ll DOC UME NT RESUME TE 000 674 ED 022 766 By.Chwasky. Noss THE CLMENT SCDE IN :PRESENT DIRECTIONS. Pub Date May 66 Note-10p. . Journal Cit-College English: v27 n8p587-95 May 1966 EDRS Price t4-S025 HC-S0A8 INSTRUCTION. *GRAMIAR.. Descriptors-DEEP STRUCTURE.DESCRIPTIVE LINGUISTICS. ENGLISH THEORY. PERCEPTION. RESEARCH *LANGUAGE RESEARCH. LEARNINGTHEORIES. *LINGUISTICS. *LINGUISTIC STRUCTURE. TRADITIONAL . METHODOLOGY. STRUCTURAL LINGUISTICS, SI.RFACE TRANSFORMATION of Two traditions aredistinguishable in modern linguistictheory: the tradition 'universal grammar' which flourished inthe 17th and 18th centuries,and the tradition of structural or descriptivelinguistics which reached itspeak 15 or 20 years ago. with (1) the relationof deep structure tosurface Universal grammar was concerned (3) forms and to the use and acquisitionof language. (2) the actof perception. and general. Structural linguistics. onthe other hand, has the acquisition of knowledge in study of been particularly valuablefor providing a methodologyfor the recording and data. The linguists oftoday can begin toutilize the methods factual problems which developedlancirstructural linguiststo scientificallyinvestigate the concernpd the-universalgrammarians..We may well witness.then. a synthesis of these langUAge study which i411 allow ourstudents to haVe insightinto ttie two traditions in the mysteries of complexities of the grammar they useunconsciously and its relation to the human intelligence itself.(DL) NS NUMMI 0. IOW. MIKAHON s MOAN *HU * NIKON

TINS OKUNIIII HAS MN MONK* (um AS *WV* NON III P11501 01 OKAIII1A11011 0101611116 II. MINIS Of VIEW 01 OPINIONS 5IA111 HO IN INCISSAIU IMMO ODKIAl OKI OfMOTION POSMON 01 POW. COLLEGEENGLISH AN OFFICIAL ORGAN OFTHE NATIONAL COUNCIL OF TEACHERS OF ENGLISH President Executive SecretaryCollege Section Chairman Mtiam. Caossir JAhtes R. SQUIRE ROBERT M. GORRELL Wilmington, Delaware, Public Schools Universityof Illinois University of Nevada Editor:James E. Miller,jr., University of Chicago,Chicago37,Illinois

Vol. 27 CONTENTS FOR MAY 1966 No. 8






A BACKWARD GLANCE: James E. Miller,Jr. 624 ROUND TABLE: The Copyright Law and T. 4:1 S. Eliot(Oscar Cargill);A Letter 4) to One More Newly-Elected Committee Set Upto Plan and Administer a Course in Freshman Composition(Ken Macrorie);A Note on Culture,or Animation and thejobless (Eric Larsen) 627 C\I REBurrAL: A Comment C\I on Richard Ohmann's "Literatureas Sentences" and Martin Steinmann's "RhetoricalResearch"(A. M. Tibbetts);Pinfeathers 0 for a Ruptured Duck(Raven I. McDavid, Jr.) CI 634 ta DEPARTMENTAL MEMO(Jerome W. Archer,Editor): Master Assistantsat the University of Wisconsin(Edgar 1V. Lacy, William Lenehan,andEdnah S. Thomas); 4. Shift in Composition Sequenceat Fort Hays Kansas State College (W. R. Thompson) 637

NEWS AND IDEAS NO (Louis H. Leiter,Editor);Verse: Man's His Own Wine(James Edmund Magner, fr.) 640 0 Books(Robert E. Knolland 40 Bernice Slote,Editors) 4/ 0 ANNUAL INDEX 661 tki I% COLLEGE ENGLISH

V olume 27 May 1966 Number 8

The Current Scene in Linguistics: Present Directions


THE TITLE OF THIS PAPER may suggest possible, and that it is, to Some extent, something more than can yae provided. It being achieved in current work. Before would be foolhardy to Jttempt to fore-approaching thc problem of synthesis, 1 cast the development of linguistics or any would like to skctch brieflyand, neces- other field, even in general tcrms and insarily,with some oversimplification thc short run. There is no way to antici- what seem to 1 i.e to b thc most signifi- pate ideas and insights that may, at anycant features in these two traditions. time, direct research in new directions or As the indicates, universal gram- reopen traditional problems that had mar was concerned with general features been too difficult or too unclear to -of language structure rather than with vide a fruitful challenge. The most thatparticular idiosyncrasies. Particularly in one can hope to do is to arrive at a clearFrance, universal grammar developed in appraisal of the present situation in lin- part in reaction to an earlier descriptivist guistic research, and an accurate under-tradition which held that the only proper standingofhistoricaltendencies.Ittask for the grammarian was to present would not be realistic to attempt to pro- data, to give a kind of "natural history" ject such tendencies into the future. of language (specifically, of the "culti- Two major traditions can be distin-vated usage" of the court and the best guished in modern linguistic theory: onewriters). In contrast, universal grammar- is the tradition of "universal" or "philo-ians urged that the study 'of language sophical grammar," which flourished inshould be elevated from the level of "nat- the seventeenth and eighteenth centuries;ural history" to that of "natural phi- the second is the tradition of structurallosophy"; hence the term "philosophical or descriptive linguistics, which reachedgrammar", "philosophical" being used, of the high point of its development perhaps course, in essentially the sense of our fifteen or twenty years ago. 1 think thata term "scientific." Grammar should not synthesis of these two major traditions is be merely a record of the data of usage, but, rattier, should offer an ,-xplanation Mr. Chomsky, whose fourth book on linguis- for such data. It should establish general tic theory, Cartesian Linguistics, is now in press, principles, applicable to all and is professor of Modern Languages and Linguis- tics at the Massachusetts Institute of Tech- based ultimately on intrinsic properties nology. Tbis paper was read at tbe NCTEof the mind, Which would explain how convention in November 1P61. language is used and why it has the par-


ticular properties to whidi thedescriptive modelisrather curious. Infact, the grammarian chooses, irrationally, to re-earliest studies of universal grammar,in strict his attcntion. France, wcrc a part of the movement to Universal granmiarians did not content raise th: status of thc vernacular,and arc themselves with merely stating this goal. conccrncd with details ofFrench that In fact, many generations ofscholarsoften do not even have anyLatin proceeded to develop a rich and far- analogue. reaching account of the generalprinci- As to the belief that modern"anthro- ples of language structure, supported bYpological linguistics" has refutedthc as- whatever detailed evidence theycouldsumptions of universal grammar,this is find from thc linguistic materials avail-not only untrue, but,for a rather impor- able to thcm. On thc basis ofthese prin-tant reason, could not be true.Thc rea- ciples, they attempted to explain manyson is that universal grammarmade a particular iacts, and to develop a psycho-sharp distinction between what we 1112V logical theory dealing with certain as-call "deep structure" and "surface struc- is pects of language use,with the produc-ture." The deep structure of a sentence de- tion and comprehensionof sentences. thc abstract underlying form which The tradition of universal grammartermines the meaning of the sentence;it came to anabrupt end in the nineteenth is present in thc mind but notnecessarilY that 1 will discussrepresented directly in the physicalsig- century.. for reasons directly.I. urthertniire, its achievementsnal. The surface structure of a sentence were very rapidlyforgotten, and an in-is the actual organization of thcphysical teresting mythology developed concern-signal into phrases of varying size,into ing its limitations and excesses. It has now words of various categories, withcertain become something of a cliché amongparticles, inflections, arrangement,and so assunipti9n of the linguists that universal grammarsuffered on. The fundamental from the following defects: (1) it wasuniversalgrammarians was thatlan- not concerned withthe sounds of ,guages scarcelyditrer at thc level of deep but only with writing; (2) it wasbased structurewhich , eflects the basic prop- primarily on a 11atin model, and was,inertics of thought andconceptionbut the much some sense"prescriptive"; (3) its assump-that they may vary widely at tions about language structure havebeen less interesting level of surface structure. linguistics refuted by modern "anthropologicallin-But modern anthror 'logical guistics." In addition, manylinguists, does not attempt to dcal withdeep struc- though not all, would hold thatuniversal ture and its relations tosurface structure. surface grammar wasmisguided in principle inRather, its attention is limited to phonetic form of an ut- its attempt to provide explanationsratherstructurcto the organization into units of than mere description of usage.the latterterance and its informa- being all that can be contemplatedby the varying size. Consequently, the direct bear- "sober scientist." tion that it provides has no concerning deep The first two criticisms arequite easy ing on thc hypotheses by thc universal to refute; thethird and fourth are morestructure postulated interesting. Even a cursory glance atthegrammarians. And, in fact, it sccms to me that was athat what information is nowavailable to texts will show that they wcnt too far in major concern of universalgrammarians, us suggests not ofunderlying and that their phonetic theories werenot assuminguniversality structure, but thatthey may have been very differentfrom our own. Nor have I what been able to discover anyconfusion ofmuch too cautious and restrained in speech and writing. The beliefthat uni-they proposed. ofuniversal versal grammar was based on aLatin Thc fourthcriticism

a THE CURRENT SCENE IN LINGUISTICS 589 grammarnamely, that it was misguidedin which the study of language should in seeking explanations in the first placedevelop. I will not discuss. It seems to mc that this The tradition of universal grammar criticism is based on a misunderstandingcame to an end morc than a century ago. of the nature of all rational inquiry. ThereSeveral factors combined to lead to its is particular irony in thc fact that thisdecline. For one thing, the problems criticism should -be advanced with thcposed were beyond the scope of thc avowed intention of making linguisticstechnique and understanding then avail- "scientific." It is hardly open to questionable. The problem of formulating thc that the natural sciences are concernedrules that determine deep structures and precisely with the problem of explainingrelate them to surface structures, and the phenomena, and have little use for ac-deeper problem of determining the gen- curate description that is unrelated toeral abstract characteristics of these rules, problems of explanation. could not be studied with any prccision, I think that we have much to learnand discussion therefore remained at the from a careful study of what was achievedlevel of hints, examples, and vaguely for- bv the universal grammarians of themulated intentions.Inparticular, the seventeenth and eighteenth centuries. Itproblem of rule-governed creativity in seems to me, in fact, that contempo-language simply could not be formulated rary linguistics would do well totakewith sufficient precision to permit re- their concept of language as a point ofsearch to proceed very far. A second departure for current work. Not only doreason for the decline oftraditional lin- they make a fairly clear andwell-guistic theoryliesinthe remarkable founded distinction between deep andsuccesses of Indo-Europeancomparative surface structure, but they also go on tolinguisticsinthe nineteenth century. study the nature of deep structure and toThese achievements appeared to dwarf provide valuable hints and insights con-the accomplishments of universal gram- cerning the rules that relate the abstractmar, and led mans,linguists to scoff at underlying mental structures to surfacethe "metaphysical" and "airy pronounce- form, the rules that we would now callments" of those who were attempting to "grammatical transformations." VVhat isdeal with a much wider range of prob- more, universal grammar developed aslemsand at that particular stage of the part of a general philosophical traditiondevelopment of linguistic theory, were that provided deep and important in-discussing these topics in a highly incon- sights, also largely forgotten, Into the useclusive fashion. Looking back now, we and acquisition a language, and, further-can see quite clearly that the conceptof more, into problems of perceptionandlanguage employed by the Indo-Euro- extremely acquisitionof knowledge ingeneral.pean comparativists was an These insights can be exploited and de-primitive one. It was, however, well- veloped. The idea that the study of lan-suited to the tasks at hand. It is, there- thc frame-fore, not too surprising that this concept guage should proceed within of language, which was then extended work of what we might nowadays calland developed by the structural and "cognitive " is sound. Theredescriptivelinguists of the twentieth is much truth in the traditional view thatcentury, became almost completelydom- language provides the most effectiveinant, and that the older tradition of lin- means for studying the natureand mech-guistic theory was largely swept aside anisms of the human mind, 3nd that onlyand forgotten. This is hardly a unique within this context can we perceive theinstance in intellectual history. larger issues that determine the directions Structural linguistics is a direct out- 590 COLLEGEENGLISH growth of the concepts that emergedgrammar. One real advancehas bccn in inIntlii-European comparative study,universal phoneticsI rcfcr herc particu- which was primarily conc..med withlarly to thc work of Jakobson. Other language as a system of p.onologicalncw and important insightsmight also units that undergo systematic modifica-be cited. But in general, the major con- tion in phonetically determined contexts.tributions of structural linguistics seem Structural linguistics rcintcrprctcd thisto me to be methodologicalrathcr than concept for a fixed stateof a language,substantive. These methodological con- investigated the relations among suchtributions are not limited to a raising of units and thc patterns they form, andthe standards of precision. In a more attempted, with varying success, to ex-subtle way, the idea that language can tend the same kind of analysis to "higherbe studied as a formal system, a notion levels" of linguistic structure. Its funda-which is developed with force and effec- mental assumption is that procedures oftivenessinthe work of Harris and segmentation and classification, appliedHockett, is of particular sirificance. It to data in a systematic way, canisolateis, in fact, this general insight and the and identify all types of elements thattechniques that emerged as it developed function in a particular language alongthat have made it possible, in the last few with the constraints that they obey. Ayears, to approach the traditionalprob- catalogue of these elements, their rela-lems once again. Specifically, it is now tions, and their restrictions of "distribu-possible to study the problem of rule- governed creativity in natural language, tion," would, in most structuralist views, I constituteafullgrammarofthethe problem of constructing 1 language. that explicitly generate deep and sur- Structurallinguistics has very realface structures and express the relations accomplishments to its credit. To me, itbetween them, and the deeper problem seems that its majorachievement is toof determining the universal conditions have provided a factual and a methodo-that limit the form and organization of logical basis that makes it possible torules in the grammar of a human lan- return to the problems thatoccupied theguage. When these problems areclearly traditional universal grammarians withformulated and studied, we are led to a some hope of extending anddeepeningconception of language not unlike that their structure andsuggested in universal grammar. Further- language use. Modern descriptive lin-more, I think that we areled to conclu- guistics has enormously enriched thesions regarding mental processes of very range of factual materialavailable, andmuch the sort that were developed, with has provided entirely new standards ofcare and insight, in therationalist philos- clarity and objectivity. Given this ad-ophy of mind that provided the intellec- vance in precision andobjectivity, it be-tual background for universal grammar. comes possible to return,with new hopeIt is in this sense that I think we can for success, to the problem of construct-look forward to a productive synthesis ing the theory o; a particular languageof the two major traditions of linguistic its grammarand tothestillmoreresearch. ambitious study of the general theory of If this point of view is correct in ! language. On the other hand, it seems toessentials, we can proceed to outline the me that the substantivecontributions toproblems facing the linguist in the fol- the theory of language structure are few,lowing way. He is, first of all, concerned and that, to a-large extent, the conceptsto report data accurately.What is less of modern linguistics constitute a retro-obvious, but nonetheless correct, is that gressionas compared withuniversalthe data will not be of particular interest THE CURRENT SCENE IN LINGUISTICS 591 to him in itself, but rather onlyinsofartheory of linguistic structure provide as it sheds light on the grammar of thevery relevant evidence for anyone con- language from which it is drawn, wherecerned with these matters; to me it seems by the "grammar of a language" I mean quite obvious that it is within this general die theory that deals with the mech-framework that linguistic research finds anisms of sentence construction, which its intellectual justification. establish a sound-meaning relation in this At every level of abstraction, the lin- language. At the next level of study, theguist is concerned with explanation, not linguist is concerned to give a factuallymerely with stating facts in one form or accurate formulation of this grammar,another. He tries to construct a grammar that is, a correct formulation of the ruleswhich explains particular data on the that generate deep and surface structuresbasis of general principles that govern and interrelate them, and the rules thatthe language in question. He is interested give a phonetic interpretation of surfacein explaining these general principles structures and a semantic interpretationthemselves, by showing how they are of deep structures. But, once again, thisderived from stillmore general and correct statement of the grammaticalabstract postulates drawn from universal principles of a language is not primarilygrammar. And he would ukimately have of intcrest in itself, but only insofar asto find a way to account for universal it sheds light on the more gcneral ques-grammar on the basis of still more gen- tion of the nature of language; that is,eral principles of human mental struc- the nature of universal grammar. Theture. Finally, although this goal is too primary interest of a correct grammar isremote to be seriously considcrcd, he that it provides the basis for substantiat-might envision the prospect that the kind ing or refuting a general theory of lin-of evidence he can provide may lead to guistic structure which establishes gen-a physiological explanation for this en- eral principles concerning the form oftire range of phenomena. grammar. Ishouldstressthat whatIhave Continuing one step higher in level ofsketched is a logical, not a temporal order abstraction, a universal grammara gen-of tasks of incrcasing abstractness. For eral theory of linguistic structure thatexample, it is not necessary to delay thc, determinestheform of grammarisstudy of general linguistic theory until primarily of interest for the informationparticular grammars are available for it proviaes concerning innate intellectualmany languages. Quite thc contrary. The structure. Specifically, a general theorystudy of particular grammars will be of this sort itself must provide a hypothe-fruitful only insofar as it is based on a sis concerning innate intellectual struc-precisely articulated theory of linguistic ture of sufficient richness to account forstructure, just as the study of particular the fact that the child acquires a givenfacts is worth undertaking only when it grammar on the basis of the data avail-is guided by some general assumptions able to him. More generally, both aabout the grammar of the language from grammar of a particular language and awhich these observations are drawn. general theory of language are of intcrest primarily because of the in ,ight they All of this is rather abstract. Let mc try provide concerning the nature of mentalto bring the discussion down to earth bY processes, the niechanisnis of perccptionmcntioning a few particular problems, in and production and thc mechanisms bythc grammar of English, that point to the which knowledge is acquired. Therc canneed for explanatory hypotheses of thc be little doubt that both specific theoriessort I have bccn discussing. of particular languages and thc general Consider thc comparative construction 592 COLLEGEENGLISH in English; in particular, such sentencesdetermine what general concept of lin- as: guistic structurc hc employs that leads (1) I have never seen a man tallerthanhim to the conclusion that thc grammar John of English treats (1) and (2) as para- (2) I have never seen a taller man thanphrases but not the superficially similar John pair (3) and (4). This still unknown prin- Sentences (1) and (2), along with innu-ciple of English grammar may lead us to merable others, suggcst that thcrc shoulddiscover thc relevant abstract principle be a rule of English that permits a sen-of linguistic structure. It is this hope, of tence containing aNoun followed by acourse, that motivates the searchfor the ComparativeAdjectivetobetrans-relevant principle of English grammar. formed into the corresponding sentence Innumerable examples can be given of containing the sequence: Comparativethis sort. I will mention just onc more. AdjectiveNoun. This rule would thenConsider the synonymous sentences (5) appear as a special caseof the very gen-and (6): eral rule that forms such Adjective-Noun (5) It would be difficult for him to constructions as "the tall man" from the understand this underlying form "the man who is tall",(6) For him to understand this would be and so on. difficult. But now consider the sentence: Corresponding to (5), we can form rela- (3) I have never seen a man taller than tive clauses and questions such as (7): Mary (7) (i) something which it would be dif- This is perfectly analogous to (1); but ficult for him to understand we cannot use the rule justmentioned to (ii) what would it be difficult for form him to understand? (4) I have never seen a taller man thanBut there is some principle that prevents 1 M ary. the formation of the corresponding con- In fact, the sentence (4) is certainly notstructions of (8), formed in the analogous synonymous with (3), although (2) ap-way from (6): pears to be synonymous with (1).Sen-(8) (i) soinething which for him to un- tence (4) implies that Mary is a man, derstand would be difficult although (3) does not. Clearly either the (ii) what would for him to under- proposed analysis is incorrect, despite the stand be difficult? very considerable support one canfindThe nonsentences of (8) are formed from for it, or there is some specific condition (6) by exactly the same process that in English grammar that explains whyforms the correct sentences of (7) from the rule in question can be used to form(5); namely, pronominalization in the (2) but not (4). In either case, a seriousposition occupied by "this", and a re- explanationislacking; thereis someordering operation. But in the case of principle of English grammar, now un-(6), something blocks the operation of known, for which we must search tothe rules for forming relative clauses and explain these facts. The facts are quiteinterrogatives. Again, the facts are inter- clear. They are of no particular interestesting because they indicate that some in themselves, but if they can bring togeneral principle of English grammar light some general principle of Englishmust be functioning,unconsciously; and, grammar, they will be of realsignificance.at the next level ofabstraction, they Furthermore, we must ask how everyraise the question what general concept speaker of English comes to acquire thisof linguistic structure is used by the per- still unknown principle of English gram-son learning the language to enablehim mar. We must, in other words, try toto acquire the particular principlethat I


explains the between (7) andshould be brought to the student's atten- (8). tion and he should be presented with the Notice that there is nothing particu-case for the various alternatives. But in larly esoteric about these examples. Thethe case of teaching grammar, the issue processes that form comparative, relative,is often confused by a pseudo-problem, andinterrogativeconstructionsarewhich Ithink deserves some further among the simplest and most obvious indiscussion. English grammar. Every normal speaker To facilitate this discussion,let me has mastered these processes at an earlyintroduce some terminology. I will use age. But when we take a really carefulthe term "generative grammar" to refer look, we find much that is mysterious into a theory of language in the sense theseveryelementaryprocessesofdescribed above, that is, a system of rules grammar. that determine the deep and surface VVhatever aspect of a language onestructures of the language in question, studies, problems of this sort abound.the relation between them, the semantic There are very few well-supported an-interpretation of the deep structures and swers, either at the level of particularthe phonetic interpretation of the surface or universal grammar. The Jinguist whostructures. The generative grammar of a is content merely to record and organizelanguage, then, is the system of rules phenomena, and to devise appropriatewhich establishes the relation between terminologies, will never come face tosound and meaning inthis language. face with these problems. They onlySuppose that the teacher is faced with arise when he attempts to construct athe question: which generative grammar precise system of rules that generate deepof English shall I teach? The answer is structures and relate them to correspond-straightforwardinprinciple, however ing surface structures. But this is justdifficult the problem may be to settle in another way of saving that "pure descrip-practice. The answer is, simply: teach tivism" is not fruitful, that progress inthe one that is correct. linguistics,asin any otherfieldof But generally the problem is posed in inquiry, requires that at every stage ofrather different terms. There has been 1 our knowledge and understandint. wegreat deal of discussion of the choice not pursue the search for a deeper explana-between competing generative gram- tory theory. mars, but between a generative grammar I would like to conclude with just aand a "descriptive grammar." A "de- few remarks about two problems that arescriptive grammar" is not a theory of of direct concern to teachers of English.thelanguageinthesensedescribed The first is the problem of which gram-above; it is not, in other words, a system mar to teach, the second, the problemof rules that establishes the sound-mean- why grammar should he taught at all. ing correspondence in the language, inso- If one thinks of a grammar of Englishfar as this can be precisely expressed. as a theory of English structure, thenRather, it is an inventory of elements of the question which grammar to teach isvarious kinds that play a role in the no different in principle from the prob-language. For example,adescriptive lem facing the biologist who has to de-grammar of English might contain an cide which of several competing theories to teach. The answer, in either case, isinventory of phonetic units, of pho- that he should teach the one whichnemes, of , of words, of lexi- appears to be true, given the evidencecal categories, and of phrases or phrase presently available. Where the evidencetypes. Of coursc the inventory of phrases does not justify a clear decision, thisor phrase types cannot be completed 594 COLLEGE ENGLISH since it is infinite, but let us put asideerativeanddescr;ptivegramniars this difficulty. tcrms of a factual assumption about Itis clear, however, that thc choicenature of langgage. Let us suppose I between a generative grammar and aa theory of language will consist n descriptive grammar is not a genuine one.definition of thc notion "grammar," Actually, a descriptive grammar can bewell as definitions of various kinds immediately derived from a generativeunits (e.g., phonological units, trrirp grammar, but not conversely. Given alogical units, etc.). 'hen wc apply sl generative grammar, we can derive thea general theory to data, wc use inventories of elements that appear atdefinitions to find a particular grami various levels. The descriptive grammar,and a particular collection of units. C in the sense just outlined, is simply onesider now two theories of this sort t aspect of the full generative grammar.differ in thc following way. In one, It is an epiphenomenon, derivable fromunits of various kinds arc dcfincd in thc full system of rules and principlespendently of the notion "grammar"; that constitutes the generative grammar.grammar, thcn, is simply the collect The choice, then, is not between twoof thc various kinds of unit. For exam competing grammars,br.t between awe define "," "," c grammar and one particular aspect ofin terms of certain analytic procedu this grammar. To me it seems obviousand define thc "grammar" to bc the how this choice shot:ld be resolved, sincelection of units derived by apply thc particular aspect that is isolated inthese procedures. In thc other thec the descriptive grammar seems to be ofthc situationisreversed. The not little independent importance. Surely the"grammar" is defined independently principles that dctermine the inventory,thc various kinds of unit; the gramma and much else, are more important thana system of such-and-such a kind. -1 the inventory itself. In any event, theunits of various kinds are defined nature of the choice is clear; it is not aterms of thelogicallyprior conc choice between competing systcms, but"grammar." They are whatever appt rather a choicc between the whole andin the grammar at such-and-such a k a part. of functioning. Although I think what I have just said The difference between theset is literally correct, itis still somewhatkinds of theory is quite an import misleading. I have characterized a descrip-one. It is a difference of factual assur tive grammar as one particular aspecttion. The intuition that lies behind of a full generative grammar, but actu-scriptive grammar is that the units ally the concept "descriptive grammar"logically prior to the grammar, wh arose in modern linguistics in a ratheris merely a collection of units. The mill different way. A descriptive grammartion that lies behind the development was itself regarded as a full account ofgenerative grammar is the opposite; i the language. It was, in other words,that thc grammar is logically prior to assumed that the inventory of elementsunits, which are merely the elements t exhausts the grammatical description ofappear at a particular stage in the fu the language. Once we have listed thetioning of grammatical processes. phones, , etc., we have given acan interpret this controversy in term ! full description of grammatical structure.its implications as to the nature of I The grammar is, simply, the collectionguage acquisition. One who accepts of these various inventories. point of view of descriptive gramt This observation suggests a way ofwill expect to b formulating the difference between gen-process of accretion, marked by grad THE CURRENTSCENE IN LINGUISTICS 595 growth in the size ofinventories, theFrammar in theschools_ My impression is generally taught as elements of the inventoriesbeing devel-is that grammar induc-an essentiallyclosed and finished system, oped by some sort of analytic or What tive: procedures. One who acceptstheand in a rather mechanical way. underlying point of view ofgenerativeis taught is a system ofterminology, a for diagramming sen- grammar will expect,rather, that theset of techniques language acquisition must betences, and so on.I do not doubt that process of the student more lik e thatof selecting a particularthis has its function, that hypothesis from a restrictedclass ofmust have a wayof talking about lan- basis ofguage and itsproperties. But it seems to possible hypotheses, on the opportunity is lost when limited data The selectedhypothesis isme that a great it deter-the teaching of grammaris limited in the grammar; once accepted, for stu- mines a system of relations amongele-this way. I think it is important of various sorts.dents to realize how little weknow about ments and inventories relation of There will, of course, begrowth ofthe rules that determine the periph-sound and meaning in English,about inventory, but it will be a rather human lan- eral and "external" matter.Once thethe general properties of heguage, aboutthe matter of how the child has selected a certain grammar, rules that will "know" whatever is predictedbyincredibly complex system of constitutes a grammar is acquired or put this selected hypothesis. Hewill, in other words, know a great deal about sentencesto use. Fewstudents are aware of the been exposed. Thisfact that in their normal,everyday life to which he has never linguis- is, of course, the characteristic factaboutthey are constantly creating new areimmediately human language. ticstructuresthat betweenunderstood, despite their novelty,by I have outlined the difference They two theories of grammarin rather vaguethost zo whom they speak or write. terms. It can bemade quite precise, andare neverbrought to the realization of the question of choice betweenthemhow amazing an accomplishmentthis is, becomes a matter of fact, notdecision.and of how limited is ourcomprehension descriptivistof what makes it possible. Nordo they My own view is that no remarkable theory can be reconciled with theknownacquire any insight into the facts about the nature and use of lan-intricacy of the grammar that they use unconsciously, even insofar as this system guage. This,however, is a matter that of this discussion.isunderstood and can be explicitly goes beyond the scope both To summarize, as the prolc...in is usu-presented. Consequently, they miss ally put, the choice between generativethe challenge and the accomplishments and descriptive grammars is not a gen-of the study of language. This seems to uine one. It is a choice between a systemme a pity,because both are very real. of principles and one, rather marginalPerhaps 25 the study of language returns selection of consequences of theseprin-gradually to die full scope and scaleof ciples. But there is a deeper anduld-its rich tradition, some waywill be mat4 factual question, to beresolvedfound to introduce students to the tan- not by decisk nbut by sharpening the thatlanguagehas withtalizingproblems assumptions and confronting them always posed for those who arepuzzled facts. of human Finally, I would like to say just awordand intrigued by the mysteries about the matter of theteaching ofintelligence.