On Some Problems of the Sysstematic Analysis of Grammar*

Vilém Mathesius

It was a proof of the true insight into the actual situation of linguistic research that the organizers of the first international congress of linguists, which met at The Hague in April 1928, designated as one of the fundamental problems of modern linguistics the necessity of establishing a really satisfactory method for the complete analysis of a given language, since the whole trend of the development of linguistic science in the last twenty years has been pointing to a revived interest in questions involved in that problem and, at the same time, has been preparing a way for their solution. A systematic analysis of any language can be achieved only on a strictly synchronic basis and with the aid of analytical comparison, i.e., comparison of languages of different types without any regard to their genetic relations. It is only in this way that we can arrive at a right understanding of the given language as an organic whole, and get a sufficient insight into the real meanings and functions of the single linguistic facts which constitute it. If we are to apply analytical comparison with profit, the only way of approach to different languages as strictly comparable systems is the functional point of view, since general needs of expression and communication, common to all mankind, are the only common denominators to which means of expression and communication, varying from language to language, can reasonably be brought. Linguistic analysis, based on these principles, must aim finally at ascertaining causal interdependences between coexisting facts of the same language. To such analysis modern linguistics (as opposed to the linguistic conception prevalent at the end of the nineteenth and the beginning of the twentieth centuries) offers, as I have repeatedly pointed out, ¹ a good access and in the conception of language as a system of signs also gives a working theoretical basis. The full importance of the new functional and structural conception was first realized, just as it was more than half a century ago in the case of the Neogrammarian conception, in the analysis of the phonic aspect of language, and structural phonology may by now be regarded as well established. An analogous task in the field of the grammar proper still remains to be performed. It is true that since 1900 we have been able to register several important attempts at working up a generally acceptable system of grammar, but even in the case of the really untiring and original efforts of Professor Jespersen the results cannot be regarded as quite satisfactory. A good illustration thereof is the fact that even Jespersen’s last word on the grammatical system of Modern English, his Essentials of English Grammar, in which, in 1933, he assembled a fine selection of results, arrived at in a long and assiduous study of that language, is disappointing as far as a complete and clear analysis of the given grammatical system in all its constituent parts is concerned. There may be several reasons for this relative failure but two of them require special attention in the context of our considerations. Jespersen, in his praiseworthy adherence to what is really accessible to our immediate observation, does not seem to realize enough the possibility of a deeper and more hidden reality, lying at the root of the perplexing variety of outward phenomena, and the necessity that scientific investigation, if it is to arrive at a clear and utilizable system of knowledge, should not refrain from seemingly disarranging reality by theoretical simplification. This, too, maybe in some connection with the uncertainty of what he regards as really essential. The relative importance of a linguistic fact within the grammatical system of a given language can be ascertained only from the point of view of the whole system, that is by considering its real function within the system, and may be set off by a well considered use of foreign comparative material. In Professor Jespersen, who has several times shown such a fine understanding of the functional point of view, and has always known very well how to make profitable use of foreign comparative material in the modern sense of the word, it would be but natural to expect that he would have made an extensive use of these two methodological helps in the book which, according to his own words is “to prepare for an intelligent understanding of the structure of a language which it is supposed that the reader knows already” (Essentials, p. 20). Unfortunately this expectation is not fulfilled, and so it comes that Jespersen’s book is a highly important collection of finely observed and finely described grammatical details, but is not an exposition of a grammatical system.² The question put eight years ago by the organizers of the Hague congress has not yet been answered, and the best method of a clear and complete analysis of the given language is still to be found.

I have been trying to find it with the help of the following considerations. Every communicative act of speech — and on communication as opposed to expression are, on the whole, based the systems of all extant languages — involves, before it comes to the real utterance, two different processes, which, of course, in the mechanized state of human speaking, merge into one with the utterance itself and can only be detected when an obstacle appears which makes their pace unusually slow or even frustrates their completion. By the one, elements are selected from the given reality, concrete or abstract, which fulfil the double condition of having focussed the attention of the prospective speaker and of being able to be expressed by means of the vocabulary existing in the given language; by the other, the linguistic signs representing the selected elements are put into mutual relations so as to constitute an organic whole, a sentence. (In the extreme case, of course, one single word may be selected from the vocabulary at hand as serving the purpose, and may function as a sentence without combining with other words.) If we are to give an organic analysis of a system of means of expression, which is called a language, we can consequently do it very well from the point of view of these two fundamental processes. So we come to two important parts of linguistic investigation, that of the ways and means of calling selected elements of reality by names, and that of the ways and means of organizing these names as applied to an actual situation, into sentences. In each case the starting point of the investigation will be the communicative needs of the speaker, and from this fact two consequences will of necessity follow: the way will lead from speech, as something which is immediately given, to language, as a system having an ideal reality only, and from the functional necessities to the formal means by which they are satisfied.³ We may then rightly call these respective sections of linguistics functional onomatology and functional syntax. Morphology, as dealing with groupings of the means of expression on the basis of formal affinities, cuts across the two fields, for different parts of the same morphological system may have different functions, onomatological and syntactical.

It would be absurd to maintain that all problems falling under the head of functional onomatology and functional syntax respectively are entirely new and have never been discussed before. It would be equally absurd, however, to deny that the functional reformulation of the questions involved can throw much fresh light upon them, and that it puts us into the position of systematically rearranging the known facts and of discovering new ones. A glance at the table of contents in Jespersen’s Essentials of English Grammar suffices to show us how much the exposition of the grammatical system of a given language may suffer in its clearness and completeness from disregard of the difference of the fundamental processes involved in the act of speech and reflected in the system called language. We may safely say that any intelligent reader will be able to draw from Professor Jespersen’s book fairly complete information on the syntactical structure of Modern English and on its morphological system, but that the ways in which Modern English satisfies the onomatological needs of those who speak it will on the whole remain obscure to him, notwithstanding all the excellent comments on word categories and related subjects which are scattered over the greater part of the book. This is to be regretted, for the onomatological aspect of Modern English is as characteristic of the language as its syntactical structure and morphological simplicity.

*

As was said above, functional onomatology deals with the linguistic repertory of names (in the broadest sense of the word) and its application in concrete acts of speech. A look at the recent linguistic literature is enough to show what important problems are involved here. One of the much discussed questions regards the very important and very difficult problem of the signification of word. Professor O. Funke’s essay On the function of naming (English Studies, XVIII, pp. 57 ss.) not only gives an instructive survey of the history of the problem, but has, in addition, the great merit of bringing the results of the respective investigation into harmony with the fundamental conceptions of modern linguistics, for in opposition to A. H. Gardiner, who overemphasizes the meaning, that is, the conceptual or lexical as opposed to contextual sense of the word belonging to language, the author clearly states that the function of naming, that is, of putting words into relation to objective reality, belongs to speech, to the context of a sentence and to the concrete situation, to which the sentence refers. Much less clear is the real linguistic significance of another distinction which has been appearing in onomat- ological research papers in the last years. I refer to the distinction between Meinung and Bedeutung which has been repeatedly expounded by F. Slotty⁴ among others. It cannot be denied that something like that exists, for it is true that the thing-meant (Meinung) is regularly represented in human speech by means of an allusion or reference to some of its aspects, real or imagined (Bedeutung). It is evident that the Bedeutung reflects the subjective conception of the thing-meant by those who were the first to give the respective thing its name. So (to use the instance quoted by Prof. Slotty) the lever in a gunlock, raised ready for release by the trigger, is called cock, because those who were to give it its name were struck by its likeness to the head and the bent neck of the bird of the same name. The difficulty, however, lies in the fact that Prof. Slotty, a specialist in Indoeuropean comparative philology, does not distinguish sufficiently between the diachronic and the synchronic points of view. The act of forming new names, to which Slotty constantly refers, is, if we do not take into account the language of poetry, on the average of very rare occurence only, and in the field of the traditional vocabulary the distinction between Meinung and Bedeutung appears in quite another light. The Meinung of a given word is always extant in the language so long as the word belongs to its living vocabulary and can always be actualized in speech, but the Bedeutung, even if it is synchronically decipherable by common speakers, and that is not the general case, is usually not realized until the attention of the speaker is directed to it. We may then rightly say that, from the synchronic point of view, the Bedeutung, as a distinct aspect of the sense of the given word, has only a rather limited role in language, and that, in addition, this role has the distinct character of potentiality. Subject to these restrictions, it cannot be doubted that a systematic analysis of Bedeutungen which can be realized by common speakers in a given time can throw an instructive light on the character of the vocabulary of the respective language.

Another group of onomatological problems, which has of late repeatedly provoked renewed investigations, centers round the question of word classes. Here again I can refer to Prof. Slotty, who, in the above quoted series of papers, discusses the nature of word classes and in the paper published in the Schrijnen volume also sketches the recent history of the problem. I quite share the opinion formulated anew by Slotty of the double-faced character of the chief word classes, which, on the one side, represent the classification of the phenomena of outer (and, by analogy, also of inner) reality into objects and qualities (constant or passing), and, on the other side, are the bearers of the chief syntactical functions, being the formal types destined to express the subject, the object, the attribute and the predicate respectively. This is, however, a mere introduction to the multitude of special problems which arise here. Not even the establishment of the list of word classes existing as distinct categories in a given language and the investigation of their morphological and onomatological character and of their mutual relations (especially of the possibility of a word passing from one class into another) covers the whole field. Word classes constitute but the fundamental classification, by means of which the speaker of a language is put into the position of mastering his or her vocabulary. The system of classification continues within the word classes themselves, and, as far as I can see, the character and the significance of the secondary categories which are the result of this further classification have as yet attracted a very insufficient measure of attention on the part of linguistic specialists.

In the analysis of these secondary classifications it will be good to keep apart two series of facts. Sometimes we see that from class to class the outward aspect only of the respective meanings changes, the basis of them remaining untouched. Another time, on the contrary, the difference between two classes is such as radically to affect the whole meaning of the word. We may call the first type of class-differences aspect modifications, the other type categorical differences. A few examples will show the correct meaning of these terms. In the substantives, aspect modifications consist chiefly in differences in number (e.g., in German Stein, Steine, Gestein) and totality (e.g., in French un pain, du pain, les pains, des pains), in definitene s s (e.g., in English, man, a man, the man) and in qualificative (as opposed to purely morphological) gender (e.g., in Modern English the difference in emotional qualification between baby referred to as it and as he or she respectively, and the difference in sex qualification as between prince and princess, Englishman and Englishwoman, etc.). In the adjectives aspect modifications are limited to the phenomena of comparison, for number and gender are in them expressions of grammatical concord only, without any independent onomatological function. Modern English is very instructive in this respect. The verbs present a great complication of intermingled functions. The character of aspect modifications is quite evident in the tenses and the moods in their primary (not syntactical) functions and in what is now called verbal aspects. Number and grammatical gender, as far as they find their expression in finite verbal forms, are clearly matters of grammatical concord, and so in my opinion are, at least in Modern English, grammatical persons, for, as is shown by the existence of personal pronouns in languages of the type of Modern English, it is primarily the function of the subject and not of the finite verb to distinguish between them. Voice, active or passive, I regard as the expression of a primarily syntactical function, and find myself in full agreement here with Prof. Jespersen (The Philosophy of Grammar, p. 164).

To anyone who knows something about several languages, the existence of fundamental differences between them in regard of aspect modifications is self-evident. The most important fact is that the same aspect modification may be in one language an essential part of its grammatical system, it being systematically applied and as a result definite formal means being in existence for its expression, while in another language it may appear only as an occasional help to a more precise qualification of the meaning in question. So, e.g., the existence of the article in the Germanic and Romance languages testifies to the category of definite- ness being an essential factor in their grammatical system, whereas the absence of the article in most Slavonic languages points to the fact that this aspect modification of the substantives has an occasional character in them only. The same difference is shown for the category of totality by the existence of the article partitif in French and its absence in English. The real character of the analysed type of aspect modifications is of high importance in the vexed question of verbal aspects in Germanic languages, for the fundamental fact is that in Germanic languages verbal aspects are not an essential part of the grammatical system of the language as they are in Slavonic languages. There, as, e.g., in Czech, which is my mother tongue, each single verb is, through its very formal character, assigned to a special aspect class, whereas in Germanic languages, e.g., in English, the given verb does not formally belong to any special aspect class and can, as a rule, pass from one aspect to another without any formal change. In Czech, therefore, the question is what formal system of verbal aspects the language possesses, and how the system is applied to special needs of concrete contexts. In English, on the contrary, the investigation must be limited to the question what formal means are used if a special verbal aspect is to be clearly expressed in the given context. In the progressive form Modern English, of course, possesses a system of forms expressing a special aspect modification of the given verbal meaning, but its basic function as it appears now is not to express duration or anything like one of the formally determined Slavonic verbal aspects. It is (from the purely synchronic point of view) primarily a means of expressing the actuality of the implied action or process, and durativeness is here (just like descriptiveness or emphasis) only a secondary aspect derived from the primary function. We see it very clearly when putting into contrast to the English sentences I am now going to school and I now go to school the corresponding Czech sentences Ted jdu do školy and Ted chodfm do školy. What matters in Czech is the difference between the action that takes place only once and the action that occurs repeatedly, and this difference, which is a constitutive element in the aspect system of the Czech verb, is expressed so that the single action is taken for normal, and the repeated action is expressed by a special formation of the verb. In English, on the other hand, the implied difference relates to the contrast between the action in its general value, and the action conceived as something actually happening. The general conception of the action is taken here for normal, and the actual happening is expressed by the means of the progressive form. In the terms of functional linguistics it means that in Czech the repeated action is the marked member of the opposition, while in English this character belongs to the action conceived as something actually happening. The opposition duration: non-duration does not come into play here either in Czech or in English.

What is meant in our context by categorical differences is pretty clearly shown by the material gathered by K. F. Sundén in the second of his two essays published in one volume at Uppsala in 1916 (Essay I. The Predicational Categories in English. Essay ?. A Category of Predicational Change in English. Essay II. on pp. 101-562). The author himself defines his task as the examination of primary transitive verbs that in the active form have adopted a secondary passivai meaning, which thus gives rise to active predications of direct object (p. 101). From the strictly synchronical point of view this means that the author investigates English verbs which in addition to their evidently transitive meaning show the possibility of such intransitive meaning as enables substantives (or respective personal pronouns) fit to function as direct object of the verbs in question when used with transitive meaning, to become the grammatical subject of them when they are used with intransitive meaning. Sundén calls this phenomenon predicational change, and he is right so far as the true character of the investigated verbal meaning can be ascertained only in the context of a sentence where the respective verb appears as predicate, but it would be wrong to regard the implied problem as a purely syntactical one. Its close connection with the secondary classification of the onomatologi- cal category of verbs is evident and Sundén himself speaks of the analysed phenomenon as a characteristic feature of the semo- logical structure of English verbs. Taken in its onomatological aspect the problem appears as follows. Transitiveness _and in- transitiveness, as marks of secondary cateogries within the class of verbs, are in some languages strictly kept apart, and a verb cannot pass in these languages from the one to the other without undergoing a formal change and so becoming an independent new item in the onomatological repertory of the language. In Modern English verbs, on the contrary, the categorical difference between transitiveness and intransitivene s s is remarkably weak, and a verb can very easily pass from the one category to the other without any formal change. The conceptual meaning of a Modern English verb must be consequently regarded as much more general and much less concrete than the conceptual meaning of the corresponding verb in a language where the difference between transitivene s s and intransitiveness is strongly felt. So in Czech we may define the conceptual meaning of the verb č¡sti as the verbal expression of the action (or the ability) of converting into the intended words or meaning written or printed symbols, whereas the corresponding Modern English verb to read clearly means only to be in some connection with such action or such ability. This fact also accounts for the ease with which passive predicates are construed in Modern English, which are quite impossible in languages with narrower and more concrete meanings of verbs. That these things must be classed with onomatological facts characteristic of Modern English on the one hand and of Modern Czech on the other hand is proved by the existence of analogical facts in the categories of substantives and adjectives respectively. Categorical differences in these two classes have unfortunately attracted very little attention as yet, and it must be accounted as a great merit of the book of N. Bϕgholm, Engelsk Betydning s lær e (Copenhagen, 1922) that it devotes some chapters to them. If we are to arrive at a complete analysis of a given language the problems of secondary classification of the onomatological repertory, and especially those of categorical differences within single word classes, must be studied systematically.

It may seem from what I have said up till now that I intentionally ignore the existence of such purely formal classifications of single word repertories as are shown, e.g., by the difference between strong and weak verbs in modern Germanic languages. That is, of course, not in the least my intention, but I am of the opinion that a thorough consideration of this type of secondary classification cannot be undertaken until the character and the functions of aspects modifications and secondary categorical differences have been properly understood. The reason is that even in the narrower field of onomatology experience proves that morphological groupings cut across functional categories. If we take, e.g., the morphological types of declension in Modern Czech, we see that those of them which relate to the difference between animate and inanimate refer to a categorical difference (e.g., balíky — packs or balls : balíci — awkward peasants), those relating to sex difference refer to an aspect modification (e.g. pán — gentleman : paní — lady) and others have no significatory function at all (e.g. dvůr — court-yard : zahrada — garden : pole — field). Of the Gzech types of conjugation, too, some are the expression of aspect modifications and others without any ono- matological function. In Modern English the functions of the morphological subdivisions of the chief word classes look a little otherwise, partly because Modern English is a much less inflectional language than Modern Czech, partly because there are prevalent in English other relations between onomatological and syntactical facts than in Czech. This I hope I shall have the opportunity of proving in detail elsewhere.

*

In syntax the central problem is of course the sentence. New attempts at giving a working definition of this principal syntactical unit have been repeatedly made in the last decade⁵ and A. H. Gardiner has in his recent book on Speech and Language (see note 3) put the highly important question of the right place of the sentence in the system of grammar. With an attractively precise dichotomy he assigns the word to the sphere of language and the sentence to the sphere of speech. This clear precision itself is apt to make us distrustful of Mr. Gardiner’s thesis, for the deeper insight we get into the organism of language the more we are persuaded of its complexity and of the impossibility of arriving at such clear-cut statements without distorting objective reality too much. A rapid analysis of the facts in question is enough to show that the instinctive distrust of Mr. Gardiner’s point of view is right. What he really means by the sentence is the sentence as individual utterance, for it is only to the sentence as individual utterance that the qualifications apply with which he tries to expound the nature of the sentence in general. Sentences, he says on page 94, are ad hoc constructions run up for momentary use, and forgotten immediately after, and are characterized by the double relevance to the thing-meant and to some audience or listener. This is all right as far as sentences as individual utterances are concerned. The question however is: Does the sentence belong entirely to the transitory moment and is it as a linguistic entity entirely determined by the individual situation in which it is uttered? The answer to this question depends on whether or not we are willing to regard as a sentence any word or set of words followed by a pause and revealing an intelligible purpose. It can be imagined that a foreigner who does now know enough German would say on seeing a horse running Píerd laufen or that a patient nursed in a hospital for mental diseases would describe the action of the doctor writing a recipe with the words Der Doktor — ein Rezept. In each of these two cases we have a set of words followed by a pause and revealing an intelligible purpose, but are we to regard these sets of words as real sentences? There are linguists who define the sentence from the purely semological point of view in such a broad way that they would answer this question in the affirmative.⁶ From the linguistic point of view, however, these sets of words are in my opinion no real sentences but only pathological sentence substitutes, for there is absent in them the outward form which the respective language has evolved for the construction of sentences. If this is true, the sentence is not entirely the product of a transitory moment, is not entirely determined by the individual situation, and, consequently, does not entirely belong to the sphere of speech, but depends in its general form on the grammatical system of the language in which it is uttered.⁷ This is, as far as I see, the opinion of Mr. Gardiner himself, for he explicitly states that syntactical forms and intonation belong to language (pp. 159 ss.). Having, however, constantly before his eyes the sentence as a concrete utterance, he does not draw the necessary conclusion from his right observation that the sentence as an abstract pattern must be classed with syntactical forms and be consequently regarded as belonging to the field of language. Combining then what has just been said here with what we cited above from Professor Funke’s paper on the function of naming, we can say that in language we have the word in its conceptual meaning and the sentence as abstract pattern, whereas in speech we have the word as referring to concrete reality and the sentence as concrete utterance.

With this conclusion we have not yet arrived at the end of what can be said of the right place of the sentence in the grammatical system. It goes without arguing that there is hardly any language with only one sentence pattern, but at a closer inspection we see that there is much uncertainty in detail as to what should be regarded as existing sentence pattern in a given language and what not. The superstition that the sentence must contain a finite verb has been done with, but verbless sentence patterns are still looked upon superciliously (see Professor Jes- persen’s terms amorphous sentences in his Essentials) and do not get such systematic attention as they deserve. Making use of analytic comparison we very easily arrive at two important conclusions regarding the role of verbless sentences in modern European languages. In some of them the verbless sentence pattern is a basic sentence pattern, that is sentence type which is an essential part of the grammatical system. It cannot be generally replaced by another type and can therefore never totally disappear. So it is, e.g., in Russian where verbless sentence patterns are the normal way of expressing certain types of qualificative and possessive predications. In other languages, English and Czech among them, verbless sentences do not belong to the basic sentence patterns, but are what might be called occasional sentence types only. They can always be replaced by verbal sentences and are completely absent from certain types of style. This, however, should not lead us to the opinion that they at least are mere ad hoc constructions and have nothing to do with the sphere of language. An attentive comparison of English and Czech shows that the contrary is true, for in each of these two languages there are verbless sentence types which are quite inimitable in the other. As two of the verbless sentence types characteristic of English in contrast to Czech the following ones may be quoted: the analytical type with postponed subject (An excellent idea, this. Much use, that) and the synthetic type expressing the central idea which, in the conception of the speaker, is at the base of the situation (“Where are you going?” “Vipers, I think.” — He comes into the room and meets the eye of Valentine, who wants him to go. “All right, sir. Only the tea-things, sir.” Taking the tray.) We see, then, that even for the occasional types of sentence structure languages evolve their own forms and from this fact we must draw the right conclusion that even these types belong to the repertory of sentence patterns of the respective language.

Notes

1. New Currents and Tendencies in Linguistic Research. (MNHMA, Prague, 1927, p. 188 — 203.)

On Linguistic Characterology. With illustrations from Modern English. (Actes du Premier Congrès International de Linguistes. Leiden s. a., p. 56— 63.)

La place de la linguistique fonctionelle et structurale dans le développement général des études linguistiques. (Casopis pro moderni filologii. Vol. XVIII. Prague, 1932, p. 1—7.)

2. This is true notwithstanding the fact that Jespersen supplemented his Essentials with a theoretical commentary published under the title The System of Grammar separately in 1933, and reprinted in the same year in his Linguistica, pp. 304 — 345.

3. The fundamental difference between speech and language has been recently expounded with an advantageous lucidity by Alan H. Gardiner in his book The Theory of Speech and Language (Oxford, 19?2). For criticism of some details of Mr. Gardiner’s exposition see the third part of this paper.

4. Das Wesen der Wortart. (Donum Natalicum Schrijnen. Nimègue 1929, p. 130 ss.)

Wortart und Wortsinn. (Travaux du Cercle Linguistique de Prague I, Prague, 1929, p. 93 ss.)

5. Good surveys of them can be found in the books of John Ries (Was ist ein Satz. Prague, 1931) and Eugen Seidel (Ge- schichte und Kritik der wichtigsten Satzdefinitionen. Jena, 1935).

6. So, e.g., VI. Skalička, who defines the sentence simply as an elementary semiological reaction. (Slovo a slovesnost. Vol. I. Prague, 1935. p. 212 — 215.)

7. As a consequence it must be required that a linguistic definition of a sentence should contain a reference to the conventional character of the sentence form. My own definition explained in detail in Časopis pro moderni filologii, vol. X. (Prague, I924), pp. 1 — 6, runs as follows: The sentence is an elementary speech utterance, through which the speaker (or writer) reacts to some reality, concrete or abstract, and which in its formal character appears to realize grammatical possibilities of the respective language and to be subjectively, that is, from the point of view of the speaker (or writer), complete. A systematic analysis of the repertory of sentence patterns, undertaken on this basis, will have to pay constant attention to the important difference between the spoken and the written form of language, for the latter very often evolves special sentence patterns or, at least, special functions of the existing patterns.

*From Travaux du Cercle Linguistique de Prague, VI, 95-107 (1936).