Actuality, Potentiality, Conventionality: Some Problematic Issues in Mathesius' Theory of the Sentence

Résumé (Fr): Cette étude analyse un aspect particulier de la dé finition que Mathesius donne de la phrase lorsque celui-ci insiste notamment sur sa dimension « coutumière » (habituelle, usuelle) ou « conventionnelle ». Il s’agit d’un asp ect problématique et loin d’être secondaire dans la réflexion du linguiste bohémien, pas facile ment réconciliable avec d’autres moments de sa pensée, surtout avec la théorie de la potentiali té et la théorie de l’actualisation (tout particulièrement selon la forme dans laquelle elle apparaît dans l’articulation actuelle de phrase). L’évolution du concept de « coutume » linguistique (ou de « conventionalité » linguistique) est saisie à travers les œuvres de Mathesius en relatio n u développement d’autres composantes de sa définition de phrase et, plus généralement, de s a théorie linguistique. L’intégration non entièrement résolue des différentes parties de la t héorie montre toute la difficulté à concilier langue et locuteurs, régularité et variation, à laq uelle de nombreux linguistes ont été confrontés dans les premières décennies du XX e siècle ; une difficulté qui demeure aujourd’hui en core irrésolue. Son examen peut contribuer à mieux compr endre l’évolution de la pensée du maître bohémien dans le milieu scientifique pragois et eur opéen au début du XXe siècle.


Introduction
The aim of this paper is to focus the attention on what seems a problematic point in Mathesius' theory of the sentence, i. e. the definition of the "customary" (or habitual, usual) nature of sentential structures. This topic is related to other aspects of Mathesius' general view of language. Its examination could contribute to understanding the development of the Czech scholar's thought in the Praguian and the European scientific contexts of the first decades of last century. My impression is that Mathesius could not reconcile the "customary" / "conventional" condition which he imposed on the sentence with other parts of his reflection on language, especially the theory of potentiality and the theory of activity, in particular the view of the actual sentence perspective. As is well known, the customary nature of the sentence has been a major theme of the modern syntactic theory. It has always been more or less overtly implied in the Twentieth century debate on the place of the sentence between langue and parole or competence and performance. Mathesius' struggle to integrate the various aspects of his reflection into a consistent general theory was no individual destiny but an interesting experience in the difficult task to reconcile language and speakers, constant regularities and variation which engaged many scholars throughout the Twentieth century.

The use of the notion of "customary" in the definition of the sentence
Mathesius attached considerable importance to the definition of the sentence, which he considered the keystone of general linguistics "for according as the sentence is defined all syntax and sometimes even all linguistics may be said to acquire their shape" 1 . In the article "Několik slov o podstatě věty" ["Considerations on the nature of the sentence"] (MATHESIUS, 1923), he criticises both Wundt's and Paul's definitions, in which he sees two limitations: (i) the recourse to a purely psychological description of the processes that originate the sentence, and (ii) an exclusive consideration of sentences with articulation in multiple components, and the exclusion of the so-called "thetic" or elliptical expressions 2 .
In his view, the psychological approach must be replaced by a functional view that aims at understanding the essential nature of the sentence. He considers three components as fundamental for a definition: (i) the sentence communicative (and expressive) function; (ii) the fact that the sentence is the result of the speaker's will to communicate and of his effective activity; (iii) the conformity of the linguistic forms used by the speaker to the "customary" (habitual, usual) structures. To these he adds another factor, which he considers less fundamental but nonetheless worth mentioning: the fact that a "normal" sentence always leaves in the hearer the impression of being complete, an impression which is mainly due to intonation: "The sentence is a communicative elemental utterance by which the speaker approaches a given reality actively and in such a way to give an impression of being customary and subjectively complete from the formal point of view" 3 It seems opportune to try to understand what Mathesius means by "customary", in view of a general appraisal of his theory of the sentence. The example he gives to clarify this property is a Czech irregular sentence like Běžet kůň, with the infinitive form of the verb, instead of the expected "predicative form" (i. e. an inflected form with Mood, Tense and Person agreement). He observes: "The two properties of the function of the sentence [i. e. the communicative aim and the speaker's effective activity] are not sufficient conditions to the sentence definition. If a foreigner with a very strong will to communicate and with the most energetic act of attribution of a property to a subject says: Běžet kůň he would not form a Czech sentence" 4 Clearly, the example refers to a kind of irregularity that no native speaker of Czech (or of any Indo-European language) would produce, in that -as Mathesius himself observes -"in the Indo-European languages the finite verb came to develop a peculiar predicative form" 5 , both in the thetic utterances and in those with the predication of a theme 6 . But it is the very notion of 'being customary' (habitual, usual) that must be questioned here.
In the same article Mathesius considers the analysis of the sentence as a decomposition in elements that can be denominated, and the synthesis as the correlation of these elements that takes place in the act of sentence formation. He then adds: "Both denomination and correlation must take place in a conventional [= konvenční 'conventional, conventionary'], that is customary (habitual, usual) way [= obvyklý 'usual, customary, in use, ordinary, general, commonplace'] for the language considered" 7 Mathesius' use of the notion of "conventionality" may have been influenced by Marty's view of the language as a system of conventional signs, as most probably was his distinction of decomposition and correlation 8 . At the end of a paper written in 1927 Mathesius explicitly mentions Marty in a context that seems to further enlighten the complexity of his thought. After criticising Wundt for regarding spontaneous expression as the sole basis of language, he observes: "If… language, instead of being simply a result of reflex processes, is, as has been held, in opposition to Wundt by W. Marty (…), a system of conventional signs, then psychology cannot be expected to afford an easy and direct help to linguistics" 9 . In this passage the conventionality issue is directly related to his anti-psychologistic view of the study of language: "Consequently there is no chance of linguistics ever becoming a mere branch of psychology" 10 . Yet he admits that "modern linguistics with its activistic conception of language will have an intensely psychological attitude towards linguistic problems so far as it will always hear or see the speaker or the writer behind the linguistic material" 11 , a statement that preludes to the wellknown passage in the article on "Functional linguistics": "The new linguistics conceives language as something living, underneath the words it sees the speaker or the writer from whose communicative intention they have resulted" 12 .
What seems particularly interesting in the passage quoted above, however, is the fact that the term "conventional" (konvenční) is paraphrased with "customary (habitual, usual)" (obvyklý). To my knowledge, this is the only place in Mathesius' work where the two terms appear together as synonyms. The paraphrase might betray a difficulty in finding a more precise definition of the sentence property.
The definition of the sentence as a "communicative utterance through which the speaker reacts to some reality or several items of the reality in a manner that appears to be formally customary and subjectively complete" is reaffirmed by Mathesius in the above mentioned paper on "Functional Linguistics" which was published in 1929, a manifesto of his linguistic research in which the influence of the recently-born Prague Circle is especially felt 13 . In this work some of the author's older ideas are reformulated within a new broader perspective.
But how important is conventionality (or consuetudinarity) for Mathesius? The weight he assigns to the various components of the nature of the sentence seems to show some uncertainties. In the article on "Functional linguistics" he reaffirms the primary role of the speaker's active attitude, which he had already expressed in the sentence definition presented in 1923: "what makes a sentence a sentence is the active attitude of the speaker to its content" 14 . Yet after observing that "the normal Indo-European sentence has both the subject and the predicate, the predicate being expressed by the finite form of the verb", he also states: "this very fact shows, however, that in each language the sentence has its definite form or forms and that without this form there is no sentence" 15 . This is a stronger and neater formulation without a parallel in the 1923 article, which gives another possible piece of evidence of Mathesius' new scientific orientations 16 .
The shift is even more evident in the article "On some problems of the systematic analysis of grammar" , where he stresses the importance of keeping the perspective of the whole system in language description 17 . Although he admits that "the best method of a clear and complete analysis of the given language is still to be found" 18 , he assumes the communicative needs of the speaker as a starting point of the analysis and suggests a procedure that "leads from speech, as something which is immediately given, to language as a system having an ideal reality only" 19 . He then rejects Gardiner's assignment of the sentence to the sphere of speech 20 . His discussion of the sentence definition has now increased in clarity and stringency. He hits the point when he poses the fundamental question: "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?" 21 . The answer, in his opinion, "depends on whether or not we are willing to regard as a 13 The reference to the passage quoted above is MATHESIUS (1929: 124). 14 MATHESIUS (1929: 124); see also MATHESIUS (1975: 79). 15 MATHESIUS (1929: 124-125); see also MATHESIUS (1975: 80): "an elementary utterance through which the speaker reacts to some reality is not a sentence unless the utterance has a form in common in the particular language", emphasis is mine). 16 A shift towards structuralist ideas is also evident in Mathesius especially in his article on "Functional linguistics" (1929: 129), where he resorts to the notion of phonological system, and is still more palpable in his later works (see for example MATHESIUS 1936). 17 MATHESIUS (1936: 307). 18 MATHESIUS (1936: 308). 19 Mathesius (1936: 308). 20 See MATHESIUS (1936: 316-317). 21 MATHESIUS (1936: 316). sentence any word or set of words followed by a pause and revealing an intelligible purpose" 22 .
Most enlightening is the ensuing criticism of Skalička's definition of the sentence as "an elementary semiological reaction". In the posthumous work A Functional Analysis of Present-Day English he adds that Skalička's definition "does not remain within the bounds of language… [it] is not essentially incorrect, but defines something much broader than the sentence of a language" 23 . Such a view would lead to consider that utterances like German Pferd laufen, or Der Doktor, ein Rezept are also sentences. The two utterances are not, of course, of the same structural type, and it may not be irrelevant here to pay attention to their different kind of irregularity (lack of agreement in the first case, lack of predicate expressed as a verb in the second) and to the different types of speaker that Mathesius associates to each of them (a foreigner, in the first case, as for Běžet kůň, a patient in a hospital for mental disease in the second). Mathesius does not consider these as real sentences "but only [as] pathological sentences substitutes, for there is absent in them the outward form which the respective language has evolved for the construction of the sentences" 24 . On this base he concludes that the sentence "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" 25 . Even more important in a theoretical perspective are the conclusions that "the sentence as an abstract pattern must be classed with syntactical forms and be consequently regarded as belonging to the field of language" and 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" 26 .
Clearly Mathesius' thought on the definition of the sentence had considerably evolved at this point and a contribution may have come from the Prague Circle's debate. It is perhaps no coincidence that the new and more lucid exposition of the problem follows Artymovyč' (1935) distinction of the sentence as a potential and as a concrete realization, the first belonging to langue, the second to parole 27 .

The consistency of the sentence "customary" property with other parts of Mathesius' theory
The condition of the customary (habitual, usual) nature of both denomination and correlation expressed in the definition of the 1923 and 1929 papers was not easily integrated to other aspects of Mathesius' thought seen in their development in time: the notion of potentiality, the relationship of language/linguistics and style/stylistics, the relationship of formal (mechanical) and actual forces in the sentence. On re-reading Mathesius' article from 1911 a century after one can perceive his struggle to systematise the various parts of the theory.
The emphasis "on the static oscillation of the speech of an individual" as "the manner in which the potentiality of language phenomena is actually manifested" 28 was a core motif of Mathesius' thought in his early work, especially in his seminal paper on the potentiality of the phenomena of language, whose aim is "to prove that static oscillation is, in many respects, an important feature of language phenomena, and that the recognition of this fact may be of some help in solving a number of important linguistic problems" 29 . Here the notion of 'potentiality' is opposed to that of 'constancy', i. e. a potential phenomenon is one that is not constant 30 . The naïve ideas of language as an objective fact and of its constancy in space and time are criticised as a distinctive feature of Nineteenth century linguistics. In Mathesius' view, potentiality was related to the structural variability that is inherent in the speech of every speaking individual. In contrast to the Nineteenth century tradition which had isolated speech from speakers, the Czech scholar identifies multiple routes in the European research of the end of Nineteenth and the beginning of Twentieth century that converged in considering the study of language as necessarily based on the real linguistic productions of the speakers and their fundamental variability.
The relationship between linguistics and stylistics (rhetoric) was another of Mathesius' constant preoccupations, which followed as a consequence from his emphasis on the static oscillation of the individual speech and his awareness that a proper scientific domain should be defined for its study. This relationship is already at the centre of Mathesius' reflection in the work on the potentiality of the phenomena of language. The Czech scholar had a full awareness of the difficulties that linguistics had to face. His representation of the irreconcilable dimensions of the phenomena pertaining to the individual speech and to the broader language community is characteristically distinct from the one that emerges from the notes of de Saussure's Cours: "Linguistics proceeds from the concrete utterances of an individual to his speech habits, to his speech and finally to dialect and language, i. e. to language usage existing in a narrower or wider language community" 31 . In a theoretical perspective the study of language covers the whole spectrum of the phenomena occurring in the utterances of all individual speakers who belong to "the same broad language community, called a nation" 32 . Mathesius, however, was well aware of the impossibility for linguistics to fulfil this theoretical task: the 28 MATHESIUS (1911: 2). On the notion of 'potentiality' and its relation to the notions of 'fluctuation' and 'variability' see the contributions in RADIMSKÝ (2007a), especially RADIMSKÝ (2007b), LAZARD (2007), FEUILLARD (2007). Also relevant to understanding the polarity of 'potentiality' and 'constancy' (or 'regularity') are the contributions in RADIMSKÝ (2010a) on the notions of "centre" and "periphery"; as has been observed by Radimský (2010b: 8), "le modèle centre-périphérie s'intègre à plusieurs théories formulées par les linguistes pragois afin de décrire le caractère asymmétrique et irrégulier du système linguistique -comme la potentialité ou synchronie dynamique [ science of language can never do justice to it, "not only on account of the astonishing richness of language phenomena in general, but mainly in view of the fact that [the linguistic] community -especially a culturally highly active onewitnesses the rise of new, even if transient, language phenomena day by day" 33 .
The analysis of individual speech with all the wealth of its phenomena must not be relegated to stylistics. It is the best way for getting to know the linguistic uses of a language community and as such it can bridge the gap between static linguistics, which investigates the language materials used by a language community at a given time, and dynamic linguistics, which explores the historical changes that have occurred in the language community across time. No less important is the envisaged link between the analysis of individual speech and the notion of potentiality: the examination of individual speech "may reveal the full extent of the potentiality of the concerned language" 34 .
This theory also has implications for the definitions of linguistics and stylistics, two disciplines that "differ in their aims, not in their materials" 35 . The first deals with the speech of the individuals to determine the language materials used in the language community, the latter investigates "concrete literary works in order to find out how the given language materials were used in making an individual work of art" 36 . The difference between the two disciplines, however, resides in the fact that stylistics only concerns single individuals (or several individuals, if the scientific target is the study of their stylistic relationships), but can never aim at the social community in its entirety 37 .
The complex relationship of linguistics and stylistics is further enlightened by other passages of the 1911 article by Mathesius. Of special interest is the definition of "styles of speech", i. e. those phenomena of language "whose examination may resemble stylistic analysis" 38 . What Mathesius means by "styles of speech" is "the fact that specimens of actual speech possessing analogous character or analogous aims display some common features in different speakers of the language" 39 . In the interpretation of these phenomena the notion of potentiality emerges again as crucial, because in combination with the continuous mixing of the social dialects existing in a community it concurs in shaping the linguistic tendencies that are shared by different speakers 40 .
Mathesius gives a few examples of the manifestation of the styles of speech in phonetics, morphology and syntax, like the use of ye in both nominative and accusative function in Shakespeare, a use that is unknown to Bacon, the occurrence in both authors of the suffixes -er, -est and the adverbs more, most in the same structure, the frequent occurrence of inasmuch as in Bacon, and the rarity of this adverb in Shakespeare. The discussion of these examples is very succinct and may also give the impression of a reflection that is still in progress: "For this reason, from its very beginnings linguistic analysis has almost invariably concentrated on the main outlines of languages, the more so that such outlines usually prove to be more accessible to primitive methods of analysis… As a result of this, the seeming simplicity of language phenomena is not infrequently regarded not as a consequence of the employed method, but as an actual quality of the examined phenomena, and this often leads to regrettable errors. The very development of linguistics thus reveals that linguists should not only try to discover regularities as general as possible but also fight, even more intensely, against the excessive, mechanical simplification of language phenomena" 41 In the overview of currents and tendencies in linguistic research published in 1927, Mathesius comes back to the problem of the relationship between what is "general" and what is "individual" in language: "Finer methods of linguistic analysis have brought to light the importance of what I should call the double-faced character of linguistic phenomena. It consists in a continuous fluctuation between the general and the individual… The communicative character of language is made possible by the conventionality of the linguistic means of expression. The need, however, of making oneself understood has in reality not been able entirely to suppress the need of selfexpression, and so it comes about that in linguistic research we cannot limit our character of language is the fact that the individual needs of expression can never be fully satisfied with the existing linguistic means. Each individual's experience is unique, and nevertheless is to be expressed by conventional means. This incongruity is, at least partially, removed by the constant adaptation of linguistic means to freshly arising needs of expression. New forms of expression are created on the model of the existing means of expression or the meaning of the old form is changed. Thus the possibilities of linguistic expression are continuously enlarged by individual efforts, which lead either to passing deviations or, if the novelty finds a ready acceptance in the linguistic community to a permanent change" 42 The conclusion is that "linguistic research work can either concentrate on what has already become a common possession of all members of the linguistic community or it can study the individual efforts of linguistic creation" 43 . Mathesius then sketches the major opposite directions of research of the previous decades: "The traditional school of linguistics has so exclusively limited itself to the study of commonly accepted means of expression that the individual speaker has disappeared from its ken" 44 . After criticising Vossler's and Croce's subjectivism as being "too general to make it clear what really its contributions to the solution of concrete linguistic problems will be" 45 , he considers Spitzer's approach to the stylistic point of view: "The proposition maintained by Professor Spitzer "nihil est in syntax quod non fuerit in stylo" very clearly shows how the greatest stress is laid by him and his friends on the individual share in linguistic expression" 46 .
Mathesius maintains that "linguistics as a whole can derive from stylistic syntax and stylistic semasiology a double benefit". The good points he sees in the contribution of the stylistic disciplines to linguistics concern two issues of remarkable interest for our discussion. The first resumes a belief he had already expressed in his 1911 article on the potentiality of linguistic phenomena: "It is good that the rule, often neglected, has been emphasized again, that linguistic analysis should always be based upon words and sentences which have actually been spoken or written, and not upon construed examples only" 47 . The second deals with the problem of everyday clichés: "It is good that the attention of linguists has been called to the fact that linguistic material does not consist in everyday clichés merely" 48 .
The two points are obviously strictly interrelated. However, Mathesius also thinks that the relationship between the study of language and the study of style should be clarified as a prerequisite for the first to benefit from the two advantages brought about by the latter. Here he expresses the opinion, which he had already presented in the 1911 paper, that the two scientific branches do not differ in their materials of investigation but in their aims of analysis. As a matter of fact, "words and sentences which have actually been used by individual speakers or writers, make up the basis of investigation in both cases" 49 . However, in the following conclusions we can see a somewhat new formulation with respect to that expressed in 1911, with a stronger and clearer emphasis on the importance of those aspects of linguistic potentiality that pertain to the whole community: "In the study of language, of course, individual utterances are analysed as specimens of the linguistic possibilities of a whole community, whereas in the study of style we try to ascertain how the linguistic possibilities common to the whole community have been made use of in a special case for an individual purpose… Linguistic analysis accordingly, always concentrates on what is common or may become common to the whole community; stylistic analysis on the other hand is concerned with what is individual and unique" 50 This formulation is interestingly different from the one found in the 1911 article, where Mathesius contemplated the theoretical possibility for linguistics to cover "all the phenomena of language that occur in concrete utterances" 51 , although he admitted the practical impossibility to put it into practice. Now instead the domain of linguistics is one-sidedly defined. In this change once again it seems possible to see a departure from the individualistic trend which was still visible in the 1911 article and the influence of the structural climate of the Twenties, with its more pronounced stress on the relevance of linguistic community. This theme had stirred a considerable debate, especially after the publication of de Saussure's Cours (think, for example, of Jespersen's criticism of the langue / parole opposition in his review of the Cours) 52 . Within the Prague Circle it was an important leit-motiv, although it seems to have been expressed with different formulations and overtones by the Czech and the Russian scholars. In the quotation reported above, however, the expression "what is common or may become common" seems to testify that -whatever new influence Mathesius had been exposed to after 1911 -he had not given up his idea of the importance of potential phenomena.
Another problematic issue with respect to the sentence "customary" aspect is the relationship of the grammatical (or mechanical) and the actual forces in the sentence. It is interesting that the grammatical factor is defined as "a mechanising tendency towards customary placement of the same sentence parts in the same sentence positions" 53 . On the other hand, the force of actuality is related to the act of speaking and to concrete utterances 54 : "it is the actual process preparatory to an actual utterance" 55 . By implication it belongs to the domain of what is nonconstant, non-fixed, but flexible. Its effects are well seen in the word-order patterns of languages like Czech: "Actual word-order patterns are not due to the absolute mechanical prevalence of a single factor, but to the interaction of several factors" 56 . Does this mean that the force of actuality does not pertain to language but to speech? Is it a process that has to be investigated by stylistics? Surely this conclusion could be hardly satisfying if we consider Mathesius' theory of "Actual Sentence Perspective" and its developments in the Prague School.
A final question which is relevant to our examination concerns correctness. Mathesius discusses Ertl's position about the criterion of language correctness for contemporary language. He agrees that it can only be found in the consensus of good writers, but with a qualification: "The consensus of good writers can make a choice only between existing possibilities, but fails where entirely new expressions are being formed in order to meet communication needs that have not yet arisen" 57 . The Czech scholar then argues in favour of the need to find a balance between language flexibility and stability, "since needless vacillation of language use makes the awkward impression of arbitrariness and lack of discipline". He maintains that "safety and certainty in the use of language can result only from unimpaired stability and in this respect the consensus of good authors […] is a better guarantee than the dictates of linguistic authority, for every linguistic authority is only temporary" 58 .
To Ertl's notion of the "spirit of language" (which he considers vague) as a fundamental factor for correctness, Mathesius opposes that of the "character of language": "The specific character of language which has passed into linguistic instinct is the highest guardian of language correctness. Where there is real language culture, it is a shared feature of people with true education and in those who are linguistically creative this criterion of language 53 MATHESIUS (1929: 127). 54 See MATHESIUS (1929: 123 and 131-132). 55 MATHESIUS (1929:132). 56 MATHESIUS (1929:126). 57 MATHESIUS (1929: 136-137). 58 MATHESIUS (1929: 138). correctness has become internalized. At this stage the requirement of correct language merges with that of good style. For in practice they are the same" 59

Conclusions
In the article "On some problems of the systematic analysis of grammar", where he seems to have reached a more systematic integration of the various parts of his theory, Mathesius preserved the open-mindedness that was a hallmark of his scientific personality. At the end of the paper he quoted a few verbless structures of spoken English as examples of patterns that have regularity and belong to the language repertory 60 . But he was aware of the riches of the linguistic phenomena, which can be even more complicated, and in fact the developments of the studies on spontaneous spoken language during the Twentieth century have shown that there can be more things between heaven and earth. Mathesius was keen to find a general model, but not at all costs, because -as he says in the same paper -"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 clear-cut statements without distorting objective reality too much" 61 .