Hjelmslev lecture, [1961] 052-0030

'T -O > H je I tns S ev Lecture°1 -31 -o I Ladies and Gent Ismers g The other day I endeavored to define my topic more precisely, ass be i ng gi ossemat i ds and other linguistic theories or gS ossemat ics as compared to other linguistic theories« j gave a brief survey of the existing theories with which glossematics has to be compared. I gave that, as you will recall, chiefly in the form,of a selected bibliography, leaving to later to point some of the essential characterist*cs of these theories as compared to q i ossernat i cs» What I ment i on eel, apart from an ear I y Dan * sh contr 5 but i on by H. G« Vi beI, which was deemed to remain {eripheraI, was mainly the traditional views, ultimately dating from antiquity, and next the Sauss-urtern doctrine represented mainly by Saussure himself, end B by Sechehaye« Next, S mentioned the two important contr5butions of Edward Sapir and of Leonard Bloomfield and at last, more recant theories like those of Harris, Hockett, Pike, Kurylowicx--that brief summary was noth?rag but a sketch« In a fairly complet© survey, a good many other names would have had to be. added« But there is one very ?i-sportant structural theory which could not be left out, even in a brief survey, and which, I realize that I did not mention« That is, the so-called phonological theory advanced by the linguistic circle of Prague« I must bag you to understand, ladies and gentlemen, that, this omission was nothing but a mere inadvertency on my part« There was no snadvartenty omission, actually, in my notes« But 1 realized after my lecture that my eye must have slipped on that point,, Well, may 1 make up briefly for this neglect, now, and tell first that the program of the linguistic circle of Prague was originally set forth before the international public at the First International Congress




of Linguistics held in the Hague in 1928, and so, chronologieafly, comes in between Sapir's book from 1921 and Bloomfield's from 1933. Bu'-c apart trom the program and the theses advanced as ©ar!y as ?958 and which played an important part in the international debate of linguistic theory in the early thirties, the so-called phono logical Prague theory has mainly beers expounded at some length and largely deepened by Karl Bffler in his book from 5934, a book in German, Sprach- Xfrgpt**® and we will have to add also one of his papers in the journal Kantstudien volume 38 (it's a little earlier than the book). St's ' r'om title of this paper Die Axiomatik der Sprachwissenschaften ■c.nai sounds a bit odd, perhaps we 'could not render that into English because "linguistics" is already a plural. So this Is a double p-ura? in a wayu Iben, of course, we have to mention the chief work W!:"‘!,n this theory, that by M* "froubetakoy Qrundxuge der Phono!ogle

w,ui published as the seventh volume of the Travaux du Circle Linguistique oe Pr>a9uc' *n *939«. I here is, by the way, for those who would prefer French to German, a French translation available with the title .»':r.L-..-?c.l.£gs....dg.. Phono i og t e • St ts a somewhat revised edition, by the way. So one had better parts in the French edition, The latest greater work, emanating from the Prague doctrine although highly miuenced by other theories and above all by modern communication theory, is the work of Roman Jakobsen, Fant and Halle Prel itntnarSea ZlL.ASSSSLASSiSælSLf 1952. Weil, in my last lecture, l also gave you ynat \ calth* biography of glossomatics including a short bibliography :f that t?»eory. For later references in my presant lectures ! think ’■r wou I d be advisable to add a volume of my own which l may not have merithoned in the first lecture, the title of the volume is "Esseis



I i ngu 5 st iques" and it was published in s959. It is not as the title seems to indicate entirely written in French,, It is a collection of papers in English and in French« A restricted number of copies of that book will be made available to the members of this group in a few days, I hope. I should like to refer particularly now generally to a paper included in that book, pp„ 27 to 35, with the title "Structural Analysis of Language" dating from 1948 and written in English. This paper would perhaps I think bo the best to refer to for the general principles, the very general principles, of the gloss&raatic theory. After these prelims nar»es, { started at the end of my first lecture, to enter upon some essential points within gSossematics, points which are crucial, particularly for the comparison with other theories. The total number of such crucial points«-the total number of these essential points which ! shail have to go through in these lectures will ba nine. IFirst, the j ytical procedure which you will recall from the first lecture, second the theory Oi function, three--commutatI on and substitution. \| think you wiIS also recall these terms from the first lecture so s will give them here in abbreviat i Qfv—commut at ion and subst itut ion * and the theory of two planes of the language, content ond,i .oxprosgjon •rønrowv.*«> (o*. - tMgnoewøOMI yr***i<c.uarv. i..«#*« v ,4 ,»«„*» Number 4 will be form am substance. ■?rrc? than nunbops 5 through 9 will be still more general points, in a way more general than strictly linguistic points. They will be more or less semiotic points. The requ*r©merits of an exhaustive description would be the 5th point; ihe sixth synchrosS c and genet ic ISncuistics; number 7, typology or ganers s tvpo l o&i cs! calculus; no. 8—dsnotat ion and connotation, and fspally, at long lest, no. 9—meta semiotics# I realise that some


of these words would be only words to some of you but I am going to explain them later« It was just to give you a general survey of the points which I shall go through« In my lecture today end in the first part of my following lecture on Thursday next I shall give a brief account of all these nine points« Three of them will afterwards be dealt with more fully and more in detail® Namely, no« 2function;no« 3~ "“commutation, substitution, content and expression; and no® 4“"“form and substance® No« 2 will be dealt with according to my schedule in the second part of the third lecture and as far as correIat Ions are concerned, correlations which are one kind of function will be dealt with in the fifth lecture« Commutation and substitution and the theory of two or several planes ordinarily two--content and expression is what will mainly be dealt with in the fourth lecture and form and substance in the sixth« The first of these nine points, the analytical procedure has already been dealt with in my first lecture so that little needs to be added to it® ft is the most fundamental one, and the one that at the outset, is most character i stic of gtoss4<mat ics, S should think* Glossematics shares some aspects of this first point with some other theories, first and foremost in that glosseraatics recognises the requirement of accounting for what ia known as immediate constituents since the paper published by Ruion S« Wells in 1947, in the 23rd volume of the journal Language« On the other hand, glossematics 'diiffiRr-s fi-om most other theories by the fact that it does not recognise syntax, even in a revised form, as a separate brooch of linguistics, but simply as an interlude, as an intermediate stage, through which the

total analysis of the entire corpus must pass on its way from units


o o

of max i roa i extent (Sarger than one complex clause) down to glossernes or minimal elements,, Thus the stages which, according to conventional use would belong to syntax, are Imbedded in the whole analysis, and have no charactersstics of their own, I mentioned in my first lecture, as an example, the relation between subordinate clause and principal clause which ! defined as what we call selection.. A compound clause may consist of a principal clause only to which you may add one or more subordinate clauses, whereas a clause of the first degree could not consist of the subordinate clause only. So we call this a selection and say that tho subordinate clause selects the principal clause, or, if you would prefer that terminology, the subordinate clause presupposes the principal clause. This case, of course, is one of the cases which conventional linguistics would class under the heading, "syntax." But the relation between subordinate clause and principal clause is exactly in the same relation as the other one ? mentioned the other day"*-that between the consonant unit and the vowel unit of a syllabic theme where we are faced with units much smaller than the so-called syntactical units. And the same will hoid good tf we take units larger than so-called syntactical ones. Say, for instance, the relation in the content between something which 5s called "third volume" and something which is called "second volume," •hese units are iarger than what is ordinarily recognized as syntactical units but the relation is the same« If you say that a book »3 a second volume, this doesn t imply that there w?ll be a third volume whereas tne reverse is true. There would not be a third volume without presupposing a second volume. Or fake the case of the relation of re New Testament and the Jewish Bible, which is also a relation in


O * I the content0 The New Testament presupposes the Jewish Bible« An analysis of the content of these large units would have to recognize "hat* to account for that, whereas the reverse is not true or, anyhow, it is not true in the same way* AH these are so many instances of presupposition, within a unit' of Sarger extent and lower degreeD In the relative enumeration we have labeled it as a unit of the first degree,, The clause of the first degree which is divided into or analyzed into secondary and principal clause« The syllabic theme, in the same way, to be labeled as a unit of the first degree in this relative sense« And so a 130 here we might label some units of a larger extent and consequently of a lower degree. So all these cases are similar. The only difference is in the stage or the analysis--in the degree to which we, in each case, have to assign the operation in question within the higher hierarchy. The so-called syntactical relations have nothing specific about them, but they are just normal instances of a general principle commanding all linguistic analysis, on any scale whatsoever. There are, of course, other general principles than this one« Or one might say there are ctner possible applications of the same general principles, Iuat is there are other possible relations than just the ones we nave just picked out here, and which I have labeled selection and symbolized by an arrow-one part presupposing the other--but not conversely a unilateral presupposition. One other possible relation its solidarity« Solidarity is not the only other possible relation but it js one. in the case of solidarity we will find one part presupposing the other and conversely« The parts presuppose each other« There is mutual or bilateral proaupposi t I on between them



? and I propose to symbolize this by this symbol u Thus when ! stated as a case of selection the relation between third and second volume of a work, perhaps it may have occurrrd to you that i expressly said the Jewish Bible and >tofc the Old ics*.ani6fit<i And perhaps you would i ike to know why I did sOo Well, there is an easy answer to that* The reason is obvious, S thSnk„ As soon as you say the Old Testament, this presupposes a New one so that in this case the presupposition is bilateral, or mutual, between the New Testament and the Olds The Old presupposing the New and the New presupposing the 01d, There could be no New Testament without an Old Testament~-that is true,, 1ou couldn't call a work an Old Testament without presupposing there be a New Testament« It is cleaa'iy to b© 3cer> from this example that what we are aiming at is an analysis of linguistic units, two texts, -»this text and the other one, two texts with a definite name assigned to each of them« Then it is, similarly, a solidarity between second volume and first volume since, t S* a volume is referred to as a second, you presuppose a first one« And if a volume Ss referred to as a first, this presupposes a second o!ume« You ws t f object now, perhaps, ladies and gentlemen, that the supposed second volume may be non existent« A too optimistic i; p I ter may publish a book and call it first volume and the second olume may perhaps never appear. This does not invalidate in the ast the linguistic truth that by the mere fact of calling something .jiLier I, one presupposes there be number two~-> whether in existence .just supposed, I shall come back to the validating of such statements in the second part of my next lecture dealing with function



-a general, under the heading of cat a 1 ysjis, which is the g I osssmat i c technical term for such an operation, (it's a kind of interpolation, of course, but a special kind of interpoJat ion) where, for functional reasons, you interpolate units or, in most cases a category of possible units* If you can't put your finger on the second volume, you would not know what it is like and so it would be safer to state that what you presuppose is a category of possible second volumes of which one may possibly appear* Through the analytical procedure, then, will be shared gradually from the largest to the smallest units through as many stages as possible depending possibly on relations between the parts resulting from the analysis, these parts being in their turn defined by the relations« Thus, the analysis Is a functional one., You will recall from my first lecture that functions are relations and correlations* Some functions are relations, others are cornel at ions* We call them relations if they take place between units within the process, as ! said the other day« Well, when we speak of something more or less similar to a linguistic: object, what we would call a semiotic, then, the process would rightly be called a syntagmet»c„ And when we are talking of a linguistic object, in the ordinary linguistic sense, we would cal I it a text, So, for a linguist, relations, in our sense of that word, would mean functions within the text, within chains in units of the text,, And correlations are functions within the system, functions contracted by members of parts of the system* Well, a semiotic system may be labeled a paradigmatic end a linguistic system may be labeled a language« A language is not a text, a language Is the system


underlying the text and on which the text is bui!t0 So correlations are, to a linguist, functions between the parts of the language, within the language and, more especially, within paradigms and categories or the language system. One kind of relations, though by far not the only one, is commutation and its counterpart, substitution—»which was the third point in my survey. At each stage of ths analysis one has to account for relations between units or between categories of units and correlations between members of paradigms including their mutual commutations, and substitutions, ihis analysis is followed, as I toid you, by a synthesis and a description or what we cal I the substance in which the fortn is mani festede The form is defined by the functions, the net of functions, in which the units under observation take part. May i point out here that a linguistic theory, to roy mind, and the glossematic theory as I'm presenting it to you, does not involve a practical discovery procedureu Such a practical discovery procedure does not form part of the theory although it presupposes the theoryo The theory gives general rules from which the practical procedure can oe logically deduced,, The actual practical procedure of the language describer must always be fine of trial and error« ; the rules of the procedure given by the theory are designed to serve the purpose of a procedure of control, rather than prescrSptions ;.r the procedure to follow —» say, during actual field work. I'm happy to see that this seems also to bo the view held by Zeiig Harris ths introduction to his book Methods in Structural Linguistic Or, the other hand, it would be too technical for those few lectures ■T; presenting to you here to go through even the whole general



procedure of control prescribed in detail by glossemat3cs» It would have been very nice to do that but it would require much more time than the time which is at my disposal here,, For practical purposes, the most interesting and the most important aspect of the analytical procedure 5s the commutation test« That's to say the test involved in my point number three, ±h® commutation test which is carried out within each particular paradigm, I think « underlined that, the last time, Glossematics does not share the methods of Prague phonology--Troubetskoy and his followers including Roman jakobson-“Qr tho.se of Bloomfield, and of various modern phonemicists according to which an inventory of elements, (say, an inventory of so-called phonemes) cart be set up first before asking the question of their occurrence within the chain, a question which is then discussed after setting dp the inventory of elements under the name of distribut ion« Ladies and Gentlemen, there is no such thing as "distribution«" The first thing we arrive at are paradigms defined by their relations to other paradigms. You might symbolize these paradigms, (i think most conveniently) you can symbolize them by vertical columns like this« (The Sines between each column)« Between each of these paradigms there is some kind of mutual relation« And each paradigm consists, of course,. of various members between which there is (within each of the paradigms) a commutation or a substitution« On condition that we constantly bear in mind these paradigms and in consequence their members, they are defined by relations, thus defined functionally and not by position the physical position of the substance unit by which a member of a paradigm may



* m>

be man i festede say, før instance, a souruio On this condition we may, perhaps, most easily convey the relevant idea by roughly and provisionally calling the paradigms places within the chain,. Here is a chain« We symbolize It ordinarily in this way« (Of course, this is completely arbitrary the chain going from left to right« This is simply according to the Roman script and nothing else«) Well, then you may cal! these paradigms, "paradigm no« l, paradigm no0 2,"«—two places within the chain« And each member of each paradigm can fill one of these pigeonholes, one of these places in the chain« Places (we can cal? them places, I mean) within the chain, places within the unit subject analysis« The members of the paradigms would then be the units which can take one end the same place« Imagine that we have, for instance, to go (l try to take English examples and you'll excuse me for writing a long "o"'-«, to some extent l think it's according to American pronunciation, not a diphthong pronunciation)« But, anyhow both would be good since the relation which interests us is that between a consonant unit and a vowel unit and the diphthong is actually a vowel unit. Well, £5© (and we can have a commutation between "g" and "I") low, so, show, sew, no, row, doe, bow« So any of these members of this paradigm, here, the vertical column can fill out one place within this unit« We can take low out, then we can add, still, a paradigm to it, say, the paradigm consisting of "s." So we obtain slow, flow. M&H# Blow* and perhaps more« And it's the same story again*--various members of this paradigm no« l taking up this place, whereas others may simply take up the second place. !n th© examples I've been giving you, "s,b, and g" are members of both paradigms. We had "so.



bow, go and we had stow, blow,, glow,, At though this is not direct i y relevant for the moment, | might add that the various paradigms are compared (must be compared) with each other and categories must be extracted for them* So wo .state there is a category consisting of s, b, g„ »»because of this double function I pointed out, and there is another category consisting of the other consonants we found here*'’!, t and r„ The first category (3 b g) can fill two places (member of two different paradigms,the second one can only fill one of the places w© have considered* This is,of course, just a small, very provisional, sketch to give a hint as to the procedure* Now within each paradigm the commutation test has to be carried out«, Within each paradigm and consequently within each category, the commutation test has to be carried out«, The commutation test is the necessary instrument for detecting the elements considered as members of paradigms and categories and the inventory of elements cannot be set up in any other way. The complete inventory of elements will result ultimately from the whole analytical procedure and then there will be no distribution* On the contrary, the task usually assigned to the distribution procedure has already been fulfiI led by labeling each element according to its possible places« Labeling it through the successive approach in analytical procedure according to its possible places (that is to say, of course, according to its occurence in categories and paradigms) i„e. to it« functions« So when we ultimately set up what we would call the complete inventory of elements, then the so-called distribution i os already taken place long ago and really it is not a distribution, ■L is, logically, just the opposite of a distribution* We may recall,




• ► 8 3

by the way, that the notion and method of so-called distribution was submitted to severe criticism at the Eighth 1nternationai Congress of Linguists held in Oslo in 1957, criticism mostly by Danish linguists, it's true0 These contributions are published in the Proceed!ngs of that congress, the Eighth International Congress of Linguists held ir. Oslo in 5957o And ! would not fail to call your attention to this, although 5 regret I'm not in a position today to give you the exact page reference to that volume. Now For the purpose of th© commutation test we would consider it necessary to recognize two planes of the Ianguage-»this terminology is in accordance with Ferdinand de Saussure~~two planes within a languages the plane of expression and the plane of content. That's how we would like to label them, Saussure's names are different-- what I her© call the expression is in Saussure's Cours "1e sign!fjant," What is here called the content is in Saussure's Cours, "!e sign if5e," 1 might warn you perhaps, talking of this terminological business, that If you read my Prolegomena you may find a complication which should not detain us here but 1 should just point it out on p, 37 of the Podegornena, planes are considered from two points of view«- from the point of view of their mutual relations, they are called lines, and from the point of view of their mutual correlations, they are called signs. Mow, between these two planes-“*the expression pilene and the content plane--which according to glossemattes are both comprised in the language, between these two planes there is presuppositIon, (that is to say) a solidarity. There would be no linguistic content without it's being expressed, without expressions having that contents Here, of course, again we shall have to make



use of* catalysis and I »hal! leave that, as I said, to a later Jecturea A sign expression and a sign content, make up a sign« This also is in accordance with Sauasure'a theory« But we think that if you would like a general statement to be made about the kind of relation taking place between sign content and sign expression, it would be safer to say that there »3 not a solidarity as is the case when talking of the planes generally, but there is combination« This is a third kind of relation« In a combination there is a relation between two terms~~two terminals of this relation but in such a way that the first terminal does not presuppose the other and the other terminal does not presuppose the first« You might call it perhaps free combination to make more clear what is meant« I think we should state that generally there can be combination-« !re© comb?natson--between a given 3ign content and a given sign express?on« At least this statement would be required if we have to account (as we do) for the existence of synonyms and homonyms« The plural in English is a content unit, according to glossematics» (We wouud call that a content unit of that language)« It has various expressions (l need not tell you that)« The Biost common is this (z) (s) business I talked of the other day« And there are others, for instance, "en" in oxen« So in this case there is a sign content and two sign expressions and if we consider one sign content and one sign expression (linked up with it), then, there is only combination, because instead of one of these sign expressions, you can have the other one« There would be solidarity if you consider the two or several sign expressions belonging to one and the same category« Then there would be solidarity between this category of the expression




and this category, with one member oniy, of the content • Vie S I the actual fact is more complicated (of course) because there is the pJura I in the verb which is expressed (if you'll excuse me for saying so) in st t H more confusing ways«, The expression (z) or (s) can convey the content unit plural but also the third person singular of the present indicative«, So here again, if we allow for this (which of course we must), then, we shall have to generalize the statement that there may be combination between sign expression and sign contento Now, the principle of the commutation test is the following« If th© replacement of one member of a paradigm in one plane of the language can entail a replacement in the opposite plane, then the two members are recognized as commutables, or mutually commutable or having mutual commutabiUty)„ Commutation we symbolize by a semi colon (In order to avoid confusion with the colon which is used also in linguistics to denote more special things and, part icularly, as you know, the grammatical alternat ions—get, got, and so on, which is also a kind of commutation, but only one kind« If the replacement in the opposite plane, then, the two members are recognized as substitutahles« (Substitutable with each other or having mutual subst t tutabi ! i ty t,) And if you would like a sign for this (l think) we just put the semi colon the other way around (*)•= An example I mentioned briefly the other day is (z) and (s)„ In English there is commutation between thorn (in the expression) ecouse a replacement of "z" with "s" (or "s" with "z") con entail a replacement in the opposite way as is actually in sea I and zea I (end so on and so forth)« Whereas, in certain other languages such as Danish, there is substitution within (z) and (s)n l better point out again (l think) that we use





in g Iossematics the term substi tution in a much narrower sense than is commonly done,, perhaps (part l cu I ar I y) , in this country,, Within the content, there Is a commutation between the content unit he and the content unit she -» in a language like English, I write content unit between inverted commas because if you replace the content unit he by the content unit she this replacement can entail a replacement in the expression and (of course) it normally does. You would say she instead of he and h_£ instead of she. Whereas there are languages where there is no such commutation but where the. replacement of he with she in the pronominal form does not entail, (cannot entail) any replacement in the expression. In Finnish he and she are invariably f9 expressed by the sign expression (hen), in Hungarian by (o), in Chinese by (t^a) and several other languages might be adduced. Other examples of this can be found in my Essais Li ngu istiques on p, 91 *•” that is the paper in Engiish’-and some others on p„ 104 (that's a paper in French), Glossematic theory (in accordance with the theory of the Linguistic Circle of Prague) (and In centradistinetion to certain modern linguistic theories) recognizes syneretisms as such, A syncretism ("neutralization") is the result of a suspended commutation; in Danish, p and b are commutabies, witness paere 'pear, bulb'; baere 'wear, carry's this commutation is suspended at the end of a syllable, resulting In the syncretism p/b, as in the word which in orthography is written kop 'cup', but may be pronounced with /—p/ or /~b/, A syncretism of this particular kind is In glossematic terminology known as a coalescence,, A different kind of syncretism

5s implication, where not both syncretized elements, but only one





of them (at a time) is an obligatory representat i ve of the syncretisms English cats with -3. dogz with -z„ It can be seen from this ending occurring in free position (after a vowel), e«fl.,(ai?.)("eyes"), that its "ideal form" is ~z, and so the ideal form of "cats" is kasts«, "actualized" as kaeteB ? gave a brief account of this in my first lecture and ! shall not insist further upon it now since S will have to deal in some detail with commutation and substitution in my fourth lecture« In the next lecture, the day after tomorrow S shall go on with the other points in my list, starting with a few words about function, (the second point) and then in the last part of my lecture, (day after tomorrow), I shall enter (more in detail) upon the functions« Thank you«


Hjelmslev Lecture g2~2<»61

Ladies and Gentlemens On my way through the nine points which 1 intend as you know, to disucss, 8 shall begin today with point number two. Functions,, What I shall do here is, first, to provide you with a set of elementary notions and terms, and these notions, are fundamental and indispensable to our theory of functions* I, then, proceed to th© systematic table of functions,, 8 shall introduce the term functive* This term has been invented in order to avoid the ambiguous use of function which we find in most scholarly work* You would say that there are two terms or two terminals A and 8 and between these terminals there 83 a function* I symbolize that by the Greek letter ip, and then in convention aS terminology you say that there this function* But you say also that for instance, B is a function of A, and A Is a function of 8* In order to avoid this ambiguous use of the term "function" 8 suggest that we us« ’’function" exclusively to mean the link, (<p) which is between the two terminals and we call those two terminals functiveb* So a functive is simply a terminal of a function* We should not 3ay that a functive is a function of the other terminal, but that a functive has a function to some other terminal0 There are, as you already know, two kinds of functives, namely relates and correlates**« exactly as there are two kinds of functions, relation and correlation*, Relation within the process, within the syntagmatic, and linguistically within the text* Correlates within the system, or the paradigmatic, and linguistically within the language* Now, from another point of view, I suggest to distinguish two kinds of functives, two kinds of relates and two kinds of correlates, namely constants and variables* We symbolize



aenslant by small c and variable by small v* (Whereas vie use capitals for consonant and vowel«) A small c mean© constant and a small v, variable,, A constant is defined? as a functive whose presence Is a necessary condition for the presence of the functive for which it has a function« So if A is a constant, then that A is a functive whose presence is a necessary condition for the presene© of B, the functive to which it has this function* And a variable is a functive whose presence is not a necessary condition for the presence of the functive to which it has a function* So a variable is the opposite of a constant« If 8 is a variable, then that implies that the presence of B is not a necessary condition for the presence of A, I suggest to use the terms constant and variable in this way0 But I may warn you against a terminological pitfall„ One should net, of course, confuse variable and variant, which is something quite different,, A variant is the same thing as what I labeled as a substitutebje* And I think that it would be much wiser to say substitutable just to avoid this confusion,, And of course to say commutabfe for invariant, that's what i suggested iast iiane„ So far we have met with three definite kinds of functions v^iich by the way are all relations and not correlations*, these are selections, solidarities and combinatIons? selections, symbolised by an arrowj solidarities by this sign and combinations-- / symbolized by a hyphen* A selection then is a relation between a constant and a variable,, If, in our symbols we include not only the function but also the functives, then what we have in a selection is a variable selecting a constant. So this is a relation between a variable and a constant* You can also write it the other way around a constant selected by a variable* My example of this so for was the




subordinate clause selecting a principal clause« And a consonant unit, selecting a vowel unit« Then in these examples, subordinate clause and consonant unit are variables, principal clause and vowel unit are constants,, In & case like this, the arrow can be read "presupposes", as l said last time« And the variable can also be called "marginal,," And the constant, "central™" There are other terminologies in other presentat ions and in other theories* "Marginal" would correspond to Hockett^-s "onset and coda,," And the constant central would correspond to Hockett^s peak™ (cf, Jesperson's "Syllabic peak,") In a unit like this, the central alone, the constant alone, can make a complete unit of the preceding degree™ A clause of the first degree can consist of a principal clause only™ And it can consist of a principal clause with added subordinate clauses« A syllabic theme can consist of a vowel unit only, or it may consist of a vowel unit with a consonant unit added« !n both cases the peak may occur alone within the unit analyzed, or the peak may be combined within this unit with the marginal parts« A solidarity then would be a relation between two constants« So we'll have to write c on both sides of the function symbol« This cannot be reversed, of course. You can reverse the order of the actual functives but you cannot reverse the function itself which is bilateral« If you distinguish two different functives, say, cl and c2, with a solidarity, between them, the c2 solidary with cl» So far we have seen one example of a solidarity, that existing between the content plane and the expression plane« A combination is a relation between two variables, and that which was said about the symbolization of solidarity holds good for the symbolization for the combination too; it cannot be reversed unless you specify the particular actual functives





which enter into that function,, So far we've seen one example of combination, the one which takes place between a sign content and a sign expression. (I presented this example the last time,,) This use of the terms constant and variable, which I have introduced now, makes it possible to avoid the term "presupposition" if one wants to. The term presupposition may involve certain metaphysical connotations or seemingly metaphysical connotations. I think the use of the terms constant and variable would be more exact than the term presupposition„ 9 shall proceed now to give you a complete table of the principal functions., The general heading is symbolized bycpo There are two kinds of relations symbolized by R., and correlation, we symbolize by three dots (£)„ On the other hand function can be divided into three kinds? determination, interdependence and constellation,, We can symbolize the detorain«>tion by an arrow of this special kind and intendependence by an arrow pointing in both directions, and a constellation by a vertical stroke0 A determination is a function between a constant and a variable,, Thi3 is the general definition. And you know already that a determination which satisfies this definition is a selection, a relation between a constant and a variable symbolized by an arrow. A correlation which is a determination is a specification. There are, of course, reasons for having chosen these terms, but I shall not really enter into that. The symbol for that »3 like this, and it Ctsn of course be reversed as can the arrow. So a specification would take place within a paradigm if the paradigm has only one member, the constant, while in other cases it has two members, the variable and the constant. « shall come back to that later. 1 shall ju3t give this general table® an interdependence of relation Ss a solidarity,. You know



I tnat already an interdependence is a function between two constants and the relation which is the function between two constants is a so? sdai’Jtyo A correlation which is an interdependence is a complementarSty. And we can symbolize it with a cross 8 Ike this? 4- . Complementarity, than, is used in a different sense from that in which this word is mostly used in modern theoretical physics, for instance,, But not absolutely different from that. A relation which is a constellation Is a combination™ That also you know aI ready--a constellation is a function between two variables, and a relation which takes place between two variables is a combs natson™ A correlation which is a constellation we shall call that an autonomy, and symbolize it by an upright cross™ So if a paradigm can comprise a member A, without B being present, and if another paradigm can comprise B without A being present, then there is an autonomy between them. This is just the general table of functions™ You will find a presentatson of all this in Prolegomena to a Theory of Language, pp„ 20-25. ihe definitions of these functions are entirely based upon the concepts of variable snd constant« And, as 2 said before, this makes it possible to avoid talking of presupposition« But it would be perfectly legitimate to say that the mere existence of a volume called "first volume" presupposes a second volume, manifested or not™ St is presupposed by the mere fact that you call the first volume, the "first«" 5 shall give you a more linguistic example« imagine that tv© says "if only I had money, then I would go to the pictures«" "!i- only I had money"-~that's the subordinate clause, "| would go to the pictures'--that's the principal clause. There Is a selection between, ' should not say this particular subordinate clause, but the category of subordsnote clauses, and then the category of principal clauses. Now



imagine someone saying, "If 5 only had money", and nothing further, so that the principal clause just does not turn up, then the very use of a subordinate clause presupposes the possibility of adding some principal clause® If you say, "If I only had money" and that is all you say, then in order to fulfill the requirement and the utility of these definitions, then I should interpolate a whole category of possible principal clauses« "I would like to go to the picture" or "I would like to buy a car", whatever it would be® This is called catalysis® If there is no principal clause, then it will have to be encotalizad by this procedure® Not a definite principal clause® (We could rot do that«,) We might, of course, in some particularly good occasions, w© might know what the talk is abouto There may have been talk about going to the pictures, then someone saying "If only I had money" then, of course, it would be easy enough, in that particular speech situation, to interpolate a definite principal clause« But in general, you can't do it, and if there is not definite speech situation but if you are just faced with the facts of the language in general, then you will have to encatalize a whole category« And it is this particular kind of interpolation that we call catalysis® I now come to the fourth point in my list-- form and substance« 1 shall deal with that at some length because they are considered very important indeed* But I shall come back to that which we mentioned not only In connection with the functions but also iaier on« As you know, l have or I am going to devote the sixth, the last lecture to form and^ubstance«. ! think it may be a good idea to niti ate this topic, by degrees and at intervals, instead of ovai'wlielming you with everything that can b© said about it This distinction between form and substance is perhaps the most essential point that should To



comprised in any presentation of the theory of glossematics* it is, ! think, the only distinctive feature that really makes 91ossamatrcs differ from ail other recognized linguistic theories,, It is the very leading idea of glossematics and specific of 9Ios3ematics„ But I realize that this idea is also hard to understand for linguists trained in other theories,, be the?ytheør los transat I ant ic of cisatlantic ones, I shall begin by giving you a few references, two papers where this topic I© specifically treated,, There is first my paper, the "Structural Analysis of language", dated 1948, a paper which is found in my Essa«a Linguistiques pp» 27 to 35c And there is also a paper in French, I would like to call your attention to if you read French at all because it is my last presentation of this topic* And also another topic, by the way, namely the two planes of the language« This paper dates from 1954 and is called "La Stratification du Lengage" (the St ratIfication of Language) and it is comprised in the same v©Juras (Essais Li nguiatt ques pp0 38-68, Then there is a third paper also in French, by the way, but the oldest one dates from 1943« The title is "Longue at parole." (That of course cannot be. translated into Engl $ sh») In the same volume Éssa i 3 Linguist I goes pp, 69*^81 • The terminology we are using is, on the point, following Ferdinand do Saussures la Langue est une forme, non une substance« Now then, what Saosaure calls wLa Langue'- ss a form not a substance^ Now i believe I should call your attention to the particular use Saussure mokes of the term "Langue" which by the way is generally accepted in linguistics in the French- speaking world« Saussure has it that l*untque ot veritable objet de 5,

; a

I s.aistlque est fa langue en af!e-«e«ft et pour el te-metee• That is, the only true object of linguistics is la langue,, considered in itself and


for jt3 own sake,, Of course, this would amount to nothing, but a too?, a truism, if langue is translated by "language«" language is the only true object of linguistics, that is obvious of course« But there is a distinction in French terminology”'' between 5argue and parole« The common denominator is language« So language consists, according to this theory, of two parts, 2angue and parole. The distinction between langue and parol© is not clear* in Saussure's posthumous. Pours de f * j I ngu i st I gue genera I a, In detail.« Langue and parole are ambiguous but they do not overlap« And whatever the interpretetion, the distinction is extremely important. But it so only really recognized by the Saussurian school, and has beers taken over by g I osseroat i cs though not with the same meanings. We just say form and substance. To make the story short, now, I shall confine myself, to insisting here upon giossematic interpretation which seams to be also ono of the interpretations that Saussure himself must have had in mind« Langue. then, is the pattern of the functional form, proper to the language as such. The Saussurian term Sangue can, then, perhaps, be rendered in English by "linguistic form" or simply "the form", in the linguistic sense of the word« The "form" consists of various linguistic forms, of course, and a form within the Form, as the snore comprehendive term, is defined by its linguistic functions, and by noth?rig else« So, l warn you that form does not mean necessarily anything morphological, but anything which is defined by linguistic function and by nothing else "by its linguistic functions", that is to say j. "by its relations and corps \ a- 'Jons, within the interior linguistic pattern «

Correlations comprising commutation and substitution



* a

which i s very important to recall since the language in our sense comprises two planes, content and expression, whose interplay must constant I y be taken into cons I derat ion,. Form, then means functional form which is linguistic and nothing but I i ngu s st i c => Form is opposed to substance*. There is a linguistic substance or several linguistic substances, that is to say, substances attached to the linguistic form and manifesting the linguistic form, but different from the linguistic forms themselvesB in order to make this distinction clear, it might be useful to point out that, if form be taken as defined by its functions only (such as the functions I have just been speaking about a few minutes ago) then it is obvious that languages (in the sense usually attributed to that word by linguists, the so-called "natural languages") are not the only objects that have these functions and this specific farm0Saus3ure'f3 pet simile, if I may 3ay so, hla favorite comparison was what of tSie so-called natural language, considered as the Iangtie with a game of ch©s3tl And Indeed, the functions observed within a game of chess have very much in common with those observed within an ordinary language« Sts 11 more so, the principles of the functional pattern of ordinary language are the same as those of the functional pattern of what mathematicians call a languagea Not only any scientific terminology but also the symbol systems of mathematics, of symbolic logic, and so on, would from our functional point of %'iew be languages in this broader sense« These may, or course, be very complicated and highly sophisticated systems« But there are also examples of very simple and everyday systems, even we f known to the man in the street, which obey the same principles of functional structure* indeed, Sadies and gentlemen, I was


astoundingly right in saying "the man in the street", since traffic lights give us a good example of this« Traffic lights, where two sheets intersect, make an excellent example of a very simple structure essentially similar to a linguistic one,. Here, likewise, we are faced with a plane of expression and a plane of content0 The succession of redt amber, green, amber, which are the sole units of the expression in that system, corresponds to the content units stop, attention, proceed, attention,, Further, the telephone dial or a telephone switchboard where letters and numbers can each be said to represent a glosserne of the expression and combine into sign expressions whose content is the exchange or the extension respectively and the number or the telephone subscriber himself« In some lectures 8 gave some years ago, in London and Edinburgh*, I developed these examples, not merely for the fun of the thing, of course, and not to serve pedagogical purposes only, but in order to gain a deeper insight into the basic structure of language, and of systems similar to language, by comparing them with ordinary language in the conventional sense„ I used them there to throw light upon the fundamental features which, according to my definition, are involved in the basic structure of any language In the conventional senseu Namely the following! A language consists of a content and an expression,, A language consists of a process and a system« Content and expression are bound up with each other through commutationD Ihere are certain definite functions within the process and within the system« A linguist would object to adopting the. logician’s and mathematician's use of the word "language" to denote structures like these and he would object to referring to thorn -3S. J have been doing up to now as languages« I shall not do that,, I




5 shall refer to them as various kinds of aenioticsa This term, of Greek origin, means simply, sign system, or, as I would prefer, systems of glossernes combining into signs-. The term semiotic may be adopted as the common denominator for any structure presenting the fundamental features I just mentioned., And a language, in the ordinary linguistic sense of the word, is just one example of one kind of semiotic« It will remain to be seen later, what are the distinctive features of a language, as opposed to other kinds of semiotics,, There is a solution to that, a solution which has not been given by me nor by any 91 ossetaat i c i an, but by the logician Tarskio {Professor Tarski who is now at Berkley«) Today S wanted to insist, not upon the differences between language and other semiotics, but upon the similarity between language and other kinds of semiotics« And the similarity will make it clear that the functional form which defines the interior structure, is the system as such, as being semiotic, That th»3 functional form is independent of any substance« Indeed, one could say, something which iin English would sound like a paradoxical joke — that the material units of the substance are immaterial! to the framework of the functional form« She colors of the traffic lights are the expression substance, (hey have been chosen arbitrarily, and any other colors or indeed any other means, for instance sounds, pictures, gestures, like those of the policeman leading the traffic at the intersect ion, might serve exactly the same purpose«, So one substance might be replaced by another without affecting the semiotic structure, Several substances may be correlated with one and the same semiotic structures. The same holds good of the. arbitrary and artificial devices known as the telephone dial and the switchboard—and their various parts-, Two similes of Saussure's


a» may interest us here0 One is his comparison with economic value systems,, A coin,, a banknote, a check or any combination of coins, of banknotes, of checks, are so many substance -=* units of the expression of a certain content value, say for instance, a U 0S„ dollar., This content remains the same, irrespective of the expression substance chosen«, The other comparison of Saussure's is that of a game of chess,, Saussure has it (and I think he's right there) that a chessman is defined exciusit'ely by its relations to other chessmen, and by its relative positions on the chessboard, whereas the externa! shape of the chessmen, and the substance of which they are made, whether ivory or wood or whatever it may be, is immaterial to the game* A chessman, say for instance, a knight, which usually has the shape of a horse's head, might be replaced by any other piece which by convention might be adopted for the same purpose«, If during the game of chess a knight is by accident dropped on the floor and goes to pieces, we can take any conceivable object of a convenient size and assign to that object the value of a knight, i« e, assign to it, or consider it as a substance unit manifesting the same function unit«, Weil, ladies and gentlemen, the same holds good of language, in the restricted sense of what in French is called la langue« 'ts structural semiotic pattern is independent of the substance by which :t J3 manifested* The expression system, defined by its functional structure only, remains the same whether St be manifested by a phonological substance or by a graphic substance, and so on* in effect that seems to be in perfect accordance with everyday usage * Perhaps you would not. expect iit to be so, but it actually is«, fit covers exactly what the man in the street would suppose a language to be, fit is nothing but a mere, commonplace to state that English when spoken, English when written, fTnglish when telegraphed by means of the Morse code, English when



signalized by maans of "the i nternat i orsa 8 flog code of the navies la, in all these cases, essent i a S ly ona and the same language, one and the same 8 angue in the French terminology, and not essentially four different languages. The units of which it is composed differ from one of these cases to another, but the framework of functions between these units remains the same and this is what makes us identify that particular language® Accordingly this framework, the entire functional system, must be the main object and the chief concern of linguists. And itr s in this sense that I mean that Saussure was right in saying 9 that "la langue est I'unique at veritable objet" -- the only true object of linguistics^ whereas the actual man i Festet s oOsS of this framework are immaterial to the language in this stricter sense of the word«, W® should not fail to observe, however, that Saussure did not mean to discard phonetics and semantics altogether. And it is not at ail our intention co do so. Saussure assigned to phonetics and semantics the modest role of ancillary sciences. Well, I have not been talking here about semantics but only about phonetics. I have chosen the expression plane, as a convenient example to illustrate the difference between form and substance. There is, however, an exactly parallel distinction to b© made within the content plane. There is a substance of the content and the form of the content® In the content plane, the substance or substances are what is generally known as meanings. It is equally important to know this, of course, and that not only phonology Or phonic theory including phonetics as well as phonemics (insofar as phcmemics is concerned with sound units and sound features)--not only these but also semantics as far as it is concerned with features of meaning--ss to be considered an ancillary science within linguistics



3! es a whote0 But as ! to!d you, I shall not enter further today upon the substance of the content but leave this to ray sixth lecture« Suffice it to say that meaning as opposed to the framework of functional form within the content snakes the content substance« The relation between form and substance, then, •s a selection, substance selecting form, form being selected by substance„ This is a logical consequence of the Fact that various substances may be coordinated to one and the same form« You can change the substance and the form can remain the same, whereas the reverse is not true« By the way, such units as form and substance should not only be considered from the point of view of their mutual relation, but also from the point of view of their mutual correlation« Although this would present a more theoretical interest while the real practical thing is to know that there is this selection« 1 he correlation between form and substance is a compiementan?ty« That *3 to say, within the total system built up by form and substance, there are two members then in that total system, sn tnat large category, form and substance and they are both necessary members«, The idea of a form presupposes, as a correlation, the idea of a substance and conversely. But from the relational point of view, substance presupposes form« And would you please note that I define substance by its presupposing the form. Because It is obvious that substances may also be studied from other points of view. For instance phonetic facts--whatever they may be, acoustic or just physical or just auditive and so on, may be studied also from other points of view, for example, from the physical point of view, which is not linguistic, and which has nothing to do, strictly speaking, with linguistics. In that case, I wouJd not use the word substance to designate such objects but



purport* It Is obvious to me that the word purport would better convey the meaning to English speaking persons when talking of the content than when talking of the expression« The purport is, then. In the general sense, the meaning involved in any statement, but without any regard to the linguistic form in which it is molded« But if it is molded in a special form, if it presupposes a special form, or has a comp Iementarity with the special form, then, we should call it a substance« (And only in that case, we should call it a substance.) I am under the impression that much in these conceptions must strike the linguist trained in America as odd or strange. At least they seem to contradict views often held within structural linguistics in this country. As far as semantics ar*e concerned, whose object is thus considered as a mere substance of the content, my views may appeal to the mind of some American linguistics, from Leonard Bloomfield to Zelig Harris, linguists who would not include semantics proper in the study of Interior linguistic structure. On the other hand, S have the Feeling that a good many linguists not only in America, but also elsewhere, are shocked by my contention that the units of the linguistic expression are not definable in phonological terms« The minsmal units of the linguistic expression are largely known as phonemes, A phoneme is, (if S should venture a definition, and allow for possible cases where what some language analysts call a phoneme would not be recognized as such by g5ossematics) a phoneme is, according to me, a stsbstance-unit manifesting a unit of the expression form. To repeats a phoneme is a substance-unit manifesting a unit of the expression form« It is not in itself a linguistic functional form, but the substance- form manifesting it^ just as a grapheme is another substance^form which can manifest the same linguistic functional form, as is the case



CS in a phonological transcription, or as S prefer to say, a phonological notation, just to avoid the idea of something being transcribed, and just as still other signals would be« Most linguists would advocate that this phonological substance represents a higher level than the graphic and the other possible substances which would thus be subordinate to the phonological ones, Glossematics would not, Glossemat«cs, on the contrary, is prepared to recognise ail these substances as being on the same level and would not allow any consideration of sound features or phonic substance or any other specific substance to enter into the definition of the functional unit of the semiotic expression. This would, according to glossematics, be a confusion and a contradiction in terfis*?. On the last point i mentioned, the exclusion of phonic features from the definitions of the functional units proper S am very much under the impression that Leonard Bloomfield in chapter 8, of his book, language, (the chapter* called unfortunately "Phonetic structure" which to my mind t) 3 a somewhat misleading title) is indeed a representat 3 ve of the view I have been advocating here, (Read chapter 8 in language and you will see how nice it meshes with the point of view l have been presenting you with here,) As far as I know, Bloomfield has hardly had anyone to follow him in this respect in his own country, (it is a paradox that, whereas Bloomfield has had, and is still having, followers in his views on meaning and semantics, thus on the content—substance, his views on the linguistic expression, with the distinction between structural pattern, as he says, and the actual pronunciation of the phonemes, have not appealed very much, if at a!! to contemporary American linguistics« Now, according to g Iossematics, form and substance should be separated« ft does not follow that we would exclude the substances






from linguistics as a whole« It is a common mi sunderstandi ng. I don^t know why this mi sunderstanding has come up that glossematics would exclude semantics, phonetics, phonology and so on, from the linguistic description. Glossematics is simply assigning to substance its proper place within the entire hierarchy. And glossematics claims an exhaustive description and an exhaustive description would of course comprise not only a description of the presupposed constants, the form, but also of the various, presupposing, variants or substances«, I shall have to say a few words on the requirement of an exhaustive description in my next lecture«, And then S shall have to go on with the next points, 6 through 9» I shall have to be very brief there and then I shall stick to my schedule and devote the fourth lecture to the topic of commutation and substStut ion«


Hjelms lev Lecture g 2-7-6!

Ladies and Gentlemen? You wtiS remember, from one of the other lectures, the list of nine points I'm intending to go through« l have been through four of them? (1) the anaI yticai procedure, (2) the TtmctForilTTfunotions have to be taken up again in the StFTecture) (3) commutation and substitution, and content*and expression" plane (which { shall take up again "today), (4) and form and substance which is intended to form the special suEject of the sixth and last lecture® I proceed now rapidly to the next points; (no® 5) exhaustive description; 5 no o 6} synch rone c" and genet o cl i ngu i at i cs; fnOo 7) genera‘P*typoTog<c^aT~caTcuTusT" (no« 8) denotation 'and connotation;*~ard ( no a 9) met a-semi ot fcs« I think that i need not take up very much time to insist on the fifth point (the requirement of an exhaustive description) although glossematics does make a point of claiming the necessity of an exhaustive description® At the outset, this necessity is most strongly felt (I think) in three respects, namely for the distinction between form and substance, for the recognition of so called immediate constituents, during the analytical procedure, and for what l shall call the broadening of the perspective® But (as far as the distinction between form and substance is concerned) I have treated that in some of the former lectures and S shall deal with it again at some length in one of the last of these lectures® ^hat this distinction between form and substance is not tcuntamount to discarding substance I think that I have already pointed outr, I'd like to make a point of it today again® We would not like to discard substance but just to assign to substance its proper place in the entire hierarchy® if a particular merit can be attributed to glossematics, as compared to other theories--this merit v/ouid be the discovery of pure form


as opposed to substance, and the claim that pure form be investigated ss the central part of linguistics« Between form and substance there is a complementarity. This is the correlation (the particular kind of relation that exists between form and substance)« They are mutually complementary. Which means that the two functives, form a substance,, have to be investigated-(both of them). It also implies that, in cases where the correlational point of view is the prevailing one, form and substance have to be taken into considerat ion at the same time. This has been pointed out particularly in Esse is I; p,c;ti j st i cues, pp. 46 ano 47« This (I mean to say the correlational point of view) is the prevailing one—this is the case in the commutation tests and in the iu vesttgation or other correlations between members of a paradigm* This particular question of the correlations between members of a paradigm will be taken up in my next lecture. }n all these cases substance has constantly to be considered together with form. If we ask not for the correlation but the relation that takes place between form and substance, there *3 a selection—Substance selects form. Substance then i3 the variable; form,the constant in the relation. And it goes without saying cue.i, in ell case® or determ5nation, including selection, it will be natural to describe the constant first and the variable next. In this particular case, then, the manifested relate first, (the form first) and next the manifesting relate, the substance. Thus a separate study of the substance which comes afté*- that of the form must bo founded on a comparison of Form and purport* So that to be complete we have to state that substance selects (substance presupposes) not only form, but also purport. So purport also enters as a constant in this complete relation or group of relations,, (The description of the substance) The


substance is (as you will remember) the purport as far* as it is subjected to a form. So the description of the substance is a description of delineations in the purport called forth by the projection of form waits on the purport; in the content plane the limits between meanings (or co-catled meanings) called forth by the form of the linguistic content; in the expression plane the limits between (soy) sound units,, phonetic units called forth by the form of the linguistic expression. But this study can only be undertaken when form and substance have already been distinguished. This distinction between form and substance is carried out through an analysis of the whole, (of the total unit) comprising form and substance. Through this analysis this total unit is analyzed into two segments (or divided into two parts, if you like)-- form and substance. And the total unit Is generated by the selection between form and substance. (The particular selection I have called the "manifestation.") To make this still clearer, one may say that the form, as the constant in this relation, is encatalyzed and that it is through a catalysis that we arrive at the form/substanee distinction. The total unit is analyzed or* divided into this complex unit. But it is a!so catalyzed (I use this sign for catalysis) and the form is detected through this catalysis. The form is v/hat we cal! encatalyzed in this totai unit. The commutation test is the main, procedure that makes this catalysis possible and makes this analysis possibles The distinction between form-elements which have mutual commutation and other form- elements (1 shall come back to that) which have mutual subststut ion. The second respect in which 1 thought that perhaps the requirement of exhaust s ve descr iption may at the outset be part icui arIy interecti. :g is the requirement of & description of all immediate constituents



3 b

during the whole analytical procedure, so that the analytical procedure snust pass through as many stages as possible« Just to account for «ti! possible units and elements-“TeI atsons and correlations occurring ?n the object under cons {deration* The third respect in which this Blight be particularly interesting was (you will recall) what S call the broadening of the perspectivet, The first perspective of a linguistic theory like ■ g5ossematics 5® that of an individual language or (more precisely) of one definite state of such a language«, what in French is called 8 uh etat de langue, a linguistic state or (as we might put it) one stage during the evolution of that language. The first perspective is that of one stage during the evolution of one particular language. Ferdinand do Saussure called this ■»diosyhehrony« I admit that it is quite a cumbersome term (of course) but I should just make a reference to this terminology of Saussure's. This first perspective, then, which is the most elementary one and In many respects presupposed by any further perspective, is the one with which we are primarily concerned in the present lectures. But this first perspective (being the contra! one and the one that one has to start from) is by no means tantamount to saying that it is the only perspective possible, nor is it the only perspective necessary,, On the contrary, the requirement of an exhaustive description implies the necessity of adding a!! formal considerations possible through a broadening of this first perspective. The study of trie individual linguistic state has to be supplemented in various ways, namely through the next points which you'll find in my 1 1st--synchronic end genetic linguistics, general typo logice? calculus, denotation and o’.notation and meta-semiotics* These are as many broader perspectives (a broadening of the first perspective)« These broader perspectives, however, cannot be dealt with in full detail in the present lectures.



I shall hav® to be brief on these points hut, on the other hand, i should not like to pass them by in silence, since this might make giossematics seem much Jess exhaustive than it intends to be and leave the path open for certain fatal misunderstandings* These broadensngs of the perspective have been dealt with a little snore in detail in my book Prolegomena to the theory of language., pp* 65 to the end of the volume. In entering upon these broadening« of th® perspective, I shall first call your attention to a point which ha3 hardly been touched upon, nor perhaps even mentioned in the Prolegomena* That this has not been done is merely accidental and does not imply that it did not enter into the author's cons*derat ions. it did so, at a very early date,, as you might infer* from certain passages in the Esso is linguistiques the first article of that volume, "An introduction to linguistics", 1947, particularly pp, !9 and 20. I maintain the idea that you will find expounded there,although the terminology has partly been changed later as will appear from what I'm going just now, if you're able to compare it with the passage S referred to in the Essaia I i ngnistd ques, the passage from 1937. By the way, there are also passages in my book from 1935, first volume on La categoric des cas. But 1 regret that I do not have the exact page reference here. What I shall be talking about now 5s, then, what I call synchronic and genetic linguistics, I should prefer these terms. Of course, these terms have not been created by me, but I should prefer them. Synchronic linguistics, can be taken as meaning simply--the description of the linguistic state. We can represent that by a horizontal line. Whereas the genetic linguistics deals with linguistic change. We can represent that by a vertical line. And what ! should



like to stress here is the fact that glossetnat ic theory is capable of accounting for both synchronic and genetic facts—as, generally speaking, any acceptable linguistic theory should be able to do« But this is one of the points that are very often misunderstoøcL I don't mean to say that people tend to attack particularly glossematics on this point, but S mean to say that there is a strong tendency (inexplicable to me) to introduce contradiction and opposition between what we call structural and what we call genetic* So that, according to this mi»understand?ng, structural linguistics should only be synchronic linguistics* This identification of structure! and synchronic is, to ray mind, completely false« Wei! the distinction itself (the distinction between synchronic and genetic linguistics) is (of course) by no means an invention of ours« Not only is it universally recognized among linguists, in our day, but the distinction is, as you know, no doubt, fairly old« The discovery of linguistic change (the discovery that language changes) was made at the beginning of the 19th century by Jacob Gr-iiitfJSo Jacob Grimm stated very clearly the distinction between the genetic and the synchronic point of view« His discovery and his great idea was the evolutionary aspect (or as I say here, the genetic aspect)* But he admitted that during the evolution of any language, there are certain stages—what he called die Niedersetzungen dcr Sprache, That would mean, of course, the setting down and establishment of the language« The language which, according to Grimm, is otherwise constantly moving and changing« For such a language, you must admit certain periods of relative establishment« This timing of language change prevailed during the 19th century, though by no means as the on,fy concern of linguistics during that period« And the synchronic point of view was re-introduced and given the place of honor (one might say) by Ferdinand de Saussure,



o but aS so before Ferdinand de Saussure by Vector Henry« (This is not so we I I known«) Victor* Henry has partly been forgotten so I shall just give this reference. His work on the subject has the title Ant i non? i e 1 i nciu i st i que.. And still earlier the same discovery of the necessity of this distinction was made by the Dane i mentioned the other day (H. 6. VibeI)* in the beginning of our century, but published in Danish unfortunate5y. It took some efforts to overcome the bias of conventional linguistics against what was considered with some horror as a revival of linguistic philosophy of antiquity and of the middle ages,~~ihe only cast in which the synchronic approach was known to have been molded. A Pew words about the technical terms and notions should not be out of p ace since there are certain biases and prejudices, even misunderstandings that are linked up with the various 3ets of technical terms arid which find their adequate expression in them. By the way, as early as 1928, in my first book on the Principles of General > Grammar ? warned;earliestly nst such mi sunderstandi ngs, although it has turned out to be more or less talking to the wind. The two points of view are often distinguished as historical versus descriptive. The genetic point of view s sometimes, by quite a few scholars, called historical and the synchronic is called descriptive. Well, to use the word historical seems to me perfectly innocent, although it's a mistake to do so because the question is not only of history but also of pre*history„ !But the term descriptive is particularly unfortunate if it is applied to synchronic linguistics only. And the term descriptive has been given (as you s|| know) a rather inconsiderate revival in many even quite recent textbooks and in modern linguistic 5iterature in general.


in the restricted sense of synchronic linguistics,, this is to my »nine!, unfortunate because any pea1 scientific treatment of an object must necessarily be descriptive by definition«. And so is not only synchronic linguistics but also genetic linguistics* Only the one is a description of a state, the other a description of a change« But they are both of the«* descriptions« It is still worse if genetic linguistics is called explanatory or the only explanatory linguistics possible« For synchrony is also explanatory, but in a different way, of course« Genetic linguistics explains by cause and effect, synchronic linguistics by reason and consequence« Now another set of terms is diachronic versus synchronic« What I have here called genetic is also called diachronic« What l have here called synchronic is called that« So I shall recommend that term, of course. But I think l should warn against calling the other one diachronic« These two terms, synchronic and diachronic, were Introduced by Ferdinand de Saussure* St is Saussurian terminology you are using, ail of you, in this country* But on© must be aware of the fact that by diachronic, Saussure really meant something more restricted than the sense given to it largely tn the present day* As far as we know, Saussure was of the opinion that thf3 genetic linguistics can concern only single facts within the language—not the total system« At least that is what ts stated more or fees expressly in the posthumous work of Saussure. Cours de I s ngu t st tque genera Ie, though j think that this is to some extent concradicted by his first work. On the Primitive System of the Vowels in the_.lndo■»European Languages. Well, if we stick to the formulations g>ven in Saussure's Cours de linguistique generals, then we find i-ha-i- there is a synchronic approach possible—(that is the great discovery Ot Saussure)p and on the other hand there is the diachronic aspect




which cmJy concerns certain single facts within the language, not the language as a whole« What Saussure had in view was (of course) primarily what is known by the name of sound laws« It has been shown (i think) sufficiently clearly that this diachronic approach in Saussure's sense is not the only possible in genetic linguistics but that we may also ask the question how a whole language changes, and how the system changes« We can consider not only a change of elements but also a change of systems« So with this l will represent the synchronic state involving also the total system« Then there would be this particular expect which was not admitted in Saussure*® theory as 1 mentioned before. (I think it was admitted in his practice but not in has theory.) We should be aware that in has theory he only used diachronic in the first sense, not in this second one,, and he never U3®d it in that sense, not even in his practical work where Ire does not use these terms at all. 1 think we should call it something else - the theory of the change of

I have invented this term, metachronic, which (by the way) has been used somewhat vaguely by quite a few linguists, mostly (? think) by Schuchart, the Austrian linguist. In my book from 1935—1st Volume of La catégorie des cas — i intooduced this term to denote the theory of the change of whole systems. Both the metachronic result and the diachronic result, in Saussure's sense, are necessary and both metachronic and diachronic problems are worthy of attention. But they should not be mixed« And us a common term for metachronic and diachronic I think that it would be suitable to take this term "genetic" as the common denominator, covering the two. Here again, there (3 a terminological revival, to my mind i neonsiderate and unwarranted,





In our day in modern writings. And this time, the revival is that of the term diachronic in a new sense which was not that of its creator, not intended by its creator. Diachronic seems to have been taken over because it is considered (as it were) simply more fashionable than just to say the theory of linguistic change. Why not say that, by the way, if that's what you mean? The way in which gføssemstica would account for genetic relationship would be by using the same notions, (functional notions) as those I have already made use of in the preceding lectures«, Say there's one stage of a language, cal! it stage number I, and then there are later stages we roav cell them 2« Then 2-1, 2-2, and so on* You may think, for instance, of Primitive Indo-European for i and the various Indo~European language families for the 2*3* or pre-Romance Latin as I and the various Romance languages as the 2's. Between these different stages there is a specification signs The later stages specify, in our sense of the word, specify to the 1 preceding one« Because the first thing we have to ask, in order to decide what functions we are faced with, is the question whether this function is a relation or s correlation. A relation will always mean a co-existence between the two relates« There is no such co-exsstance between one stage and the next stage, (This is obvious*)« On the other hand, the second stage presupposes the first as the necessary or desired interpretation, (explication of the 2nd stages). Between the various related languages (what l cal! the 2's) (between the various 2 ) there is a relation* They may not always co-exist, and then there Is a combination between them* You can have one or two or three, etc* There is no necessity (there is no possible constant)


there is only combination* At the same time there is a complementer ity* You will often have to state complementerity between themD You know (of course) that in many cases we have to reconstruct the first stage (what l have here called the first stage). Often we have to reconstruct it* No, we have a I ways to reconstruct it* There is no first stage without a reconstruction« I should not (of course) adduce Primitive indo European as an example, because it is obvious that a reconstruct ion is necessary* (This was my point when I was speaking of pre-Romance Latin* Because what we really need in order to explain a relationship, between the various Romance languages is a reconstructed language which w® reconstruct on the basis of the existing Romance Languages* This language is very near to classical Latin but not identical with it* And we may distinguish between the two* So there is always a reconstruct ion involved in genetic considerat ions. And a reconstruction means nothing but (if you would like for me to translate it into glossemat ic language)-- it means a catalysis* We encataiysize the constant which is exactly the first stage specified (presupposed) by the variants that are represented by the following stages* The genetic classification of languages into families is not the only one* It is also possible (though this has not been done to the same extent) to classify languages into types, not families but types, the genetic classification into families is a cl assificat ion based upon a comparison of the elements of the expression of the languages concerned* The classification into types is based on a comparison of the categories of the languages compared* We would like in linguistic theory. and perhaps we would stress that most


emphatically in glossematScs, that a real theory of language should have to be built up as a general calculus of possible linguistic structures« This means certain axiomatic® (i »hall not enter on this question) but on the basis of certain fundamental assumptions a general calculus of the possible linguistic types has to be built up« And this is necessary because a theory should not only be based on pure observation (not only on pure empiricism) but the theory should allow for semiotic structures that have not been observed so far and are hitherto underscrsbed« This goes to show that it will be desirable (even necessary)' to build up the whole theory as a general typological semiotic calculus« This was in Saussure's mind when he created another cumbersome term, that of pansynchrony» In this way (and this way only) the theory will have one of the distinctive features of any good theory, namely that of being able to predict« Not only to observe, but also to predict« This is ultimately the essence of the theory In its accomplishment, more than in its pre-requisite (it's true) and th© pre-requisite of the theory Is what I've had mostly to confine myself to In the present lectures« Within a typological hierarchy, the functions between the objects compared would be the followings--the types are of various hierarchic stages, of course, and this is relative, so that we can only talk of general and special types, relatively to each othero A relatively general type, then, is selected by the relatively more special types« We may call them S-l, S-2, etc« Because (as you will recall) the first thing we have to ask iss "is there a co-existence or not between the general type and the special type?" She special type may be represented by one well known language of course,--that is one of the possibilities« Is there a relation between


a ti the type and this language*, 3s there a relation or a correlation? Is there a co-existence or not? There is (of course) a co-existence«, The type is in that very language« It can be observed within that language« The language is a sample of that type, so the special, relatively lisore special, types will be samples of a co-existing relatively more general type imbedded (so to speak), or embodied in them,. Between any types of the same hierarchic stage, there will be a compleinentarity -- one presupposing the other, the other presupposing the first one« Here again we are faced with the necessity of catalysis« What we do when we state that these are representatives of this general type is just to encataSyze this general typo -«* just to interpolate it as being within these languages and then sorting it out exactly as we do for instance with form relative to substance. Wei I, I now coma to the 8th point that of denotation and connotation« This has been treated in the Prolegomena pp„ 73*"76« I think S can be brief also because there "is a list of terms and definitions and also bibllographicai references which is going to be distributed to yon, ladies and gentlemen, during these lectures« There you will find the references and the definitions of (l should not say all but most) of the specific terms I have here been 15 få 3 SI 0 u, The solidarity between the content and the expression (content plane and expression plane) which generates a language, or more generally a semiotic, if the content and the expression arc not of a particularly complex character, is what we cal 1 a denotation. The sign function -- the solidarity between the content plane and the




<b A O 4 u expression plane of an ordinary language, for instance, would be called (according to this terminology) a denotation« But now a more genera! calculus may foresee that the expression plane of a semiotic is in itself a semiotic (or a language) so that we would have a more complex situation where one content (we could possibly call C2) has an expression, of course« Otherwise it would not be a semiotic (not be a language) and this expression is »-equal to a language in itself, a content Cj , solidary with an expression« Sn this case, where the expression is»~ equal to a language or s semiotic, then in this case we call the solidarity between content 2 and expression 2, a connotation« This may seem odd to you, but it ts really a very simple thing« We would account, by means of this theoretical device, for various styles, vernaculars, national languages, regional languages, individual and personal physiognomies, as concerns the expression, different voices or organs« All these would express a content of their own« It seems obvious that the solidarity that exists between such things and what they express is exactly a function. St is obvious that the Danish language can be said to be expression for the corsnotator Dan i sh or Dane or Denmark, Likewise, it is the semiotic complex (the semiotic structure) which we designate as the linguistic physiognomy, that is the expression for the real physiognomy, that person. And correspond?ng!y in all the cases I have here in view« Not for nothing does the national language stand as a so-called symbol for the nation, the local dialect as a symbol for the region, a certain jargon as a symbol for a social class (and so on and so forth). Thus it seems appropriate to view the cQnnotators as content for which the denotative semiotics are the expression. And to designate this content and this expression as a semiotic, namely as a connotative and not a denotative semiotic« In


other*'.words after* the analyses -of the-denotative semiotics is completed, the connotat?va semiotic must be subjected to an -analysis according to just the same procedure« Met a™ sets sot i cs, my 9th point is roughly the same thing but the other way round« Here we consider in our calculus the possibility of their being a semiotic whose content plane consists of a content and an expression, whose content plane is a language or a semiotic*. Vile consider an expression (call it expression 2) expressing a content*—cal I it content 2-** this content being equal to a whole language with two planes C-I and E-I« Well, this possibility was actually foreseen a long time ago, or relatively a long time ago by modern logistics, particularly first in Poland, now all over the world« . r

Any terminology is the expression of such systems« Any linguistics, ladies and gentlemen, is nothing but an expression of this, (an expression, a description in expression terms of course, a terminology being involved) art'expression of a content which is equal to a content and an expression in mutual solidarity. So there's nothing really remarkable about this, but l would point this out and that this seems to fit in the glossematic theory and so our theoretical calculus can account for and may mesh very well with modern logistics and various modern epistemological approaches* ) have been through these points and I shall net come back to them very much because they all concern not what are called the first perspective, but broader

perspectives which I thought it expedient to touch upon witnout reaiSy entering upon them. It was my intention to enter into some detail on commutat I on and subst i tut ion which has been dealt with in some of the former lectures. I thought it expedient to give more examples of commutations and substitut ions-"if you would allow only for the


possibility of there being a commutation not only in the expression plane but also in the content plane, and for the existence of a content p:ane and the necessity of taking that into considerat#on, to account For the structure of the content plane unraveled through the commutation tests« These are so many points in which g Iossematics may differ from various modern linguistic theories that 1 think I shall have to enter at least briefly upon this at the beginning of the next lecture, then 3 shall have to tel! yow about the special functions called correlations between the members of a paradigm.


Hjelmsiev Lectures 2«'9>0I 51 Ladies and Gentlemens I would like to hand out a list of definitions and references (terms, definitions and references)« So I might at least refer to that in my last two lectures,. As I told you the other day, I thought it might be expedient to give some examples of commutation aid substitution, I do think I shall have to do so although the principle itself is (no doubt) well known to all of you. And glossematics has not really much of original ideas to present you with in this connection, Perhaps there are a few facts I ought to touch on again in this respect, Glossematics would recognize syncretisms better known I guess by the name neutralizations. Whatever you call it, a syncretism is a suspended commutation or what results from a suspended commutation. And we also recognize what is known as participations»»I shall have to go into that in some minutes. But ! shall Hke to say that sometimes glossematics would carry out the reductions differently from other theories or applications of theories. Perhaps one day we may come to agree upon such reductions. Glossematics goes as Far as possible in reducing and having a really irreducible inventory of final units or elements. So in some cases glossematics would tend to apply further reductions than those mostly introduced in the analysts of particular languages {of definite languages), I have adduced the instance of the correlation between the unit manifested by unvoiced js and the unit manifested by voiced ss, There is, as you have seen, a commutation between the two in the case of English sealj zeal, hi33s his, for example. Whereas in Danish (for instance) there is no commutation, just a possible substitution of the two elements. We have in Danish substitutabIes



* a ør (as some people would say) variants, 3n English they are commutables Or» invariants« It is obvious that French Is in this respect a language like. English,, There is the difference between polsson, (fish); poison (poison)* and so on and so forthn Now one might ask the quest ion, "What is the position of this problem as far as Modern High German is concerned?" We have (as you know) reisen (nayzn) to travel! whach 3n (Standard H§ gh German has a voiced (z) end on the other hand re is asen 'raysn) to tea£ with the unvoiced« So if we just stick to these examples then, of course, we have to recognize simply a commutation between the two units just like in English or In French, but there 13 just this difference«»that in German at is possible to go a step further. At the beginning of a word In German it would be possible to have (z) for instance (zi) (zo) and so on but as far as Standard High German is concerned, not (s)0 So this seems to show that at the beginning of a syllable, (z) is only possible one of the two elements« And if can generalize this rule (without any contradiction with other facts observed within that language,)then we should do so. As far as 1 can see there is no objection to this general!zation and we could then introduce a syllabic limit (boundary) here between rei~ and »senj here between reis** and -00« That would account actually for the difference of form in all cases« Where we just state the form, of course, we presuppose certain rules as to manifestation,, And then you must have to note, once for all, that at the end of the syllable you can*’t have (z), but only (a), whereas at the beginning of the syllable, It's just the other way around,, Witness also, for instance, the difference between der Weise (Vaiz&)'the wise one , There is something to support this assumption (l think) because



the lexical form of this word is actually weise (yaisa) like this in the orthography, in two syllable© whereas the lexical form of the second one is just weiss with one syllable, pronounced with (s)» in Danish we have two phonemes (&) and (y) (well I intended to call them phonemes, perhaps you would not and sine© I'm aiming at a further reduction, than our common terminology would be different)« By phonemes, I mean certain manifestations or classes of phonic manifestations. (■$) and (y) these are only found in the interior of a word, whereas at the beginning of the word they can't occur, but (d) and (g) occur under such conditions that it is possible to reduce, to identify them from the point of view of the form the (%) and (d) as well as (y) and (g)„ This also means that we have to refer to different divisions of syllables, in the two cases. But there is no contradict ion (this can be done without any contradiction) with other facts« There Is quite an interesting example, that is the word (medisfn) 'medicine'« This word can be pronounced in two ways, 1 did it now (medialn) or (me’6 isin) wstb ($). Well, that would, according to one theoretical snterpreaation, correspond to two different syllabic limits, two different ways of delimiting the syllables. Well, 2 know, ladies and gentlemen, that you would all be thinking of what is known as juncture. I'm not so sure that it is the 3ame thing and ? must confess and tell you that 1 have never understood rightly what juncture means. Perhaps it's ewe same thing but I have the feeling that juncture (this term, this

motion) or these notions which have been introduced recently in some kinds or American linguistics have a rather doubtful foundation (dubious foundation), and that they are still controversial» What we do in 9os sc mat \ cs in such cases is just'to assume a syllabic limit (arbitrary ly)


where it comes in useful for structural purposes and is not contradicted by other facts« There are also other cases of reductions within glossemat ic analysis which would perhaps go further than most people would acknowledge, would recognizes A (p, t, k) in Danish end there is a commutation between these units and (b) (d) (g) roughly as in English, French and German, altnough the pronunciations are not th© same« But in Danish and perhaps in some other languages, it proves possible (to reduce) to identify p partly with' bh" and partly with^hb* t partly with "dh" and partly with "hd% and k with "gh" and partly with "hgo" So this reduction *s actually carried out in analyses of modern Danish from the glossematic point of view« VieII, as I told you, all or most or many of these reductions might even be accepted by adherents of other theories but I can't tell yet„ So far we have cur own analysis and reduction that go as far as possible* And it is not only a question of knowing how far to go* it is also a question of distinguishing form and substance as we are used to doing within glossernatics* This is one of the conditions, not the only one, but on® of the conditions for some of these reductions« A very important condition would also be the defining of the syllable, in general and lire I 2 ! shall riot go into that now, I think that meanings would differ considerably about syllables, this problem of th© syllable« As far as I know so far no phonetician has been able to tell us what a syllable is, and so it would be up to the structuralists really to tel I what a syII able is« We have our definitions l just gave you a sketch of this the other day, but this syllable, thus defined, need not coincide necessarily with the syllable recognized in






other theories« You know what the commutations are in general. Perhaps you would not be used to the term commutation. But of course the procedure and the thing you know well, as far as the principle is concerned« So in a way then, the commutation between (s ) and (s ) in such languages as English and French might suffice to make clear to you what we would understand by commutation in the expression plane of the language« And similarly the opposite, the subat»tution--that*s to say the lack of commutation between the two elements. If in French or in English (f) and (s) be exchanged mutually, in the expression, then this exchange can entail some -exchange in the language content« It is the definition of a commutation whereas a substituts or is tho case where the exchange of two unit3 cannot entail an exchange in the opposite plane, in the linguistic content« You know that this difference, (the difference between commutation and substitution) is what contributes most to make language differ, to make languages differ from one another. So certain languages have this commutation« Others don't have it but just have a substitution. I was just talking of the further reduction *n Danish, but (well apart from that) there are many languages with an obvious commutation between (pi) and (bi) like English /per/ and /her/ and so on,, Whereas there are languages which have a substitution in that case“-for instance Finnish. It is a well known feature in the structure of Finnish, very well known to anyone traveling in Finland, that a bank is called (pankkt) and that a ticket, the French word billot is (p! I'etti) „ Now In these cases where we write a p, they would mostly pronounce a (p), an unaspirated p in such words, A (b) does occur is* the orthography in some other foreign words, but perhaps more recent i ,, benei ini, baktl Ire* • tho orthography reeogn 20s



a ‘,b) but the ordinary prommciation la with (p)* So one might say (just jtdging from the pronunciation only) that Finnish has a (p) and no (b) but the fact is that there is no possible commutation between (p) and a possible (b) which might arise in pronunciation in some cases« So this makes a considerable difficulty to a Finnish speaking person who acquires a language with this commutation between (p) and (b)0 He carsnot distinguish very easily,, for example, between an orobre SI a and lead when speaking Swedish« So there are many difficulties and many jokes when speaking about this« It's a well known fact that the plosives going from palatal (k) and further back in the mouth to the uvula (<^) make a continuum« You can make a plosive in any place here, phonetically speaking* But the limits are different and the commutations are different in different languages* In Danish we would 9 pronounce something like a palatal (k) in a word like (kil ) meaning a fountain or a source« We would pronounce a normal velar (k) in a word like (kaoiia; call more backwards in (kuls) cold',and uvular or nearly uvular (k) in a word like (krut)'powder'» These are positional variants as you would call them« That is to say there is no possible commutation in Danish between these four k's. You can replace one by the other experimentally without being able, to change the content in any way. Whereas, for instance, in Latvian there is a commutation frontier between the palatal "k" (k) in a word like (kulis) "mace or club' or something like that, and on the other hand a "k" further back in a word like (kulis) * sheaf'? In Esk i mo th«r«v i a a well known frontier between uvular (q) and the other possible "k's*" There is a commutation there and only there* You have (<>omac) 'tapeworn' and (komak) 'louse', These examples, or rather the principles involving these examples are so well


known to any modern linguist that \ shall not insist further upon this., I should like to add that it ia not only in the phonic manifestation or substance but also in the graphic mansfestation or substance that you would find distinctions in some cases which we have to interpret as real formal distinctions« For instance, the two s"s in normal German orthography, the bhort s* % Aj , end the "long s"i /j , short s occurring only at the end of th® syllable, long s, in all other positions, according to norma I prescript sons« Then there ar« individual variants that is to say, substitutions between variants manifested in the graphic substance like these two kinds of "r", r and Jv and two kinds of t and 4r and so ona These examples have been adduced, I think by Saussure first. Then of course, we have syncretisms and we shall hav« to add this as a suspended commutation. Languages which distinguish (z) and (s) may also have (under definite conditions) a syncretism, a neutralization, of the two as is the case in Russian, In the middle of a word, Russian ha© this commutation, but it has a substitution at the end of the word and at the end of a syllable before an unvoiced contoid, that is to say before a consonant manifested by an unvoiced contoid. Now, since we recognize two planes (which w© have to do in order to carry out the commutation test) we shall aiso have to recognize commutations and substitutions in the content. As soon as we discover that an exchange of one content ©3«w»©nt with another can entail «©»*>• Exchange in the linguistic expression, then we have to 3tate a commutation between these two content elements. And vice versa (the other way around) »f there is no entailing of an exchange in the linguistic expression, then we have to state a substitution between the two content units. And this again gives rise to some very important, perhaps the most important — structural differences between languages« You will recall from one


of my lectures here the example of the pronouns he and she. We have i?« English, a commutation of these two content forms« Because there are just two different expressions,, So they may be, actually nearly always are, expressed by two different expression units* Whereas in Chinese and Finnish and Hungarian, for instance, there is a substitution between the two. These languages have only one pronoun, one would say, one personal pronoun or anaphoric pronoun, that is one sign which signifies or can be translated he or she indifferently or (to put it another way) they have the two substitutables in the content he and she according to the content«, Well, similarly, the familiar European languages distinguish between brother and sister, as conteirst-units, because they are expressed differently«, Malay, however, does not make this distinction and possesses only one sign (one word if you would like) signifying brother and/or sister indifferently. So that only the content or paraphrasing can make it clear which of the semantic variants is intended,, (Which substitutables are intended)«, Again there are other languages which do not merely distinguish between male and female siblings, if i may use this word, but also distinguish between eider or younger brother or sister« Thus, Chinese and Hungarian, for instance, have four words correspond#ng to our two words, brother and sister,namely one word signifying "elder brother", one signifying younger brother", one "elder sister", and one "younger sister«" That is, the content form of these languages draws more distinctions within the content substance than do other languages. We also find that one and the same substance area or substance zone in the content is formed in different ways in different, languages so that though there are distinctions in most languages, they are drawn in different places and are thus obliquely placed In relation to each other« A familiar* illustration of this which | have given in several papers is the relation


of ii>te two Engl sah words Tree and wood,. And the two Danish words (tre) d (stok)c We have just this distinction in both languages but differently«, There are the same lines drawn between them« What the content-unit expressed by (tre) really means sn Danish is a tree and wood in the meaning of wood as material„ This table is made of wood* for instance,, There you would use the word (tre), not the word as has beer* suggested by the French linguist Sauvageot recently. He knows Danish very well but perhaps not well enough,, VI11 is a little archaic and so there would be a particular stylistic effect produced by that word« Well* then (tre) means tree and wood but if you translate from English to Danish, then, wood means two different things«, Wood is not only wood in the form of material but ai30 wood as a collection of trees, a collection of plants, vegetables and then we use the word (sko7)u Well, it is still more complicated of course, shies in English you also have the word Forest which we would have to translate also by tssko”) in Danish® Weil, ladies and gentlemen, I shall nox, take up your time with such exxtnples since it is so well known from t?je works of Benjamin Whorf and Korxibsky a I so,{General semantics) that there are such different limits in different languages, A good example which has often been adduced, also by myself, is that of the adjectives denoting colors, I have mentioned it on pp„ 33 in my book Prolegomena ani it is adduced by the comparison of various American Indian languages, by Ho Ao Gleason in his recent introduction to Descriptive Linguistics 19^5 PP» 4 and 5« And many other lexical examples could be Introduced to show that there are commutations and substitutions in the content, and different frontiers between the units of the contentVarious lexica) examples may be adduced, but also what would be called mostly


morphological examples, And this contributes to the fact that different languages have a different number of members of one and the same morphological category? a different number of degrees of comparison, jf they have any degrees of comparison, which is rare in the world of language; a different number of voices of the verb where Greek has 3, Danish 2, and English,German and French zero if we count what I think we ought to do only the simple forms, not the compound forms; the grammatical numbers may comprise a dual, not only a singular and plural but also a dual; in some languages the grammatical moods, that category comprises an optative in Greek but not in various other well known languages; there are three genders in Greek, Latin and German, two genders in some languages, zero in a good many; there are six tenses (I think) in Latin, 5 in French, 2 (that is to say two simple, not compound) tenses in English, German and Danish, An ancient Danish grammarian, Eric Pontoppidan In 1644 set up this funny grammatical paradigm (for Danish) of cases, St's a word meaning 'man'« Nominative (man ), accusative (mas ),., dat i ve (mas) ablative (ia«a)„ genitive { mos ), (so there is something) vocative (o mae )0 Of course, we would never say that, o a Well, you might perhaps in English use oh in that way but in Danish we would never do that. Of course, this is wrong. But what is wrong about It is exactly that Poirtopps dan (?) did not recognize the distinction between commutations and substitutions in the content„ If fie had been able to recognize that, then he would have seen immediately that from the point of view of the form these four items are one and the same, They are different in Latin, but not in Danish, This is what is notorious in introducing Latin grammar in Danish and I understand that in this country nobody would do such a thing. There may be various


reasons, but 5 think that one of the reasons would be that you know how to distinguish what I call commutations and what I call substitutions« Sn the Sight of what has been said about commutations and substitutions so far, and about form and substance, i shall now have to come back to functions and I shall! be mostly concerned with correlations You will recall that we distinguish the process ' H I ' and the .system when the object under observation is a semiotic. Then we call the process a syntagmet j c and the system a paradigmatic,, When the semiotic under observation is a language in the ordinary linguistic sense, then we caiI the syntagmatic a text, and the paradigmatic a language What is immediately observable, of course, is the process, the syntagmatic,, the text, whereas the whole analysis consists in a catalysis, where the system is encatalyzed, the paradigmatic or the language underlying the text and by which the text has to be explained. So this is, by the way, one kind of explanatory synchronic linguistics. Sn the text we have £i3s±SLS anc! when they are generated in a specific way, we call them units. In the system (that is, linguistically sn the language) we have correspond!ngJy paradigms and when these paradigms are generated in a specific way, we call them categor S es« This is just by way of repetition. The functions within the process are the. relations “-the functions within the system of the correlations. You will also recall that the correlation* units between them, the paradigms and categories may be represented graphically by vertical columns forming a right angle with the chain. Tj;is is to say, this is a paradigm. So that each member of the paradigm is just like a single picture in a moving film strip. Each member can lake a definite place, as we would say, In the chain. Go, so, low, toe, bow, doe, mow, no, row. And in a!I these cases you may also (for Instance),


have, gay, say* lay, bay, day, may, pay, nay, ray0 Thea® would (I thin!) always convey meanings in English« And perhaps there would be more consonants to add« The situation we find in this graphic representation is what one might call the Ideal practical situation, because here ail members of the first category can enter into a relation with all members of a category so that all this would make up units which are actually observed in the text« In this ideal situation, then, paradigm and category coincide« Here w© can see that there is not only a paradigm but also a category, and so do chain and the unit coincide,, We can say that there's not only a chain but also a unit« In other cases, two or more paradigms must be compared and thrown together into one„ and two or more chains must be compared and thrown together into one and the same unit« If instead these representat ions of vowels we would take (ay), then I don't think that we could have all the consonants »« you may have guy, (which I understand is a very common word in this country)« You can have lie; you can have "by, die, my, rye" and so ont, And then you would have to compare this with the first situation« And then in that way one can build up a general category of consonants which are able to be members of that paradigm to take up such a place in the chain, and similarly with the vocalic units« In the content it would be the same thing« We can have the theme go----that was my example before from the expression« Now we take the content unit 30« Together with many other so-called verbs, it can be combined with the present tense or the past tense« This is an analysis of content, not an analysis of tine expression„ jjn the expression there happens to be



for* thi s English word a go and a went« From the point of view of the expression fch®&@ have nothing to do with each other. They are just two members of the same paradigm of tenses and this is a paradigm of the content. The expressions of the various content-units may vary considerably within one and the same language as is actually the case in English« There are some very easy examples (of course) in English like die, died (that*3 a very easy case)„(And there you can tell what is the formant of one morphene and what is the formant of another morpheme. You may remember that a formant is just an expression of a morpheme and a morpheme is an inflectional content unit«) You asav have more difficult situations in the expression of such a morpheme'. • t 2° and went is an extreme case. Write and wrote is already a complicated case. Now, here the formant is expressby a vowel unit in the middle of the root. That*s what is known as an alternation or (in some jargons, "discont?nuous signs" or even "morphemes.")Th© discontinuous, sign expression is (r-t). So (r-t) means scribøre if I may talk Latin, The English expression for sera bere Is (r-t). The vowel between R and T varies and the vowel is not the expression of seribere but is just the expression of the various morphemes added to seribere. This variation of the vocalic unit (the vowel unit) expressing morphemes, may be combined with an ending as in "writes", for instance. An example of a discontinuous inflexional expression would be what we have in German past tense part s c ipIe geschrieben. Of course, we have the same thing here because we have schreSb,3chr >i<eb, etc. This is just like write, wrote. But ge and -en



together are the expression of the discontinuous expression of what I call the morpheme,. Then m the content, as we 11 as sir* the expression, we have suspended commutations under definite conditions, that is what. 1 cal 1 seyncretIsms., There is sometimes in definite English verbs a syncretism between the present tense and the past tense« Witness let and put and so on« Only in the third person singular present indicative, of course, there is a difference because of the ending s« We have also to add a thing which may be important to understand some of the correlations I have gone through som« days ago, namely the fact which is known as defectivism, a defective paradigm,, in Latin we hafce various tenses« We can organize them in a table of three X two, (This is often done by Latin grammarians 8 think and I think they are right«) There Is a distinction, then, between an *nfectum and perfectum« So infectum is scribo and other forms, I write. Whereas perfectum is scrips! and i other forms« But there are certain verbs which only have the perfectum odi *1 hate* mem* r* I *8 remember*. This is the so-called perfectum praesentia or praeteriturn praesentia. So in that case perfectum, the Latin perfect, is specified by infectum« The perfectum can be the only member (can remain as the only member) of the paradigm whereas the »nfectum cannot, A very well known example is the pluralis tantum (the nouns which only have the plural form, not the singular); then, singular specifies plural« There is another well known case of the impersonal verbs, verbs meaning * it rains*,and so on« One cannot say, "I rain", "you rain", and so on« But in ail these cases, the question ■remanus whether this belongs to the linguistic form (the linguistic pattern In the strict sense or whether this belongs just to usage, usage which, to some extent may ever* be governed by external non- I i nguiiatic



There* are reasons why I can *t say,

"I rain, you

rain", but you might say "you rain", if you addressed Jupiter for instancso And Jupiter might say "I rain"" So this might be cases of usage«. But there are other cases where it is definitely a question of the linguistic pattern itself, just like these verb forms S mentioned f s rato Well, Sadies and gentlemen, £ thank this will be ail for today«. If time permits 3 should like to go a I little into a consideration, in my last lecture, of the particular correlations between members of paradigms-. But first and foremost my last lecture shall be devoted to the theory of form and substance«


Hjelmslev Lectures 2-8 4=61

! SixgMhrfc that everybody has a copy of the list of terms and definitions. This in a way provides you with a resume of the whole thing that I have been expounding in these lectures; and even with a few additions, because one cannot easily fill all the formal definitions orally* But in a list like this, St would be natural to give them* So this is why you find, for instance, on the second page, the formal definition of analysis and the formal definition of function, which ! have not given you express IyaOn page 3, there is a definition of catalysis which would not give you any difficulty* But it is Just a formal statement of what I have been expounding, and so are the following definitions of class segments* Some other definitions will be found below such as hierarchy and some other definitions also* On page 3, you find, about the middle of the page, the term "incompatabiIityo" S have not mentioned that at ail in the lectures, but of course it is a very well known term* And it really does not belong to the giossematic theory* Incompatability is the absence of function between objects* That is how we would state it in giossematic terms, if we stated it at ail* So incompatibility is just the negation of any positive function, the absence of a function* So we can state, if we like, an incompeti bi Iity between k and g in English within one and the same consonant cluster, of the syllabic theme. But incompat»biHty is nothing but the negative side of the positive functions which I have given in my list and which I have been giving in my lectures« So there is really nothing new about it, it's just a negation of any function,, On the next page, page 4, there is a definition given, of language and semiotic« I should like to change this a little,:, It is entirely my own fault that it has been stated in these terms« The only thing that I



' # have to change is the definition of semiotic because we usually comprise languages in semiotics, so that a language in the linguistic sense is a kind of semiotic and that is not what is stated here, so that definition ought to be as follows; Language is taken in the ordinary linguistic sense; a hierarchy which is a language, or whose structure is fundamentally similar to that of languages, is called a semiotic.. Now, that leads me to say a few words about languages as a particular kind of semiotic -- I've been talking about that in my lectures, and in one of my lectures I promised to come back to the definition of language:,, Of course what you find on p* 4 of the list of definitions is not a definition of language« It just states that language is taken in the ordinary linguistic sense, but the problem is to know what is the "ordinary linguistic sense" of a "language"; and what is the proper character?stic of a language or the differentia specifics of a language as opposed to other semiotics. This is very simple and this has been found, by the way, independently of glossematic theory by the Polish logician now active in the University of California in Berkley, Professor Tarski,, According to this theory, (which I think would be convincing to everybody) a language in the ordinary linguistic serf.se of that word is a semiotic into which all semiotics may be translated. You can render anything you would like into a linguistic language -- you could not render it in a semiotic which has been, for instance, devised for some particular purpose such as, say a mathematical or logical semiotic, system of symbols, or, say, the traffic signals and so on and so forth. All such systems have a far more limited application. Whereas the linguistic language has the wonderful quality oh being able to form anything. It's a passkey semiotic«. It can be used for everything. To give a more formal definition, one could state


that a language is a semiotic into which ail other* semiotics may be translated* That is to say, a language, in this sense is a paradigmatic whose paradigms are manifested by alf purports« A text, correspondingly, is a syntagmetic whose chains, If expanded »rsdef s nitel y, are manifested by all purports* So all other 8anguages,and all other conceivable semiotic structures, can bus translated into any language in the linguistic sense of the word* Whereas any linguistic language cannot be translated into a semiotic that is not a language., This trans5atabi I ity rests gws the fact that languages and they af®no are in a position to form any purport whatsoever« Sn a language and only in a language you can do what Kierkegaard used to caf I '’Work over the inexpressible until It is expressed,? It is this quality that makes a language usable, as a language, capable of any gsven, even unpredicted situations, giving satisfaction to any unpredicted situations* This, no doubt, rests on a structural peculiarity of languages, on which we might be able to cast better light If we knew more about the specific structure of non- linguistic semiotics* But it seems to be an obvious conclusion that the basts lies in the unlimited possibility of forming signs (for instance, of forming new words) according to the rules accepted in the language, and the relatively free rules for forming units of great extension-'-sentences, and the like* You can form nearly any sentence you would like to according-to certain rules. And this has the consequence that it is possible for a language to permit false, inconsistent;, unpreclse, ugly, and unethical formulations, as well as true, consistent, precise, beautiful and ethical formulation«.. Indeed as the old German linguist Hermann Steinthal has it "even the most crazy nonsense can be expressed in correct language and even in a beautiful sentence-structure„"


o a

"Selbst der tolste Unsinn (asst sich sprachlich {correct und sogar in ff schonen Satzbau ausdrucken«" II would only add that it is not only a question of expressing but also of formulating something, not only In the expression but also in the content«, This really is true« I think S experienced that when once I was a member of a group translating Kant*3 Kritik der Reinen Vernunft into Shawnee« This can be done; it's perhaps, a cumbersome thing to do« But it can be done and Franz Boas has stated that in his introduction to A Handbook of American Indian languages, first volume, that anything can be very well rendered into any American Indian Language-~a so-called "primitive language«," A translation then is a function existing between different semiotics* It's a transposition of one and the same purport into a different semiotic form* And if there are two languages, there is a mutual translatabiJity between them* This translatabiIity, this function, between the two languages may be considered a complementarity or an autonomy, according to the situation. But if we add a semiotic which is not a language, then there is a specification, this way* A non-language can be translated into any of these real languages, but not conversely, On page 5, the next page of the list, you will find the term direction and also the term government.,, which is better known to you, St is stated there that I shall enter upon that in the present lecture. There is a reference to Essais Linguist agues, so ! should like to state in a few words whet this is about, Direction is a kind of cohesion, (that is to 3ay, as you will recall) a function, and (in this case) a relation, where at least one of the functives is a constant« Imagine that we have two clauses which we have found through the analytical procedure« We add a synthetical procedure and we find that there is actually a cohesion between these two clauses*




/o We may fsind a8 so that this cohesion may be narrowed down to a cohesion not between the whole clauses, but between certain parts of them,, These would be inflectional elements,, contracting a government or, as we would say, a direction* Actually we can say, in ordinary linguistic terminology, that the whole sentence is in the subjunctive moodo And we can say that the verb is in the subjunctive mood, in both cases, the subjunctive mood may be dependent on a cohesion with something In the preceeding or the following clause,, In this case, where we have this situations In the synthetic procedure, we will call that a nexl a, We will call each of these entities, a nexus* This is Otto Jespersen*« termino!ogy,but w<* have changed his definitiono* And these small units can be made responsible for this cohesion we call them actually morphemes, (but that may disturb you. You might, call them morphemes, but we would call them, and only them, morphemes) inflectional morphemes and in this case.*extense* Now later in the analysis, we find smaller units where we can find the same thing^-a cohesion which, like this one, generates (can generate) nexsa« And where a smaller unit ini both of these relatively larger units, can be made responsible for this cohesion* |n that case we call these smaller units in the content plane, we call them (we call the whole) a junction^ which also is Jespersen's terminology* We call each of them, what they really are, a noun* And here we have two different nouns contracting a cohesion and mutual government which we call a direction* In the linguistic expression, that which corresponds to the extense morphemes of the content are relations* What corresponds to intense inflectional morphemes are accents, (We are not interested $n knowing how modulations and accents are manitested--whether by 3tres3


o 7 i or- by pitch or whatever it may be*) The units of the expression that correspond to nouns in the content* are syllables in the giossematic sense of this word* Actually* that is what you find on the proceeding page of the list* on p,, 4f"“definitions of independent clauses*, of sentence* modulations* syllables, and accents« So syllable is the expression parallel to noun as a content unitj accent is what in the expression corresponds to intense inflectional morphemes of the content* and there are nexus and next as in the expression as there are naxias and nexus in the content« What corresponds to the «xtens® morphemes of the content* are the modulations in the expression« So there is an absolutely parallel structure in the two planes of the language« This has been called by some authors, the isomorphism between content and expression~~That is not really a term which has been created by us but it has been used as a label by some authors* mostly by the Pole, Kurylowics„ Now* ladies and gentlemen* I shall very briefly enter upon a matter which l mentioned in the last lecture« I promised to tell you what a participation is, but 8 did not do so« But I insisted upon the fact that* as far as correlations are concerned* gIossematics posits a part i c i pati ®n between the correlates« We do that because of a consideration of form and substance at the same time* as we always do when the question is of commutation and substitution, But we arrive at completely formal statements« What I'm saying about this is really outside the theory in the strict sense of this word* 8t is more of a hypothesis, or a system of hypotheses* which are empirical laws, and which have to be verified, of course, I thought it expedient, just to show you this in a very brief presentat £ on,) Consider simple paradigms




of two, within the general category of adjectives, in well-known languages, Consider, for instance, short, long, the correlation between short and long,, between young and old, between small and big, between thin and thick, and between narrow and broad. We can define, approximately and roughly, the semantic design covered by each of these pairs, by calling this length, this one age, this one size, this one thgckness (if you will allow for this) and the last one breadth or width.® We can give a graphic statement (graphic presentation of such an opposition) by drawing a rectangle with two compartments **** a compartment A and another compartment which is the opposite B. So A Ss a slot which might be filled by short, young, small, thin, narrow, j? Is the one which might be filled out by long, old, big, thick, broad. Now in other cases we may need three compartments. There is A and B. And ther© i« C between A and B aowhen A, Itcoid and 8^»hots.C^is warm (lukewarm)* when A is wet and E? Ss dry, then C is humid or moist. One would expect all correlations between such members to be logical exclusions, Young means "not old"£ old means "not young", and so on. And so all the compartments (that is the compartments I've been drawing here) contract mutual exclusion. We call them sub-li ngu i st i c or sub**-1 og* ca I units. But linguistically they are represented by members of a category. These are all short, young, old, warm., thick and so on*-»they are all members of th#Ir categories. Experience shows that linguistically the members of these categories are participations. One member of the category is not only B, not only the contradictory opposite of _A, but can also act as a representative of the entire zone A8-ABC, a function as a representative of the whole zone. Thus,long if you say, "How long is this object?" (And it may actually be a short one) "How long is the road, the table, and so on."

"How old is he"* and he may be quite




young« "How big is the object, how thick is it"even if it Ss small and "feh>on "How broad is the road, the tabled even sf it $© actually narrow« "How warm is it?"-»»even if it is more cold than warm, and so on« So one can expand one of these two or three to the entire zone« One of the members of the category can cover the entire zone, and replace any of the other ones,, The converse is not true« That*® what we call a participation« In an occasional participation, wo call the member of a category which has the faculty of expanding, of covering the entire zone, the extens*ve member, and the opposite member, the intensive« (Now we have worked this out, or I may say, more modestly, I have worked this out, not to make anyone else responsible« (I cannot go into the details, of course today, which is my last lecture«) But we recognize very often two members*e<(alpha) symbolizing the intensive, which covers one of the three and A which has the faculty of covering the entire zone, and which is extensive« And V we can expand this theory so as to allow for categories comprising not only two members, but even three, four, five,and six, and more® If we have more dimensions than one, then we will have more than six members possible« One of my reasons for mentioning this is that it is the entire basis of U!daHe,a glossematic algebra in the second part of Outline of Giossematics« You well see there, if you open that book or you may have seen there, that UIdaII is operating on A, B and C, and stating correlations between them« And that is actually what one can do« I must add now that a syncretism can only take place between an intensive and an extensive mewiber of a category. But not between 2 intensive members or between two extensive members of a category« It would be difficult to give examples of this where there are two members only, so I think I shall leave it at that for now.



and * nvitc you to read about St, The coinmutet ? on test* the use of the commutation teist, (and of course the use of the commutation test os unavoidable for everybody, whether you are an adherent of one or the other theories) the use of the commutation test, then, puts us under the obligation of analyzing the plan© of content as well as that of expression, and of recognizing the form of content as part and parcel of la langue itself^ This is why we proceed to an analysis of the content (l have 9»v*n you a short example showing you the so-called isomorphism between the analysis of the content and the analysis of the expression) And we are famous or notorious (according to your taste)for proceeding to the analysis of the content units expressed by words. So that, for Instance, w© reduce what is comprised in the word ram if we reduce it) to the combination of he and sheep, or male and sheep. And ewe to she / sheep or female / sheep. Stallion is he horse or male horse and mare Is she’horse or female horse, and so on, These are very, very simple examples, But here again there Is an exact parallel with the analysis In the expression., Th s seemingly unanalyzable unit Is exactly parallel to this one. It can be reduced to ft (flow), to ®r» (snow) and so on.

ihese units of the expression are Isomorphic with the unit of the content. In this reduction of content entities into groups: ram is a group with sheep just as f! is a group f plus I, In this reduction, the sign content is equated with a chain of sign contents having certain mutual relations. Definitions in which words are translated in a uni lingual dictionary are, in principle, of this kind, although dictionaries are not able to aim at a reduction, and therefore do not yield definitions that can Immediately be taken over


by & consistently perforated analysis« But that which is established a© equivalent to a given entity, when that entity Sa so reduced, is actually the definition of that entity, formulated in the said language, and in that plane to which the entity itself belongs« In this way we can say, then, that a definition is an analysis of a sign content or of a sign expression« This is the definition of a definition« I promised to say a few words in my last lecture about form and substance« What I would like to insast upon is that the distinction between form and substance is nothing particular to glossematice (St may be if glossematics is compared to other linguistic theories) that it is not when the glossematic method is compared to ordinary scientific methods« Professor Kahne, in one of his earlier works. Der togjsche Aufbau der Welt defines structure in a way which seems to me in accord with the views I have been advocating, namely as purely formal and purely functional« According to Kahne, all the scientific statements must be structural statements, in this sense of the word, that is to say, formal and functional statements,, A scientific statement must always be a statement about relations, what I call functions without involving a knowledge or a description of the relator itself. This use, perhaps, confirms completely the results which have been obtained by glossematics (if we can claim certain results)« It's obvious that the description of language must begin by stating relations between relevant units and these statements cannot involve a statement of the inherent nature or essence or phonic substance of these units themselves« This must be left to phonetics on the one hand, to semantics on the other, which accordingly presuppose the structural analysis of the entire language pattern. One might state then, that linguistics in the glossematic sense of that word is a meta ^i.ngu«igo of the first degree, and phonetics


and semantics ar® metalanguages of the second degree,, The distinction of Form/substanee is due to analysis of the total consisting of form and substance,, The study of substance is a catalysis* So that the form is encataiyzed,., .A study of the substances and the functions on the strength of fact that substance presupposes form* Also that both form and substance are encatalyzes on the strength of the fact that there is isomorphism between substance and form* Some people think that there is only one formal system in a language, I wonder why* I can easily be made to believe that in most cases, in nearly all cases we can prescribe a procedure which would allow the arriving at the formal system only* S think that the principle of simplicity would be a good guide for arriving at such an ultimate statement* (But of course simplicity would be now*) But one has to allow for ambiguous and equivalent analyses, analyses which are equally simple and lead to equally simple results* So I agree with Chao, Yuen-Ren, in his paper on The non°uniqueness of phonemic solutions of phonetic systems. (Tape UNCLEAR;, approximately l page.) In come respects th© view® of the Polo Kurylows cz go to the right. Other® might be mentioned, such as the London school of phonetics (phonetics may also mean phonemics) under the leadership of Danis! Jones,, i should like most to introduce the contricutSon of structural linguistics given by that early and great pioneer toward oapa r, in papers sike the one on "Sound Patterns in Language", and in his book Language (1921) which is to my mind a wonderful and inspiring book, and a real masterpiece, together with essential chapters of Saussure's Cours de H ngul.at i gue. general®, Sapir*s i © perhaps the best book on linguistics (in general* Next comes, of course, if not on an equal level, Leonard Bloomfield*® Language which I would rank with Joseph Vandrye's, 1, Langage.> i have hardly,, ,if at all.


«f , h

* . fc

made any use of the ready-made and too easy I abe! '"American linguistics«." American linguistics would be just as unjustified as "European linguistics,;." Some Europeans may have the impression, now and then, that certain structural descriptions, for instance of American Indian languages, which find their way to periodicals and other publications in this country, are a little too rough in their efforts and have rather an „ « <. (Tap© UNCLEAR - J page) . One reason why such an important work was not known, was, of course, that Sapir's writings were published in a so little known language as English used to be in Eastern Europe, at that time. One often meets the statement that American linguists have the advantage having the American Indian languages as a good object of description and that calls very much for a synchronic approach«, But in Europe we have dialects, many dialects, which have not been described. And as far as languages are concerned both groups, that is to say Americans and Europeans, are still in need of a general theory of linguistics and a procedure which can be applied. The unification of several contrary approaches in all this, is largely due to American hospitality (which) makes cooperation possible across the big ocean., And I would add that the hospitality bestowed upon me this time by the Department of Germanic Languages of The University of Texas is an example which I very highly appreci ate«,