofstutter-inGeneticBackground ing, ofhyperfluencies,ofvoice qualities, andofmanyothertraits,whichcon-Manisanunsatisfactorysubjectforstitute supporting thoughnotconclusive thestudyofgeneticinfluences; wecan-evidencethatinheritance playsarolein notdobreedingexperimentsonhimlanguageacquisition.I n addition t o such andcanuseonlystatisticalcontrols.familystudies,muchresearchhasbeen Practicallyanyevidenceadducediscarriedoutontwins.Particularly susceptiblet oavarietyofinterpreta- notablearethestudiesofLuchsinger, 0 1 0I5I10I15 I2 0 I 25I 3 0 VOICES CLOSE TO SUBJECTPERCENT OCCURRENCES DURING DAYFig.3.Relationbetweentheamountofparents'noisesheardbythebabyandthe amountoffussingnoisesmadebythebaby.Eachsymbolisonebaby'sday;(solid circles)deafparents;(triangles)hearingparents. whoreportedontheconcordanceof dmelopmentalhistoriesandofmany aspectsofspeechandlanguage.Zy-gositywasestablishedinthesecasesby serology(Fig.4). Developmentaldata ofthiskindare,i nmyopinion,of greaterrelevanceto our speculationson geneticbackgroundthanarepedigrees. The nonbiologistfrequentlyandmis- takenlythinks ofgenes as beingdirectly responsible for one property or another; thisleadshimtothefallacy,especially when behavior is concerned, ofdichoto- mizingeverythingasbeingdependent oneithergenesorenvironment.Genes actmerelyonintracellularbiochemical processes, although theseprocesses have indirecteffectsoneventsintheindi- vidual'sdevelopmentalhistory.Many alterationsinstructure andfunction in- directlyattributabletogenesaremore immediatelytheconsequenceofaltera- tionsinthescheduleofdevelopmental events.Therefore,thestudiesontwins areimportantinthattheyshowthat homozygotesreachmilestonesinlan- guagedevelopmentatthesameage,in contrastt oheterozygotes,inwhom divergencesarerelativelycommon.It isalsointerestingthatthenatureof thedeviations-thesymptoms,ifyou wish-are,inthevastmajority,identi- calinhomozygotesbutnotinhetero- zygotes. Suchevidenceindicatesthatman's biologicalheritageendowshimwith sensitivitiesandpropensitiesthatlead tolanguagedevelopmentinchildren, whoare spoken t o (in contrasttochim- panzee infants, whodo notautomatical- ly develop language-eitherreceptive or productive-underidenticaltreatment). Theendowmenthasageneticfounda- tion,butthisisnotto say thatthereare "genesforlanguage,"orthattheen-vironmentisofnoimportance. AttemptsToModify LanguageDevelopment Letusnowconsiderchildrenwho havethecapacityfor languageacquisi- tionbutfailtodevelopitforlackof exposure.Thisisthecasewiththe congenitallydeaf,whoareallowedto growupwithouteitherlanguageor speechuntilschoolage,whensuddenly languageisbroughttotheminvery unnaturalways.Beforethistimethey mayhavehalfadozenwordstheycan utter,read,write,orfinger-spell,but1 haveknownofnoprofoundlydeaf child(in NewEngland, where my inves- tigationswereconducted)withwhom ,ldenttcaltwins,Fraternaltwins, Fig.4.Theonsetofspeechanditssubse- quentdevelopmenttendtobemoreuni- formamongidentical twinsthanfraternal twins. onecouldcommunicatebyuseofthe Englishlanguagebeforeschoolage. Whendeafchildrenenteranoralist school,lipreadingandspeechbecome themajorpreoccupationoftraining. However,in mostchildren theseactivi- tiesremainpoorfor manymoreyears, andinsome,throughoutlife.Their knowledgeoflanguagecomesthrough learningtoreadandwrite.However, teachersintheoraltraditionrestrict expressioninthegraphicmediumon thehypothesisthatitinterfereswith lipreadingandspeechskills.Thus,ex-posuretolanguage(i)comesmuch laterinthesechildren'slivesthanis normal,(ii)isdramaticallyreducedin quantity,(iii)ispresentedthrougha differentmediumandsensorymodality, and(iv)is taught to the children rather asasecondlanguageistaught,instead ofthroughthesimpleimmersioninto aseaoflanguagethatmostchildren experience.Thedeafchildrenareim-mediatelyrequiredto use grammatically correctsentences,andeverymistakeis discussedandexplainedtothem. Theresultsofthisprocedureare interestingbutnotveryencouraging fromtheeducationalpointofview. During the early yearsofschooling, the children'sspontaneouswritingshavea veryunusualpattern;thereislittleevi- dencethattheteachers'instructionin "howtocomposecorrectsentences"is ofanyavail.Yet,carefulanalysisof theircompositionsshowsthatsome subtletiesofEnglishsyntaxthatare usuallynotpartofthegrammartaught inthe schooldo make theirappearance, sometimes quiteearly.Therecanbeno questionthatthe childrendo notsimply imitatewhattheysee;someofthe teachingsfallbythewayside,whereas anumberofaspectsoflanguageare automaticallyabsorbed from the written materialgiventothechildren. Thereareotherinstancesinwhich effortsaremadetochangeachild's languageskills byspecialtraining,asin the mildlyretarded,for example.Many parentsbelievethattheirretardedchild wouldfunctionquite normallyifsome- bodycouldjustteachhimtospeak.At Children'sHospitalinBostonIunder-tookapilotstudyinwhichaspeech therapistsawasmallnumberofchil-drenwithDowne'ssyndrome(mon-golism)for several hours each week,in an efforttospeedup language develop- ment.Later,twograduatestudentsin linguisticsinvestigatedthechildren's phoneticskillsandtriedtoassessthe capacitiesofeachchildforclearer enunciation.Throughouttheseattempts, it wasfoundthatifachildhadasmall repertoire ofsinglewords, itwasalways possibletoteachhimyetanotherword, butifhewasnotjoiningthesewords spontaneouslyintophrases,therewas nothingthatcouldbedonetoinduce himtodoso.Thearticulatoryskills weresomewhatdifferent.It wasoften possibletomakeachildwhohadal- wayshadslurredspeechsayaspecific wordmoreclearly.However,themo-mentthechildreturnedtospontaneous utterances,hewouldfallbacktothe stylethatwasusualforhim.Themost interestingresultswereobtainedwhen theretardedchildrenwererequired simplyto repeatwell-formedsentences. Achildwhohadnotdevelopedtoa stageinwhichheusedcertaingram-maticalrulesspontaneously,whowas stillmissingthesyntacticfoundations andprerequisites,couldnotbetaught torepeatasentencethatwasformed bysuch higherrules.This was trueeven insentencesofveryfewwords.Similar observationshavesincebeenmadeon normalchildren( 6 ) ,withuniformly similarresults;normalchildren,too, canrepeatcorrectlyonlythatwhichis formedbyrulestheyhavealready mastered.Thisisthebestindication thatlanguagedoesnotcomeaboutby simpleimitation,butthatthechild abstractsregularitiesorrelationsfrom thelanguagehehears,whichhethen appliestobuildinguplanguagefor himselfasanapparatusofprinciples. WhatSetsthePace ofLanguageDevelopment? Thereisawidespreadbeliefthatthe developmentoflanguageisdependent onthemotorskillsofthearticulating organs.Somepsychologistsbelievethat speciesotherthanmanfailtodevelop languageonlybecauseofanatomical differencesin their oral structures. How- SCIENCE,VOL.164 ever,wehaveevidencethatthisisnot SO. It is important that weare clearabout theessentialnatureoflanguage.Since myinterestsareinlanguagecapacities, Iamconcernedwiththedevelopment ofthechild'sknowledgeofhowlan-guageworks.Thisisnotthesameas theacquisitionof"thefirstword."The besttestforthepresenceanddevelop- mentofthisknowledgeisthemanner inwhichdiscourseisunderstood.In mostinstances,it istruethatthereisa relationbetweenspeech and understand- ing,butthisrelationisnotanecessary one(7). Byunderstanding,I meansomething quitespecific.Intherealmofphonol- ogy,understandinginvolvesaprocess thatroughlycorrespondstothelin-guists'phonematization(incontrast, forexample,toa"pictographic"un-derstanding:phonematizationresults inseeingsimilaritiesbetweenspeech sounds,whereaspictographicunder-standingwouldtreatawordasanin- divisiblesoundpattern).Intherealm ofsemantics,understandinginvolves seeingthebasisonwhichobjectsare categorized,thusenablingachildto nameanobjectcorrectlythathehas neverseenbefore.(The childdoesnot start out witha hypothesis that "table"is thepropernameofauniqueobjector thatitrefers to allthingsthathavefour appendages.)Intherealmofgrammar, understandinginvolvesthe extractionof relationsbetweenwordclasses;anex-ampleistheunderstandingofpredica- tion.By applicationofthesetests, itcan beshownempiricallythatAuntPaul-ine'sfavoritelapdogdoesnothavea littlelanguageknowledge,but,infact, failsthetestofunderstandingonall counts. Asurvey ofchildrenwitha varietyof handicaps shows that their grasp ofhow languageworksisintimatelyrelatedt o theirgeneralcognitivegrowth,which, inturn,ispartlydependentonphysical maturationandpartlyonopportunities to interactwithastimulus-richenviron- ment.Inmanyretardingdiseases,for example,languagedevelopmentispre- dictedbestbytherateofadvancement inmentalage(usingtestsofnonverbal intelligence). In an investigation ofcon- genitallyblindchildren( a) ,weare againfinding thatmajormilestonesfor languagedevelopmentarehighlycorre- latedwithphysicaldevelopment.A naiveconceptionoflanguagedevelop- mentas an accumulation ofassociations betweenvisualandauditorypatterns wouldbehardputtoexplainthis. BrainCorrelates Inadults,languagefunctionstake placepredominantlyinthelefthemi- sphere. Anumber ofcortical fields have beenrelatedtospecificaspectsoflan- guage.Thedetailsarestillsomewhat controversialandneednotconcernus here.Itiscertain,however,thatpre-centralareasofthefrontallobeare principallyinvolvedintheproduction oflanguage,whereasthepostcentral parietalandsuperiortemporalfields are involvedinsensoryfunctions.These corticalspecializationsarenotpresent atbirth,butbecomeonlygradually establishedduringchildhood, ina proc-ess verysimilar to thatofembryological history;thereisevidenceofdifferen-tiationandregulationoffunction.In theadult,traumatacausinglargeleft- sidedcentralcorticallesionscarrya highlypredictableprognosis;in70per-centofallcases,aphasiaoccurs,andin abouthalfofthese,theconditionis irreversible( Iam basingthese figures on ourexperiencewithpenetratinghead injuriesincurredinwar). Comparabletraumaticlesionsin childhoodhavequitedifferentconse-quences,theprognosisbeingdirectly relatedtotheageatwhichtheinsultis incurred.Lesionsofthelefthemisphere inchildrenunderage2are nomorein- jurioustofuturelanguagedevelopment than are lesionsoftherighthemisphere. Childrenwhosebrainistraumatized aftertheonsetoflanguagebutbefore theageof4usuallyhavetransient aphasias;languageisquicklyreestab-lished,however,iftherighthemisphere remainsintact.Oftenthesechildren regain languagebygoing throughstages oflanguagedevelopmentsimilart o those ofthe 2-year-old, butthey traverse eachstageatgreaterspeed.Lesions incurred before the very early teensalso carryanexcellentprognosis,permanent residuesofsymptomsbeingextremely rare. The prognosisbecomesrapidly worse forlesionsthatoccurafterthisperiod; theyoungmenwhobecomecasualties ofwar have symptorns virtuallyidentical withthoseofstrokepatientsofad-vanced age.Experience withthe surgical removalofanentirecerebralhemi-spherecloselyparallelsthispicture. Thebasisforprognosticatingoperative successis,again,theageatwhichthe diseasehasbeencontractedforwhich theoperationisperformed. Ifa disturbance in the lefthemisphere occursearlyenoughinlife,theright hemisphereremainscompetentfor languagethroughoutlife.Apparently thisprocess iscomparableto regulation, asweknowitfrommorphogenesis.If the disease occurs after acertain critical periodoflife,namely,theearlyteens, thisregulativecapacityislostand languageis interfered withpermanently. Thusthetimeatwhichthehemispher- ectomyisperformedislessimportant thanthetimeofthelesion. CriticalAgeforLanguageAcquisition The mostreasonable interpretationof thispictureofrecoveryfrom aphasiain childhoodisnotthatthereisvicarious functioning, or taking over, bytheright hemispherebecauseofneed,butrather thatlanguagefunctionsare notyetcon- finedtothelefthemisphereduring earlylife.Apparentlybothhemispheres areinvolvedatthebeginning,anda specialization takesplacelater(which is thecharacteristicofdifferentiation), resultingina kindofleft-rightpolariza- tionoffunctions.Therefore,there-coveryfromaphasiaduringpreteen yearsmaypartlyberegardedasarein- statmentofactivitiesthathadnever beenlost. There is evidence thatchildren atthisagearecapableofdeveloping languageinthesamenaturalwayas do veryyoung children. Notonlydo symp- tomssubside,butactivelanguagede-velopmentcontinues to occur.Similarly, we see that healthychildren haveaquite differentpropensityforacquiringfor- eignlanguagesbeforetheearlyteens thanafterthelateteens,theperiodin betweenbeingtransitional.Forthe youngadult,second-languagelearning isanacademicexercise,andthereisa vastvarietyindegreeofproficiency. It rapidly becomesmore andmore diffi- culttoovercometheaccentandinter- feringinfluencesofthemothertongue. Neurologicalmaterialstronglysug-geststhatsomethinghappensinthe brain during the early teensthatchanges thepropensityforlanguageacquisition. Wedonotknowthefactorsinvolved, butitisinterestingthatthecritical periodcoincideswiththetimeatwhich thehumanbrainattainsitsfinalstate ofmaturityintermsofstructure,func- tion,andbiochemistry(electroenceph-alographicpatternsslightlylagbehind, butbecomestabilizedbyabout16 years).Apparentlythematurationof thebrainmarkstheendofregulation and locks certain functions into place. Thereisfurtherevidencethatcor-roboratesthenotionofacritical period forprimarylanguageacquisition,most importantly, thedevelopmentalhistories ofretardedchildren.I t isdangeroust o makesweepinggeneralizationsabout allretardedchildren,becausesomuch dependsonthespecificdiseasethat causestheretardation.Buti fwecon-centrateondiseasesin whichthepath- ologicalconditionisessentiallystation- ary,suchasmicrocephalyVeraor mongolism,itis possiblet o makefairly generalpredictionsaboutlanguage development.I f thechild'smentalde-velopmentalageis2whenheis4 yearsold(thatis,hisI.Q.is50), one maysafely predictthatsomesmallpro- gresswillbemadein language develop- ment.Hewillslowlymovethroughthe usualstagesof infantlanguage,al-thoughtherateof developmentwill graduallyslowdown.Invirtuallyallofthesecases,languagedevelopment comestoacompletestandstillinthe early teens, sothatthese individualsare arrestedinprimitivestagesoflanguage development thatare perpetuatedfor the restoftheirlives. Trainingandmotiva- tionareoflittlehelp. Developmentinthecongenitallydeaf isalsorevealing.When theyfirstenter school,theirlanguageacquisitionis usuallyquitespectacular,considering theenormousoddsagainstthem. How- ever,childrenwhobytheirearlyteens havestillnotmasteredalloftheprin- ciplesthatunderlietheproductionofsentencesappeartoencounteralmost unsurmountabledifficulties inperfect-ingverbalskills. There is also evidence ofthe converse. Childrenwhosuddenlylosetheirhear- ing(usually aconsequenceofmeningi- tis)showverydifferent degreesoflan- guageskill,dependingonwhetherthe diseasestrikesbefore theonsetoflan- guageorafter. I f itoccursbefore they are18monthsold,suchchildrenen-counterdifficulties withlanguagede-velopmentthatarevery much thesame asthoseencounteredby the congenitally deaf.Childrenwholosetheirhearing after they haveacquiredlanguage, how- ever,atage3to4, haveadiffer-entprospect.Theirspeechdeteriorates rapidly; usuallywithinweekstheystop using language, andsofar ithas proved impossibleto maintaintheskillby edu- cational procedures[although newtech- niquesdevelopedi nEnglandandde-scribedbyFry( 9 ) givepromiseofgreatimprovement]. Manysuch children thenlivewithoutlanguageforarela- tivelylongtime, often 2to3years,and when they enter the schools for the deaf, mustbetrainedinthesamewaythat otherdeafchildrenaretrained.How- ever,trainingismuchmoresuccessful, andtheirlanguagehabitsstandout dramaticallyagainstthoseoftheirless fortunatecolleagues.Thereappearst o beadirectrelationbetweenthelength oftimeduringwhichachildhasbeen exposedto languageandtheproficiency seen atthe time ofretraining. BiologicalApproach: Defining LanguageFurther Someinvestigatorsproposethatlan- guageis an artifact-atoolthat man has shapedforhimselftoserveapurpose. Thisassumptioninducestheviewthat languageconsistsof manyindividual traits,eachindependentof theother. However,thepanoramaofobservations presentedabovesuggestsabiological predispositionforthedevelopmentoflanguagethatisanchoredin theoperat- ingcharacteristicsofthehumanbrain ( 10) . Man'scognitiveapparatusap-parentlybecomesalanguagereceiver andtransmitter,providedthegrowing organismisexposedtominimumand haphazardenvironmentalevents. However,thisassumptionleadstoa viewdifferentfrom thatsuggestedby the artifact assumption.Insteadofthinking oflanguageasacollectionofseparate andmutuallyindependenttraits,one comestoseeitasaprofoundlyinte- gratedactivity.Languageistobeun-derstoodasanoperationratherthana staticproductofthemind.Itsmodus operandireflects thatofhumancogni- tion,becauselanguageisanintimate partof cognition.Thusthebiological viewdenies that language is thecause ofcognition,orevenitseffect, sincelan- guageisnotanobject(likeatool) thatexistsapartfromalivinghuman brain. Asbiologists, weareinterested i n the operatingprinciplesof languagebe-causewehopethatthiswillgiveus somecluesabouttheoperatingprin- ciplesof thehumanbrain.Weknow thereisjustonespeciesHonzosnpiens, anditisthereforereasonabletoas-sumethatindividualswhospeak Turkish,English,orBasque(orwho spokeSanskritsomemillenniaago)all have(or had)thesamekindofbrain, that is, a computer with the same operat- ingprinciplesandthesamesensorium. Therefore, inabiologicalinvestigation onemusttry to disregardthe differences betweenthe languages oftheworldand todiscoverthegeneralprinciplesofoperationthatarecommont oallof them. This is not an easy matter; in fact, therearesocialscientists who doubtthe existenceof languageuniversals.As studentsof languagewecannotfail to beimpressedwith theenormousdif- ferencesamonglanguages.Yet every normalchildlearnsthelanguaget o whichheisexposed.Perhapsweare simplyclaiming that common denomina- torsmustexist;canweprovetheir existence?I f wediscoveredatotally isolatedtribewitha languageunknown to anyoutsider, howcouldwefindout whetherthislanguageisgeneratedby a computerthathasthesamebiological characteristicsasdoourbrains,and howcouldweprovethatitsharesthe universal featuresofalllanguages? Asastart,wecouldexchangechil- drenbetweenourtwoculturestodis- coverwhetherthesamelanguagede- velopmentalhistorywouldoccurin thoseexchanged.Ourdatawouldbe grossdevelopmentalstages,correlated with the emergenceofmotor milestones. Abioassayof thiskind(alreadyper- formedmany times, always with positive results)givesonly partoftheanswer. Intheory, one mayalsoadduce more rigorousproofof similarityamong languages. Theconceptionoflanguage universals is difficult to graspintuitively, becausewefinditsohardtotranslate fromonelanguaget oanotherandbe- cause thegrammars appear, on thesur- face, to besodifferent.Butitisentirely possiblethatunderneaththestructural difference thatmakesitsodifficult for theadultspeakerto learna secondlan- guage(particularlyonethati snota cognateof hisown)therearesignifi-cantformalidentities. Virtually everyaspectoflanguageis theexpressionofrelations.Thisistrue of phonology(asstressedbyRoman Jakobsonandhisschool),semantics, andsyntax.Forinstance,inalllan-guagesoftheworldwordslabelaset ofrelationalprinciplesinsteadofbeing labelsof specificobjects.Knowinga wordisneverasimpleassociationbe- tween an objectandan acoustic pattern, butthesuccessfuloperationof those principles,orapplication ofthoserules, thatleadtousingtheword"table"or "house" for objects never before encoun- tered.Thelanguageuniversalinthis instanceisnotthetypeof objectthat comes t o havea word, northeparticu- larrelationsinvolved;theuniversalis thegeneralitythatwordsstandfor relationsinsteadofbeing uniquenames foroneobject. SCIENCE.VOL.164 Further,nolanguagehaseverbeen describedthatdoesnothaveasecond orderofrelationalprinciples,namely, principlesinwhichrelationsarebeing related,thatis,syntaxinwhichrela-tions betweenwordsare beingspecified. Onceagain,theuniversalisnotapar- ticularrelationthatoccursinalllan- guages(thoughthereareseveralsuch relations)butthatalllanguageshave relationsofrelations. Mathematicsmaybeusedas ahighly abstractformofdescription,notof scatteredfactsbutofthedynamicin- terrelations-theoperatingprinciples- foundinnature.Chomskyandhisstu- dentshavedonethis.Theiraimhas beentodevelopalgorithmsforspecific languages, primarilyEnglish,thatmake explicittheseriesofcomputationsthat mayaccountforthestructureofsen-tences.Thefactthattheseattempts haveonlybeenpartiallysuccessfulis irrelevanttotheargumenthere.(Since everynativespeakerofEnglishcan tellawell-formedsentencefromanill- formedone,itisevidentthatsome principlesmustexist;thequestionis merelywhethertheChomskyiteshave discoveredthecorrectones.)Thede- velopmentofalgorithmsisonlyone provinceofmathematics,andinthe eyesofmanymathematiciansarela-tivelylimitedone.Thereisamore excitingprospect;onceweknowsorne- thingaboutthebasicrelationaloper- atingprinciplesunderlyingafew languages,itshouldbepossibleto characterizeformallytheabstractsys- temlanguage as a whole. Ifour assump- tionoftheexistenceofbasic,structural language universals is correct, one ought tobeabletoadducerigorousprooffor theexistenceofhomeomorphismsbe-tweenanynaturallanguages,thatis, anyofthesystemscharacterizedfor-mally.Ifacategorycalculuswerede-velopedforthissortofthing,there wouldbeonelevelofgeneralityon whichacommondenominatorcouldbe found;thismaybedonetrivially(for instancebyusingtheproductofall systems).However,ourpresentknowl- edgeoftherelations,andtherelations ofrelations,foundinthelanguagesso far investigatedindepthencouragesus toexpectasignificantsolution. EnvironmentandMaturation Everythinginlife,includingbehav- iorandlanguage,isinteractionofthe individual with its milieu. Butthe milieu 9 MAY1969 isnotconstant.Theorganismitself helpst oshapeit(thisistrueofcells andorgansasmuchasofanimalsand man).Thus,theorganismanditsen-vironmentisadynamicsystemand, phylogenetically,developedassuch. Thedevelopmentoflanguageinthe childmaybeelucidatedbyapplying toittheconceptualframeworkofde-velopmentalbiology.Maturationmay becharacterizedas asequenceofstates. Ateachstate,thegrowingorganismis capableofacceptingsomespecificin- put;thisitbreaksdownandresynthe- sizesinsuchawaythatitmakesitself developintoanewstate.Thisnew statemakestheorganismsensitiveto newanddifferenttypesofinput,whose acceptancetransformsittoyetafur- therstate,whichopensthewayto still different input,andso on.Thisis called epigenesis.Itisthestoryofembryo-logicaldevelopmentobservableinthe forniationofthebody,aswellasin certainaspectsofbehavior. Atvariousepigeneticstates,theor-ganismmay besusceptible to more than one sortofinput-itmay besusceptible to two ormore distinctkindsorevento aninfinitevarietyofinputs,aslongas theyarewithindeterminedlimits-and thedevelopmentalhistoryvarieswith thenatureoftheinputaccepted.In otherwords,theorganism,duringde- velopment,comes tocrossroads;ifcon- ditionAispresent,itgoesoneway; ifconditionBispresent,itgoesan-other.Wespeakofstateshere,butthis is,ofcourse,anabstraction.Every stageofmaturationisunstable.Itis proneto change intospecific directions, butrequiresatriggerfrom the environ- ment. Whenlanguageacquisitioninthe childisstudiedfromthepointofview ofdevelopmentalbiology,one makes an efforttodescribedevelopmentalstages togetherwiththeirtendenciesfor changeandtheconditionsthatbring aboutthatchange.Ibelievethatthe schemaofphysicalmaturationisap- plicabletothestudyoflanguagedevel- opmentbecausechildrenappeartobe sensitive to successively different aspects ofthelanguageenvironment.Thechild firstreactsonlytointonationpatterns. Withcontinuedexposuretothesepat- terns astheyoccurinagivenlanguage, mechanismsdevelopthatallowhimto processthepatterns,andinmostin-stancestoreproducethem(although thelatterisnotanecessarycondition forfurtherdevelopment).Thischanges himsothathereachesanewstate,a newpotentialforlanguagedevelop-ment.Nowhebecomesawareofcer-tainarticulatoryaspects,canprocess themandpossiblyalso reproduce them, andsoon.Asimilarsequenceofac-ceptance,synthesis,andstateofnew acceptancecanbedemonstratedonthe levelofsemanticsandsyntax. Thattheembryologicalconceptsof differentiation,aswellasofdetermina- tionandregulation,areapplicableto thebrainprocessesassociatedwithlan- guage developmentisbestillustratedby thematerialdiscussedaboveunderthe headings"braincorrelates"and"critical ageforlanguageacquisition."Further- more,thecorrelationbetweenlanguage developmentandothermaturational indicessuggeststhatthereareanatorn- icalandphysiologicalprocesseswhose maturationsets the pacefor bothcogni- tiveandlanguagedevelopment;itisto thesematurationalprocessesthatthe conceptdifferentiationrefers.Weoften transferthemeaningofthewordto theverbalbehavioritself,whichisnot unreasonable,although,strictlyspeak- ing,itisthephysicalcorrelatesonly thatdiffe~entiate. Pseudo-Homologies andNaive"Evolutionizing" Therelationbetweenspeciesises-tablishedonthebasisofstructural, physiological, biochemical,and often be- havioralcorrespondences, calledhomol- ogies.Theidentificationofhomologies frequentlyposesheuristicproblems. Commonsensemaybeverymisleading inthismatter.Unlessthereiscogent evidence that the correspondences noted areduetoacommonphylogeneticori- gin,onemustentertainthepossibility thatresemblancesarespurious(though perhapsduetoconvergence).Inother words,notallcriteriaareequallyreli- ableforthediscoveryoftruehomol- ogies.Thecriteriamustpassthefol- lowingtwotestsiftheyaretoreveal commonbiologicalorigins.(i)They mustbeapplicable to traitsthathavea demonstrable(oratleastconceivable) geneticbasis;and(ii)thetraitsto whichtheyapplymustnothavea sporadicandseeminglyrandomdis-tributionover thetaxa oftheentire ani- malkingdom.Homologiescannotbe establishedbyrelyingonsimilaritythat restsonsuperficialinspection(awhale isnotafish);onlogicalratherthan biologicalaspects(animalsthatmoveat 14milesperhourarenotnecessarily relatedtooneanother);andon anthropocentricimputationofmotives (a squirrel'shoardingofnuts mayhave nothingincommonwithman'spro-visionsfor hisfuture). Comparisonsoflanguagewithani-malcommunicationthatpurportto throwlightontheproblemofitsphy- logeneticoriginsinfringeoneveryone oftheseguidelines.Attemptstowrite generativegrammarsforthelanguage ofthebeesinorder to discoverinwhat respectthatlanguageissimilartoand differentfromman'slanguagefailto passtest(i). Syntaxdoesnothavea geneticbasisanymorethandoarith- meticoralgebra;thesearecalculiused todescriberelations.Itmaybethatthe activitiesorcircumstancesto whichthe calculiareappliedareinsomeway relatedtogeneticallydeterminedca-pacities.However,merelythefactthat the calculus mayormaynotbeapplied obviously doesnotsettle thatissue. Thecommonpracticeofsearching theentireanimalkingdomforcommu- nicationbehaviorthatresemblesman's inoneaspectoranotherfailstest(ii). The factthatsome birdspecies and per- hapstwoorthreecetaceanscanmake noisesthatsoundlikewords,thatsome insectsusediscretesignalswhenthey communicate,orthatreconlbinationof signalshasbeenobservedtooccurin communicationsystemsofadozen totallyunrelatedspeciesarenotsigns ofacommonphylogenyorgenetically basedrelationshipto language.Further- more,thesimilaritiesnotedbetween humanlanguageandanimalconlnluni- cationallrestonsuperficialintuition. Theresemblancesthatexistbetween humanlanguageandthelanguageof the beesand the birdsare spurious.The comparativecriteriaareusuallylogical (12) insteadofbiological; andthevery ideathattheremustbeacommonde- nominatorunderlyingallcommunica-tionsystemsofanimalsandmanis based on an anthropocentric imputation. Everythinginbiologyhasahistory, andsoeverycommunicationsystemis theresultofevolution.Buttraitsor skillsdonothaveanevolutionaryhis- toryoftheirown,thatis,ahistorythat isindependentofthehistoryofthe species.Contemporaryspeciesaredis- continuousgroups(exceptforthosein theprocessofbranching)withdis-continuousconlmunicationbehavior. Therefore,historicalcontinuityneed notleadtocontinuitybetweencontem- porarycommunicationsystems,many ofwhich(includingman's)constitute uniquedevelopments. Anotherrecentpracticeistogive speculativeaccountsofjusthow,why, andwhenhumanlanguagedeveloped. Thisisasomewhatfutileundertaking. Theknowledgethatwehavegained about the mechanismsofevolution does notenableustogivespecificaccounts ofeveryeventofthepast.Paleonto- logicalevidencepointsto thenatureof itsfauna,flora,andclimate.Thepre- cursors ofmodernmanhaveleftforus theirbones,teeth,andprimitivetools. Noneofthesebearsanynecessaryor assuredrelationtoanytypeofcom-municationsystem.Mostspeculations onthenatureofthemostprimitive sounds,onthefirstdiscoveryoftheir usefulness,onthereasonsforthehy- pertrophyofthebrain,ortheconse-quences ofanarrowpelvisareinvain. Wecannolongerreconstructwhatthe selectionpressureswereorinwhator- dertheycame,becauseweknowtoo little thatissecurely establishedbyhard evidence about the ecologicalandsocial conditionsoffossilman.Moreover,we donotevenknowwhatthetargetsof actualselectionwere.Thisisparticu- larlytroublesomebecauseeverygenetic alterationbringsaboutseveralchanges atonce,someofwhichmustbequite incidentaltotheselectiveprocess. SpeciesSpecificities andCognitiveSpecialization Inthe19thcenturyitwasdemon-stratedthatmanisnotinacategory apartfromthatofanimals.Todayit seemstobenecessarytodefendthe view(beforemanypsychologists)that manisnotidenticalwithallotherani- mals-infact,thateveryanimal species isunique,andthatmostofthecom-monalitiesthatexistare,atbest,ho- mologies.Itisfrequentlyclaimedthat theprinciplesofbehavioralfunction are identical-inallvertebrates,for ex- ample-andthatthedifferencesbe-tweenspeciesaredifferencesofmag-nitude,ratherthanquality.Atother times,itisassumedthatcognitivefunc- tionsarealikeintwospeciesexcept thatoneofthetwomayhaveaddition- allyacquiredacapacityforaspecific activity.Ifindfaultwithbothviews. Sincebehavioralcapacities(I prefer thetermcognition)aretheproductof brainfunction,mypointcanwellbe illustratedbyconsideringsomeaspects ofbrainevolution.Everymammalian specieshasananatomicallydistinct brain.Homologiesarecommon,but innovationscanalsobedemonstrated. Whenman'sbrainis compared withthe brainofotherprimates,extensivecor-respondencescanbefound,butthere aremajorproblemswhenitcomes totheidentificationofhomologies. Dramaticdifferencesexistnotonlyin size but also in details ofthe developmen- talhistories;togetherwithdifferences in cerebrocorticalhistology,topography, andextent,thereare differencesinsub- corticalfiber-connections,aspointed outbyGeschwind(13)mostrecently andbyothers beforehim.The problem is, whatdo wemake oftheinnovations? Is is possiblethateachinnovation(usu- allyaninnovationisnotaclear-cut anatomicalentity)islikeanindepen-dentcomponentthatissimplyaddedto thecomponentscommontoallthe moreold-fashionedbrains?Andifso, isitlikelythatthenewcomponentis simplyaddingaroutinetothecompu- tationalfacilitiesalreadyavailable? Bothpresumptionsarenaive.Abrain isanintegratedorgan,andcognition resultsfromtheintegratedoperation ofallitstissuesandsuborgans.Man's brainisnotachimpanzee'sbrainplus added"associationfacilities."Itsfunc- tionshaveundergonereintegrationat thesamepaceasitsevolutionaryde- velopments. TheIdenticalargumentappiiesto cognitivefunctions.Cognitionisnot madeupofisolatedprocessessuchas perception,storing,andretrieval.Ani- malsdo notallhaveanidenticalmem- orymechanismexceptthatsomehave alargerstoragecapacity.Asthestruc- tureofmostproteins,themorphology ofmostcells,andthegrossanatomyof mostanimalsshowcertainspecies specificities(as dodetailsofbehavioral repertoires),SOwemayexpectthat cognition,too,inallofitsaspects,has itsspeciesspecificities.Myassumption, therefore,isthatman'scognitionisnot essentiallythatofeveryotherprimate with merelythe additionofthecapacity forlanguage;instead,Iproposethat hisentirecognitivefunction,ofwhich hiscapacityforlanguageisanintegral part,isspecies-specific.Irepeatonce morethatImakethisassumptionnot becauseIthinkmanisinacategory allofhisown,butbecauseeveryani- malspeciesmustbeassumedtohave cognitivespecificities. Conclusion Thehumanbrainisabiochemical machine;itcomputestherelationsex-pressedinsentencesandtheircompo- SCIENCE,VOL.164 nents. Ithas a print-out consisting ofacoustic patterns that are capable ofsimilar relational computation by ma-chines of the same constitution usingthe same program. Linguists, biologists,and psychologists have all discussedcertain aspects of the machine.Linguists, particularly those develop-ing generative grammar, aim at a for-mal description of themachine's behav-ior; they search mathematics for acalculus to describe it adequately. Dif-ferent calculations are matched againstthe behavior to test their descriptiveadequacy, This is an empirical proce-dure. The raw data are the way aspeaker of a language understands col-lections of words or the relationshipshesees.A totally adequate calculus hasnot yet been discovered. Onceavailable,itwill merely describe, in formalterms,the process of relational interpretationin the realm of verbal behavior. It willdescribe a set of operations; however,it will not make any claims of isomor-phism between the formal operationsand the biological operations theydescribe.Biologists try to understand the na-ture, growth, and function of the ma-chine (the human brain) itself. Theymake little inroads here and there, andgenerallyplaycatch-as-catch-can; every-thing about the machine intereststhem(including the descriptions furnishedbylinguists).Traditionally, learning theory hasbeen involved neither in a specific de-scription of this particular machine'sbehavior norinitsphysical constitution.Its concern has been with the use ofthe machine: What makes it go? Canonemake it operate moreorlessoften?What purposes does it serve?Answers provided by each of theseinquiries into language are not intrinsi-cally anatagonistic. as has often beenclaimed. It is only certain overgeneral-izations that come into conflict. Thisisespecially sowhenclaimsaremade thatany one of these approaches providesanswerst oall the questions that matter.DrugSafety:ExperimentalProgramsProblems and solutions of the past 10years areOn 6 November 1958 3. Lehman,chief of the division of pharmacologyof the Food and Drug Administration(FDA), addressed the research and de-velopment section meeting of the Phar-maceutical Manufacturers Associationat Sea Island, Georgia. Although Leh-man's views on the subject were wellknown, through his work with his col-leagues at FDA (I), rumor had it thatnew and far-reaching official rules fortesting drug toxicity were about to beproclaimed. For those who had fearedthe introduction of minimum stan-dards, thespokesmanof FDAprovidedno cause for immediate concern. Al-though he did pronounce certain rulesfor the toxicologic evaluation of ex-9 MAY 1969critically reviewed.Gcrhard Zbindenperimental dnigs he made it clear thatthese were only meant asflexible guide-lines (2).An abstractof Lehman's 1958talk was published in a journal withlimited distribution. The concept be-came generally known after Lehmanspoke at a joint American MedicalAssociation, Society of ToxicologySymposium on 17 June 1963, whencopies of hisprojected slideswere madeavailable and gained wide distribution.To this day, Lehman's unofficial ruleshave decidedly shaped the industry'sand FDA's approach to toxicity test-ing.The 1958-1963 guidelines recom-mended various types of experiments:short-term studies in which the acuteReferences and Notes1. E.H. Lenneberg, in The Slruct~rre of Langlrrrge, Readings in the Philosophy of Lrmgrtage, J. A. Fodor and J. J. Katz, Eds. (Prentice-Hall, Englewood Cliffs, N.J., 1964). 2. For complete treatment, seeE. H. Lenneberg, Biological Forindafions of Lcmgztage (Wiley, New York, 1967). 3. E. H. Lenneberg, I. A. Nichols, E. F.Rosenberger, i n Disorders of Co~nt~trinicutiorzD. Riocll, Ed. (Research Publications ofAssociation for Research i n Nervous andMental Disorders, New York, 1964), vol. 42.4. E. H. Lenneberg, F. G. Rebelsky, I. A. Nichols, Hum, Develop. 8, 23 (1965). 5. R. Brown, C. Cazden, U. Bellugi, in The 1967 Minnesota Symposiunz on Child Psy- chology. J .P . Hill, Ed. (Univ. of Minne- iota Press, Minneapolis, in press). 6. D. Slobin, personal communciation.7. E. H.Lenneberg, J. Ahnornz. Soc. Psychol. 65, 419 (1962). 8. -.S. Fraiberg, N. Stein, research in progresq. 9. D.I).Fly, in The Genesis of Langiiage: A Ps~~cholirrauistie A ~ ~ r o n c h . Smith and G. F.A: ~i l l r r ' ;Eds. (MITPres;, Cambridge, 1966).10.For dctails, see E. H. Lenneberg, Perceplionand t uvg~rage,in preparation.11. N. Chomsky, "The formal nature of lan-guage" (in 2, appendix A).12. See, for instance, C. F. Hockett, in AnimalComnitinication. W. E. Lanyon and W. N.Tavclga, Eds. (American Institute of Biolcgi-cal Sciences, Washington, D.C., 1960); and i nSci. Anrer. 203, 89 (1960).13. N. Ceschwind, Brain 88, 237, 585 (1965).14.I thank H. Levin and M. Seligman forcomments and criticisms.toxicity was to be evaluated by singleadministrationof the drugs to fourani-mal species and two sets of long-termtests in which the substances were tobe given repeatedly. Two animalspecies were suggested for the long-term studies and three dosages were tobe tested. The duration of the experi-ments was 2 weeks to 1 month for thesubacute and up to 6 months for thechronic toxicity tests, with the optionfor an extension up to 2 years. Drugswere to be administered by the sameroutes as anticipated in man. Clinicalte\ts included hemograms, coagulationtests, limited tests on liver and kidneyfunction, and determinations of bloodsugar. Gross and microscopic examil~a-tions were confined to major organs inshort-term studies or were to be donein considerable detail in the longer ex-periments. No specific recommenda-tions were made for the number oftreated animals and controls, the fre-quency of laboratory tests, and theper-centage of animals included in the Iab-oratory studies. Less extensive proce-dures were suggested for the testing ofdrugs administered by inhalation andby the dermal, ophthalmic, vaginal,and rectal routes.Lehman's guidelines became rapidlyThe author is visiting research fellow at theDepartment of Med,cine. University of Cam-htidpe, Cambridge, England.643


Top Related