European Linguistics since Saussure
Abstract and Keywords
The chapter firstly deals with Saussure, then with the scholars most directly influenced by Saussure’s linguistic thought, the schools of Geneva, Prague, and Copenhagen, and finally with other European linguists essentially independent from it, but who are nevertheless labeled as “structuralist”: the Frenchmen Guillaume and Tesnière, and the London school. Saussure’s views are summarized by means of his four classical “dichotomies” (langue vs. parole, synchrony vs. diachrony, signifiant vs. signifié, associative vs. syntagmatic relations). They were differently developed by Saussure’s followers: the Geneva school (Bally, at least) and the Prague school choose a functionalist approach. The Prague school also aimed at overcoming the synchrony/diachrony dichotomy, especially in the domain of phonology, where its most important contributions lie (by Trubetzkoy, Jakobson, and Martinet). On the other hand, the Copenhagen school (whose leader was Hjelmslev) adopted a strictly formal approach. Guillaume’s and Tesnière’s approaches differ from those of the just mentioned schools by focusing on syntax rather than on morphology and phonology. The London school (especially its leader, Firth) worked out a particular approach to phonology (“prosodic phonology”).
20.1 Saussure and the Cours de Linguistique Générale
Despite Saussure's undeniable links with nineteenth-century linguistics (on which see Koerner 1973, Allan 2010a), the appearance of his posthumous Cours de linguistique générale (Saussure 1962 [1916/1922]) undoubtedly represented a radical change within the discipline, which exerted a powerful influence on subsequent scholars, especially within Europe (but, to a lesser extent, also in the United States). The main lines of Saussure's linguistic thought will therefore be discussed first in what follows. Subsequently, an overview of the several European linguistic schools more or less directly connected to it will be traced. This connection is especially close for the schools of continental linguistic structuralism, namely those of Geneva, Prague, and Copenhagen (see §20.2); it is somewhat looser for other scholars, such as the Frenchmen Guillaume and Tesnière, or the British linguists belonging to the so-called ‘London school’ (see §20.3). As will be seen, however, none of these scholars can avoid referring to some of Saussure's ideas and proposals: this is the reason for dealing with them under the same heading.
20.1.1 Saussure's Life and Work
Ferdinand de Saussure (1857–1913), born and educated in Geneva, subsequently studied historical-comparative linguistics at the University of Leipzig, which in that epoch (the end of the 1870s) was the centre of the Neogrammarian school. He obtained his Ph.D in 1880, but his scientific activity had already started a couple of years earlier, with a book (p. 470) devoted to the Indo-European vowel system (de Saussure 1879, actually published in 1878). It was analysed in a way so innovative that the book came to be fully appreciated by the scientific community only much later. After ten years of teaching in Paris, in 1891 Saussure became professor for Sanskrit and Indo-European Languages at the University of Geneva, where he remained until his death. During his Geneva years, he became more and more uncertain about his ideas and his results, as is witnessed by the fact that his publications were increasingly rare. His fundamental aim was a reconsideration of the methods and the goals of linguistics: ‘montrer au linguiste ce qu'il fait’ (‘to show to the linguist what he is doing’), as he wrote in a letter of 4 January 1894 to his colleague and former student Antoine Meillet (1866–1936). Saussure first presented his ideas on such matters during three courses in general linguistics which he gave in the academic years 1906–7, 1908–9, and 1910–11. After his death, two of his former students, Charles Bally (1865–1947) and Albert Sechehaye (1870–1946), edited a book which they named Cours de linguistique générale ‘Course in General Linguistics.’ It first appeared in 1916, and the final edition in 1922. Since Bally and Sechehaye had never attended Saussure's courses in general linguistics, they based the text on the notes taken by other people: hence, their work was unavoidably arbitrary to some extent. In the second half of the twentieth century, an essential means for the knowledge of Saussure's authentic thought came from the work of two Swiss scholars, Robert Godel (1902–1984) and Rudolf Engler (1930–2003). Godel 1957 discovered the original notes from Saussure's lectures which had been the sources of Saussure 1922; Engler (1967–74) published such sources together with the text edited by Bally and Sechehaye. Starting from Godel's and Engler's work, Tullio De Mauro (b. 1932) added to his Italian translation of Saussure 1922 an extensive commentary which, since 1972, has also accompanied the original French version of the volume. A reader interested in deepening her/his knowledge of Saussure's thought should therefore refer to it.
20.1.2 The Saussurean Dichotomies
It is standard to summarize Saussure's thought by resorting to his four ‘dichotomies,’ i.e. four pairs of concepts opposed to each other: (1) langue ‘language’ vs parole ‘speaking’; (2) synchrony vs diachrony; (3) signifiant ‘signifier’ vs signifié ‘signified’; (4) syntagmatic vs ‘associative’ (later called ‘paradigmatic’) relations. This kind of exposition will also be adopted here, but Saussure's theories are much more complex than they appear from such a simplistic presentation. It has been maintained that traces of each of these four dichotomies can be found in the work of earlier linguists, from nineteenth-century scholars such as Hermann Paul (1846–1921) going back to the Stoics (from the fourth century bc: see Chapter 13 above): their essential novelty, however, cannot be denied. In particular, what distinguishes Saussure's thought from that of the preceding scholars is its systemic approach: every linguistic (p. 471) unit can be defined only by virtue of the system of relations it has with the other units. Another characteristic feature of Saussure's thought is his attempt at building an autonomous linguistics, namely independent from psychology, sociology, or any other discipline, contrary to the methods of most linguists immediately preceding him. Actually, this attempt was not completely carried out by Saussure: it however became the trademark of all later schools of European structural linguistics, together with the systemic approach to linguistics, as will be seen in the following sections.
18.104.22.168 Langue vs Parole
It is necessary to point out that the English word language has two different equivalents in French, namely langage and langue. According to Saussure (1959 ), langage (rendered as ‘speech’ or ‘human speech’ in the English translation) ‘is many sided and heterogeneous’ and it has ‘both an individual and a social side’ (p. 8). The opposition between langue and parole as introduced in Saussure (1922) appears somewhat oversimplified with respect to its handwritten sources (cf. §20.1.1 above). In his class lectures, Saussure actually distinguished not two concepts, but three (langue, parole, and langage): the last concept is not only presented as a purely ‘many sided and heterogeneous’ phenomenon but, in a more positive way, also as the faculty which allows humans to acquire any language; and Saussure also speaks of a faculté de langage ‘faculty of language,’ a concept which appears rather close to the homonymous Chomskyan one (see e.g. Chomsky 1975c). However, Saussure only focuses on the langue/parole dichotomy. Langue is defined as the ‘social side’ of language (langage). It is the common code shared by all the speakers belonging to a given linguistic community: it is ‘a storehouse filled by the members of a given community through their active use of speaking’ (de Saussure 1959: 13). What Saussure calls parole ‘speaking’ denotes both (a) the usage of this common code by the different individuals and (b) the psycho-physical device which allows them to put such code into use (cf. de Saussure 1959: 14).
22.214.171.124 Synchrony vs Diachrony
In Saussure's own words, ‘synchrony and diachrony designate respectively a language-state and an evolutionary phase’ (1959: 81). In itself, this opposition was nothing new: as Saussure himself recalls (p. 82), the ‘programme’ of traditional grammar, such as Port-Royal grammar, ‘was strictly synchronic.’ On the other hand, ‘since modern linguistics came into existence, it has been completely absorbed in diachrony.’ The difference between Saussure and earlier linguists therefore lies in their respective views of the opposition between synchrony and diachrony. Traditional grammar almost totally ignored diachrony; nineteenth-century historical-comparative grammar (‘modern linguistics,’ in Saussure's just quoted passage) subordinated synchrony to diachrony, stating that only a diachronic study of language can be really scientific (cf. Paul 1920: 20). Saussure's position is wholly opposite: ‘it is evident that the synchronic point of view predominates, for it is the true and only reality to the community of speakers’ and if the linguist ‘takes the diachronic perspective, he no longer observes language (langue) but rather a series of events that modify it’ (de Saussure 1959: 90). Synchronic facts cannot be (p. 472) accounted for in diachronic terms. For example, ‘historically the French negation pas is identical to the substantive pas “step,” whereas the two forms are distinct in modern French’ (p. 91). Saussure's opposition between synchrony and diachrony can be condensed (as is standard) in the following way: synchronic facts are systematic and meaningful; diachronic facts are isolated and ateleological (i.e. without a goal). As will be seen below (§20.2.3), this polar opposition between synchrony and diachrony was rejected by some schools of European structuralism, especially by the Prague school, which attempted to overcome it: but Saussure's dichotomy had anyway the effect of producing a radical change in the goals and methods of twentieth-century linguistics with respect to those of nineteenth century. While the latter was mainly of a diachronic kind, the former preferred synchronic studies.
126.96.36.199 Signifier vs Signified
The signifier and the signified are the two ‘sides of the linguistic sign’ (de Saussure 1959: 66). Saussure (p. 67) states that ‘the bond between the signifier and the signified is arbitrary. Since I mean by sign the whole that results from the associating of the signifier with the signified, I can simply say: the linguistic sign is arbitrary’ (original emphasis). This is the so-called ‘doctrine of the arbitrariness of the linguistic sign,’ which has sometimes been misunderstood. First of all, it must be kept in mind that this doctrine not only states that the relationship between a given sequence of sounds (e.g. /buk/) on the one hand and a given object (a pile of printed sheets bound together) on the other has no natural basis, since it derives from a convention: such a ‘conventionalist’ conception of the linguistic sign can be found in many linguists before Saussure, and can be traced back at least to Aristotle's treatise De interpretatione. Saussure's conception is new and deeper: according to him, the relationship between the signifier and the signified is not, in the first place, a relationship between language and reality, but a relationship internal to language itself. What singles out a linguistic sign, in Saussure's perspective, is its value, namely its relationship with the other signs of the linguistic system to which it belongs. The signifier is the value from the point of view of the expression (of the ‘sound,’ in Saussure's words), the signified from that of the content (the ‘concept’). The value of a sign is therefore not intrinsic, but is simply the outcome of its differences from the other signs belonging to the system. This ‘differential’ conception of the sign is the basis for the Saussurean notion of arbitrariness: ‘arbitrary and differential are two correlative qualities’ (de Saussure 1959: 118). ‘In language there are only differences. […] differences without positive terms’ (p. 121; original emphasis). For example,
Modern French mouton can have the same signification as English sheep but not the same value, and this for several reasons, particularly because in speaking of a piece of meat ready to be served on the table, English uses mutton and not sheep. The difference in value between sheep and mouton is due to the fact that sheep has beside it a second term while the French word does not (pp. 116–17).
(p. 473) If linguistic signs are not simply conventional labels which designate a given reality, differing only in their sounds, but rather are values determined by their reciprocal relationships, what warrants them? The fact that they belong to a system shared by a social group, to a given language (langue): ‘the social fact alone can create a linguistic system’ (p. 113). However, as any other social fact, also a language is historically conditioned: the sign systems may change in the course of the time, as the opposition between synchrony and diachrony shows. This social and historical conditioning is the basic reason for the arbitrariness of the sign, in Saussure's sense.
188.8.131.52 Syntagmatic vs Associative Relations
This dichotomy may be exemplified by means of the word ‘instruction.’ The combination between the stem ‘instruct-’ and the suffix ‘-ion’ is an example of a syntagmatic relation, namely of a combination between two signs. That between e.g. ‘instruction’ and ‘education’ is an example of an associative relation: each of the two words refers to the other, and any of them can replace the other in a given context (e.g. ‘higher instruction’ vs ‘higher education’). The associative relations are surely considered by Saussure as belonging to a language (langue); the place of the syntagmatic ones is not wholly clear. Saussure states that the sentence, which is ‘the ideal type of syntagm,’ ‘belongs to speaking (parole), not to language’ (1959: 124). On the other hand, ‘to language rather than to speaking belong the syntagmatic types that are built upon regular forms’ (p. 125). De Mauro, in his commentary on Saussure's text (1972: n. 251), resorting to its handwritten sources, has come to the conclusion that Saussure considers also that sentences, insofar as they realize general patterns, also belong to la langue. This is probably the most accurate interpretation of Saussure's real thought: it is a fact, however, that the linguists immediately following Saussure, not having the handwritten sources at their disposal, maintained the view that the he ascribed the sentence to speaking (parole), and not to language (langue).
20.2 The Schools of Geneva, Prague, and Copenhagen
20.2.1 General Features
The schools of linguistics most directly influenced by Saussure's thought differ considerably from each other, and such a differentiation often occurs even among scholars belonging to the same school. However, they share some significant ideas about the nature of language and the aims and methods of linguistics, which are essentially a critical development of some Saussurean basic insights. One such idea is the conception of the language as a structure, namely as set of entities defined not in themselves, but by virtue of their reciprocal relationships: it must also be stressed that the labels (p. 474) ‘structure,’ ‘structural,’ etc. became key terms of linguistics due to the work of post-Saussurean schools (Saussure very seldom employed ‘structure,’ rather speaking of ‘system’). Another consistent development of Saussure's ideas by the European schools of structural linguistics, which marks their difference with respect to most preceding trends, is the abandonment of psychologism: according to such linguists, language has to be described only on the basis of its structure and its functions, without any reference to psychological entities or processes. This ‘anti-psychologistic’ attitude fully characterizes the Prague and Copenhagen schools, while some remnants of psychologism can still be detected among the Geneva scholars.
20.2.2 The Geneva School
It may seem somewhat paradoxical, but the first members of the Geneva school, Bally and Sechehaye (§20.1.1), were possibly the European structuralist linguists least influenced by Saussure's thought. This paradox is, however, only apparent: the views of Bally and Sechehaye were already formed when Saussure gave his classes on which the Cours de linguistique générale is based. As we hinted above, Bally's and Sechehaye's connection with pre-Sassurean linguistics is shown by their residual links to nineteenth-century psychologism.
A decidedly psychologistic attitude characterizes Sechehaye's first book, (Sechehaye 1908). For example, Sechehaye maintained that one of the major tasks of theoretical linguistics was the solution of what he called the ‘grammatical problem,’ which would consist in the investigation of the ‘psychophysical basis’ of the laws and of the functioning of the grammar (cf. Sechehaye 1908: 24). Actually, Sechehaye never solved this big problem: nevertheless, his analyses of many grammatical phenomena (especially those contained in Sechehaye 1926) are very insightful and still deserve attention. After the appearance of Saussure's Cours, Sechehaye also turned to more general problems, such as the relationship between the social and the individual sides of language. Detaching himself from Saussure's concept of langue, intended as a common code shared by a community of speakers, he denied the legitimacy of assuming an entity of ‘a language in itself,’ over and above the languages of the individuals (Sechehaye 1933: 65).
Bally's psychologism is especially to be found in the opposition between what he called the ‘intellectual’ and the ‘affective’ components of language. According to different circumstances, either of the two components can prevail over the other, but they always occur together: e.g. an affective (or ‘emotional,’ or ‘expressive’) element occurs even in the apparently most neutral utterances, such as ‘it is raining’ (cf. Bally 1926: 23). The reason for this lies in the fact that the concrete use of language always consists in a dialogical relation, where the speaker not only transmits some intellectual content to the hearer, but also expresses his own emotions, attempts to reach some goal, etc. (p. 33). Bally calls the discipline which has to deal with the combined effect of (p. 475) both the intellectual and the affective side of language ‘stylistics.’ In his sense, therefore, stylistics is not limited to the analysis of literary texts: e.g. to say John, I cannot bear him stylistically differs from I cannot bear John: the intellectual content is the same, but the affective element is stronger in the first sentence than in the second.
Bally's notion of stylistics is surely an original facet of his thought, which distinguishes it from Saussure's. Other notions directly derived from Saussure, in their turn, are somewhat modified by Bally. In particular, he reshapes the opposition between langue and parole: he defines parole as the ‘actualization’ of langue. All elements of langue are ‘virtual,’ and to be applied to the reality they have to be ‘actualized’: e.g. book as an element of langue is a virtual concept, which becomes actualized by means of the ‘actualizer’ this, in a phrase such as this book (Bally 1965: §119). The phenomenon of actualization shows that parole follows langue from the point of view which Bally calls ‘static.’ From the ‘genetic’ point of view, however, this relationship is reversed: parole precedes langue in the genesis of language.
Among the scholars belonging to the second generation of Geneva linguists, we limit ourselves to quoting Bally's pupil Henri Frei (1899–1980). His most important work is Frei (1929), whose title (‘The Grammar of Faults’) is due to the fact that it deals with aspects of contemporary French normally treated as errors by French prescriptive grammars. Frei shows that such alleged errors are actually explained as effects of different needs, which he names ‘assimilation,’ ‘differentiation,’ ‘shortness,’ ‘invariability,’ and ‘expressiveness.’ In this same volume, Frei also worked out a ‘syntagmatics’ or theory of syntagms which resumed and developed Saussure's ideas about syntagmatic relations (§184.108.40.206 above). Frei's investigations in this domain were further deepened in some essays of the 1950s and the 1960s.
20.2.3 The Prague School: the Functional Approach and the Phonological Theory
The Prague school flourished in the 1920s and in the 1930s and was centred around the Prague Linguistic Circle, founded in 1926 by Vilém Mathesius (1882–1945), professor of English at the University of Prague. Mathesius' own work is especially important for his studies of the sentence structure, which he developed in an essentially communicative framework: from his earliest essays on this matter, Mathesius opposed the ‘actual’ to the ‘grammatical’ analysis of the sentence. The latter is the traditional analysis into subject and predicate; the former subdivides the sentence into ‘theme’ and ‘enunciation’ (later called ‘rheme’; see Mathesius 1929). Both analyses are necessary, in Mathesius' view, since they do not always coincide: e.g. the theme is not always identical with the grammatical subject, nor the rheme with the grammatical predicate. The differences are also cross-linguistic: Modern English tends to make the subject coincide with the theme much more than do languages like Czech. Mathesius' investigations, which did not make any great impact at the time of their appearance, became (p. 476) instead very popular after the Second World War, under the label Functional Sentence Perspective.
Mathesius' organizational work was extremely important: the Prague Linguistic Circle counted among its members the most influential European structural linguists, among whom the Russians Roman Jakobson (1896–1982) and Nikolaj S. Trubetzkoy (1890–1938) are especially worthy of mention. The research fields of both scholars were very wide: Jakobson had also investigated folklore and was especially interested in poetic language; Trubetzkoy, after his early studies on ethnographic matters, turned to historical-comparative linguistics and eventually became professor of Slavic philology at the University of Vienna. He was always an active member of the Prague Linguistic Circle (for more information, see Toman 1995). Another important member of the Prague circle was Sergej Karcevskij (1884–1955), who had left Russia after the 1905 Revolution and migrated to Geneva, where he was in contact with Saussure: he can therefore be considered as the link between the school of Geneva and the Prague School. The link between the two schools became official at the First International Congress of Linguists, held in 1928 at The Hague, where both Bally and Sechehaye, on the one hand, and Jakobson, Karcevskij, and Trubetzkoy, on the other, expressed their reciprocal agreement on several points that emerged during the discussion.
On that occasion, Prague scholars stressed the importance of Saussure's conception of langue ‘as a system of reciprocal values,’ hence on the structural conception of language. In their view, however, Saussure's limitation lay in restricting this systematic, structural perspective to synchronic linguistics: a ‘teleological’ and ‘systematic’ view of linguistic change had therefore to replace Saussure's ‘atomistic’ one (cf. Jakobson et al. 1929: 35–6). This position was restated at the congress of Slavists held the following year in Prague, where the famous ‘Theses of 1929’ (see Steiner 1982) were presented by the members of the Prague Linguistic Circle. In particular, the need for overcoming the sharp opposition between synchrony and diachrony is restated in the first thesis: not only does linguistic change show a systematic character, but also any linguistic stage contains some traces of the preceding ones. Among the other arguments dealt with in the theses of 1929 (ten in total, but the most relevant for general linguistics are the first three), the distinction introduced in the second thesis between the sound considered as ‘an objective physical fact’ and as ‘an element of a functional system’ deserves special attention. This distinction became the basis of the opposition between phonetics and phonology, presented and developed in Trubetzkoy (1958 ), which we will take up in a moment. The third thesis deals with the different functions of language: the Prague linguists maintain that the study of language, on both the synchronic and the diachronic plane, cannot be adequate if the different linguistic functions (communicative, referential, poetic, etc.) are not taken into account (the problem of the different functions of language and of their definition was returned to by Jakobson some decades later; see Jakobson 1960). The content of this thesis clearly illustrates the ‘functional' view of language that already underlay Mathesius’ kind of sentence analysis (see above) and can therefore be considered as the forerunner of the several ‘functional frameworks’ which developed in the second half of the twentieth century.
(p. 477) As has been alluded to above, the topic focused upon by the Prague linguists (and which gave them the greatest renown) was phonology; in their usage, phonology not only indicates a field of analysis, but also a specific theory, namely that worked out by them during the 1930s, and whose most detailed presentation—unfinished, because of the premature death of the author—is Trubetzkoy (1958 ). Here only two aspects of Prague phonology will be presented: the distinction between phonetics and phonology and the notion of ‘phoneme.’1
Trubetzkoy (1969 : 4) defines phonetics as the science of sounds ‘pertaining to the acts of speech’ (so, approximately, to Saussure's parole), and phonology the science of sounds ‘pertaining to the system of language’ (Saussure's langue). Actually, the opposition between the two sciences of linguistic sounds derives from their relationship with meaning: phonetics does not take it into account, while phonology singles out the sound differences which bring about differences in meaning (pp. 10–11). Hence phonetics only investigates ‘the material side’ of the linguistic sounds, their acoustic and articulatory properties, which instead concern phonology only insofar as they have a ‘distinctive function’ (they distinguish meanings). This distinctive function differs across languages: some different sounds which bring about a meaning difference in a given language do not produce it in another. It is on this possibility/impossibility of meaning that the opposition between ‘sound’ and ‘phoneme’ is based. ‘Phoneme’ was a term coined in the 1870s by a rather obscure French phonetician, A. Dufriche-Desgenettes, to contrast linguistic sounds with other kinds of sounds (e.g. those of music); it had also been employed by other linguists, including Saussure and Jan Baudouin de Courtenay (1845–1929), with a different meaning, which only partly anticipates Jakobson's and Trubetzkoy's. These latter scholars define phonemes as the units which, in a given language, bring about a meaning difference and cannot be analysed into smaller units. In other words, the phoneme is the smallest distinctive phonological unit in a given language (cf. Trubetzkoy 1969: 35). Hence phoneme is opposed to speech sound: the first entity only contains distinctive (‘relevant’ is Trubetzkoy's word) features, the latter the non-distinctive features as well (Trubetzkoy 1969: 36–7). This opposition is immediately related to that between phonetics and phonology: sounds are the ‘material’ entities, phonemes the ‘functional’ ones.
Trubetzkoy (1969: 46–65) works out several rules to discover the phonemes of a given language. Their effect can be summarized as follows: two sounds of a given language realize two different phonemes if (a) they occur in the same position; (b) they bring about a meaning change. So, for example, English /p/ and /f/ realize two different phonemes, since they distinguish at least two meanings: e.g. pat vs fat (a ‘minimal pair,’ it will later be called). Phonemes are always defined with respect to a given language, since some sounds can be distinctive in one language but not in another. As Trubetzkoy (p. 67) says, a phoneme is determined from its position in the system to which it belongs, namely from its relationships to the other phonemes to which it is opposed. (p. 478) Think e.g. of velar nasal consonants, like the final sound in English sing which distinguishes it from sin, with a final alveolar nasal. Velar nasal consonants also occur in Italian (e.g. in a word like sangue, ‘blood’), but in this language there is no minimal pair brought about by the contrast between velar nasal vs alveolar nasal, like the English case just cited. Therefore the velar and the alveolar nasal consonants, in Italian, are not phonemes but, in Trubetzkoy's terms, ‘variants’ of the same phoneme: more exactly, they are called ‘combinatory variants’ (allophones, in American structural linguistics). The other kind of variants are ‘optional variants’: they do not differ according to the phonetic context in which they appear (like the Italian velar or alveolar nasals, which occur before velar or alveolar stops, respectively), but they are different sounds which may occur in the same position. One example of optional variants is Italian /r/: its standard realization is as an alveolar sound, but several Italian dialects realize it as a uvular phone (like in Parisian French).
The notions of system, phoneme, and variant lie also at the basis of historical phonology, which Prague linguists (especially Jakobson) worked out during the 1930s. Jakobson (1971 ) lists four types of phonological change: (a) ‘extraphonological,’ i.e. when the change has no phonological effects, but only changes the number of the variants of a given phoneme; (b) ‘dephonologization,’ i.e. when the change deletes a phonemic opposition; (c) ‘phonologization,’ i.e. when two combinatorial variants become two different phonemes; (d) ‘rephonologization,’ i.e. the transformation of a phonological opposition into another, which has a different relationship to the phonological system (Jakobson 1971: 209). Actually, the notion of system is essential to account for any kind of phonological change: ‘any modification must be treated as a function of the system within which it has occurred’ (p. 203; original emphasis, my translation). Jakobson's sketch of historical phonology was fully consistent with the Prague 1929 theses, namely the goal of superseding Saussure's neat opposition between synchrony and diachrony, and of showing the systematic character of the latter as well as of the former.
20.2.4 The Copenhagen School
The Copenhagen school was centred on the Copenhagen Linguistic Circle, but, in contrast to the Prague school, the Copenhagen school was not theoretically unified. The most important Copenhagen linguists, Viggo Brøndal (1887–1942) and Louis Hjelmslev (1899–1965) shared the structuralist approach to linguistics (strongly argued for in the opening article of the journal they founded together, Acta Linguistica: see Brøndal 1939), but their similarities do not go beyond this common core. In a nutshell: while Brøndal maintained that language is based on logic, and therefore attempted at analyse it by means of logic (largely borrowed from Leibniz), Hjelmslev's programme was that of basing linguistics on logic, in the sense of the ‘logic of the science’ worked out during the 1930s by the Neo-positivist philosophers. Since Hjelmslev became more (p. 479) celebrated than Brøndal, we will limit our presentation to Hjelmslev's theories; yet Brøndal's analyses are often very perceptive and therefore deserve attention (see esp. Brøndal 1943).
Hjelmslev's theory of language and linguistics was worked out mainly in the 1930s and 1940s. He coined for it the wholly new label ‘glossematics’ in order to distinguish it from preceding theories, which all (with a partial exception of Saussure's) shared the fault—as Hjelmslev saw it—of basing themselves on some discipline from outside linguistics, such as psychology, sociology, etc. They are therefore defined by Hjelmslev as ‘transcendent’ while his own theory, on the contrary, is ‘immanent’ (Hjelmslev 1961 : 4–5). The aim of linguistic theory, according to Hjelmslev, is the analysis of the system of dependences which form the structure of a given language (cf. pp. 21–8). Such dependences are called ‘functions’ (p. 33). As can be seen, this term is identical with that employed by Prague linguists, but its meaning is quite different for the two schools: for the Prague linguists, it designates something external to the structure of language (e.g. its poetic, or its communicative, function); in Hjelmslev's framework it denotes the internal dependencies which constitute the structure itself.
A function closely analysed by Hjelmslev (1961: 47–60) is the ‘sign function,’ the reciprocal dependence between expression and content. These notions approximately correspond to Saussure's signifier and signified, respectively (§220.127.116.11 above). Hjelmslev's starting point is Saussure's statement that the combination of such units ‘produces a form, not a substance’ (de Saussure 1959: 113). Hjelmslev interprets this statement by assuming that the same ‘factor common to all languages’ (which he calls ‘purport,’ p. 50) can be differently shaped across languages, because of the different form that the sign function has in each of them. Lets us give an example both for expression and for content. On the expression plane, the articulatory space of the nasal consonants is differently partitioned in English vs Italian: English opposes three nasal phonemes /m/, /n/, /ŋ/, Italian only two /m/, /n/. On the content plane, consider the way in which some different languages denote the purport of matters to do with the trees: English and French employ wood and bois both for ‘a collection of trees growing more or less thickly together’ and ‘the substance of which the roots, trunks, and branches of trees or shrubs consist.’ German, instead, denotes the latter substance with a special word Holz, while the other meaning of English and French ‘wood’ is expressed in German by Wald. In its turn, German Wald covers also the meaning of English and French forest/forêt (cf. Hjelmslev 1961: 54).
Hjelmslev maintains that the expression plane and the content plane of natural languages, unlike other symbolic systems, are not ‘conformal’ (i.e. isomorphic) because there is no one-to-one correspondence between the minimal elements of each plane. If the two planes were conformal, there would be no reason to distinguish them: it would be a violation of the ‘simplicity principle’, which linguistic theory must strictly follow (Hjelmslev 1961: 18). Any system which is formed by two planes is called by Hjelmslev a ‘semiotic’: natural language is therefore a special case of semiotic. In a semiotic, a given plane (that of expression, that of content, or both) can in its turn consist of a semiotic. In the final part of his main theoretical work, Hjelmslev (1961: 114–27) works out a (p. 480) complicated hierarchy of the different kinds of semiotics which, in his view, should allow one to treat all scientific matters in a semiotic perspective. This enthusiastic conclusion appeared (and still appears) as too easily drawn: nevertheless, it exerted a great fascination on several scholars (e.g. Greimas 1966: 13–17).
20.2.5 Developments of European Structuralism after the Second World War: Jakobson's Binarism, Martinet, Benveniste
18.104.22.168 Jakobson and Martinet
With the outbreak of the Second World War, the activity of the Prague Linguistic Circle was put to an end: Trubetzkoy had died the year before, Jakobson was obliged to leave Czechoslovakia after the Nazi invasion, and Mathesius died in 1945. Nevertheless, the insights of the Prague school were subsequently developed, in different and often contrasting ways, by Jakobson and by another scholar who had been in close contact with the Prague circle during the 1930s, the Frenchman André Martinet (1908–99).
Jakobson's binaristic theory of phonology was presented for the first time in 1939, but its full development dates to the 1940s, when Jakobson settled in the United States. The final formulation of the theory is found in the first part of Jakobson and Halle (1956). Trubetzkoy had already observed that the phoneme is actually constituted of several ‘features.’ Jakobson's basic innovation is the statement that such features are binary, namely that the phonemes are uniquely characterized by the presence or the absence of given ‘distinctive’ features, respectively indicated with the signs + and −. Two such features are e.g. [± vocalic] and [± consonantal]: vocalic phonemes have the features [+vocalic] and [−]; consonantal ones the features [−vocalic] and [+consonantal]; liquid ones (/l/ and /r/ in languages like Czech, where they can form the nucleus of a syllable) have the features [+vocalic] and [+consonantal]; glides (i.e. /j/ and /w/), the features [−vocalic] and [−consonantal]. According to binary theory, any phonemic opposition is to be represented as an opposition of features values: e.g., /p/ and /t/ are both [−compact], and such feature opposes them to /k/, which is [+compact], while they are different from each other since /p/ is [+grave] and /t/ [−grave]. These binary features (twelve in the earlier formulations of theory, fourteen in the final ones) are the same both for vowels and consonants and they are assumed to be universal. In other words, the phonemes of any language cannot be constituted but by these features: cross-linguistic differences are accounted for by the fact that not all features occur in all languages, and that some phonemes can have a positive value in one language and the opposite value in another (e.g., /l/ is [+vocalic] in Czech, but [−vocalic] in Italian).
Binarism was fiercely opposed by Martinet. He allowed that the phoneme is not actually the minimal phonological unit, since it is further analysable into features, but he rejected Jakobson's assumption that such features have only a positive or negative (p. 481) value. According to Martinet, the binary hypothesis is aprioristic, i.e. not sufficiently proved (Martinet 1955: 74). On the other hand, Martinet shares with Jakobson the common Prague School assumption that diachrony is a system: he, however, tends to explain the phenomena of diachronic phonology already dealt with by Jakobson (§20.2.3 above) as due to the need for economy. This means that the phonological system of any language, in its changes across time, always tends to preserve a balance between the number of its members, which cannot be too high (‘minimal effort’), and the need to keep the different signs distinct from each other (‘communicative efficiency’; see e.g. Martinet 1955: 93–7).
Emile Benveniste (1902–76) can be considered the European structural linguist whose beliefs come most directly from Saussure (with the obvious exception of Bally and Sechehaye): at the Collège de France he was a student of Antoine Meillet, who in his turn had been a student of Saussure. Meillet was essentially an historical-comparative linguist, as was Benveniste, but his work also contains important theoretical insights which point in a different direction from the Prague scholars. His first important contribution to general linguistics was an essay devoted to Saussure's doctrine of the linguistic sign (Benveniste 1939; reprinted as Benveniste 1966: ch. 4), where he states that it is the relationship between the linguistic sign and the reality which is arbitrary, while that between the signifier and the signified is necessary. In this way, Benveniste contributed to the clarification of some issues which appeared somewhat obscure in the Saussure 1922 text edited by Bally and Sechehaye; today, on the basis of the handwritten sources (§20.1.1), it is possible to say that Saussure's authentic thought was close to Benveniste's interpretation. In the 1950s, Benveniste was the author of several essays which deal, among other things with the problem of performative utterances, which were investigated more or less contemporaneously by Austin (see especially those reprinted in Benveniste 1966: chs 18–23).
20.3 Other European Schools of Structural Linguistics
20.3.1 Guillaume and Tesnière
The most significant works of the two French linguists Gustave Guillaume (1883–1960) and Lucien Tesnière (1893–1954) were published posthumously as Guillaume (1971–2010) and Tesnière (1959), respectively. Both scholars were essentially independent of the mainstream of European structural linguistics: this is also shown by the fact that both concentrated more on syntax than on phonology and morphology. This is the reason for dealing with them both within the same section. However they differ greatly (p. 482) from each other both in their systems of linguistic analysis as well as in their style, often obscure in Guillaume, extremely clear in Tesnière.
Guillaume starts from Saussure's assumption that language is a system, but he criticizes the Geneva linguist for not having worked out the proper technique to analyse this system. He therefore attempts to build a new theory of language which relates the notion of system to his own notion of time (possibly connected to that worked out by the French philosopher Henri Bergson): this theory is called by Guillaume ‘psychomechanics.’ Guillaume also reshapes the opposition between langue and parole: langue is ‘potential,’ while parole (which Guillaume proposes to replace with discours, ‘speech’) is ‘actual.’ Many different linguistic entities are opposed in this way: e.g. the word and the sentence are considered as the typical units of language and of speech, respectively, and morphology has to do with language, and syntax with speech. The language/speech dichotomy also opposes many categories of traditional grammar: e.g. nouns as a word class are the ‘nouns of language,’ while subordinate clauses with a nominal function are the ‘nouns of speech.’ Guillaume investigates not only intra-sentential syntactic phenomena but also inter-sentential relations: e.g. he opposes two sentences such as I spoke to Peter and It is Peter I spoke to as a ‘basic sentence’ vs a ‘new expressive sentence’ (Guillaume 1971–2010: iii. 175). This inter-sentential relationship is an example of what Guillaume names ‘genetic syntax’: to that, he opposes the ‘syntax of result,’ namely linear word order.
A partly similar distinction is traced by Tesnière, although in a very different conceptual framework. Tesnière is not particularly interested in developing a general doctrine of language, but in building a new system of syntactic analysis, which is characterized by a fundamental opposition: that between ‘structural order’ and ‘linear order.’ The latter consists of the linear sequence of words; the former derives from what Tesnière calls ‘connection.’ Connection is essentially a hierarchic fact: in the syntax of human languages, any two elements are in a dependency relation, since one of them is the governing element and other its subordinate. For example, in a sentence like John speaks, ‘speaks’ is the governing element and ‘John’ the subordinate one (Tesnière 1959: ch 2, §7). The hierarchic relations deriving from connection are represented by Tesnière in the format of tree diagrams (called by him ‘stemmas’), the highest node of which is always the main verb. The verb is therefore the central category of syntax, according to Tesnière: and his classification of verbs according to the number of participant roles (actants) they can take is his best-known contribution to syntactic theory, his so called ‘valency grammar.’ So there are ‘0-valency verbs’ (such as the meteorological ones), ‘1-valency verbs’ (the traditional intransitives), ‘2-valency verbs’ (the traditional transitives), and ‘3-valency verbs’ (such as the verbs of telling and giving).
Long after his death, Tesnière's syntactic model gained much greater success than Guillaume's: it had been worked out in the 1950s, hence more or less contemporarily with Chomsky's Syntactic Structures, and both maintain that human language syntax is organized along two dimensions, the hierarchic and the linear. However, Tesnière's model of grammar, unlike Chomsky's, fully overturned the traditional model of sentence analysis based on the dichotomy between the subject and the predicate in (p. 483) favour of the valency properties of the verb. It is for such reasons that Tesnière's dependency syntax was taken as a starting point for syntactic theories alternative to the Chomskyan generative grammar towards the end of the 1960s.
20.3.2 The London School
The label ‘London school’ in the strict sense refers to the group of linguists formed by John R. Firth (1890–1960) and his students at the School of African and Oriental Studies. Among the latter, M. A. K. Halliday (b. 1925) is the most influential, with his Systemic-Functional Grammar, one of the functionalist theories to present itself as an alternative to Generative Grammar in the second half of the twentieth century.
The London school shares the general features of British structural linguistics, which, while taking account of Saussure's and Prague School research, developed an autonomous framework. This autonomy can be seen in the work of scholars such as Daniel Jones (1881–1967) or Alan H. Gardiner (1879–1963). Jones, a leading phonetician, dealt with the notion of phoneme in the second decade of the twentieth century, earlier than Trubetzkoy and the other Prague linguists. Unlike Trubetzkoy, however, Jones did not define the phoneme in terms of ‘relevance’ (§20.2.3 above), but as ‘a family of uttered sounds […] in a particular language which count for practical purposes as if they were one and the same’ (Jones 1957: 22). Gardiner, an outstanding Egyptologist, devoted a book to ‘the theory of speech of language’ (Gardiner 1951; 1st edn 1932), terms which respectively translated Saussure's parole and langue. However, while Saussure (1959: 14) defines parole as something ‘accessory or more or less accidental,’ the notion of speech is central for Gardiner: he defines it as ‘a set of reactions’ to external stimuli in order to obtain cooperation from other people (Gardiner 1951: 20).
In a rather similar vein, Firth's analysis of the communication process does not aim at discovering an abstract and over-individual structure such as langue, but rather at analysing the behaviour of individuals in particular special situations: and indeed the notion of ‘context of situation,’ which he borrowed from the anthropologist Bronislaw Malinowski (1884–1942), has a key role in his system. Firth therefore views language as a set of socially defined behaviours, of which it is possible to build a typology (cf. Firth 1957: 190). This attitude seems rather close to Bloomfield's and other American linguists' behaviourism, but one has to remark that Firth always declared himself neutral with respect to the behaviourism/mentalism controversy. Rather, his analyses seem to parallel some aspects of linguistic pragmatics which was beginning to develop in those same years (the 1950s) with the work of Austin and late Wittgenstein: for example, Firth's statement that meaning does not belong to words and/or sentences in themselves, but is a function of the context of situation of the speaker and the hearer, is surely close to Wittgenstein's (1953) view of meaning as use.
It is therefore not surprising that Firth's most original contributions, in the fields both of phonology and of syntax, are related to the notion of context. According to (p. 484) Firth, Prague phonology restricted itself to a ‘paradigmatic’ approach, namely to the delimitation and classification of sounds and phonemes, while a ‘syntagmatic’ approach is also necessary, namely the analysis of the context where the sounds occur. ‘Prosodies’ are therefore to be added to phonemic entities (and actually Firth's model of phonology is named ‘prosodic phonology’): by ‘prosody,’ Firth does not only mean accents, tones, or intonation, but also any other entity defined on the basis of its function within the spoken chain. For example, the English central vowel schwa (which occurs in words as can and was when they are unstressed) is not a phoneme but a prosody, the occurrence of which is determined by the rules of English syllabic structure. Hence, according to Firth, the sound chain is not formed by a simple combination of elements independent from each other, but is governed by autonomous rules: the syntagmatic axis crosses the paradigmatic one.
Firth's key notion in syntax is that of ‘collocation,’ which has a meaning different from ‘context,’ since it specifically applies to the syntactic environment, while context, as has been seen, more generally refers to the cultural and situational environment. Collocation is defined by Firth (1968: 181) as ‘an order of mutual expectancy’ between two or more words: e.g. a word like ass is (or was) often collocated with a word like stupid, in phrases such as You stupid ass! (Firth 1957: 195, 1968: 179).
20.4 Closing Remarks
Coming back to Saussure's dichotomies, summarized in §§22.214.171.124–4 above, we can now ask ourselves: to what degree did each of them influence subsequent linguistic research? The dichotomy between signifier and signified seems to have been that most poorly understood and, as a consequence, that which had the least impact on the later linguists: only Hjelmslev and Benveniste appear to have fully realized its real meaning. On the contrary, the dichotomies between langue and parole and between synchrony and diachrony (and, to a lesser extent, those between syntagmatic and associative relations) opened a new era of linguistic thought. It has been seen that the interpretation of such dichotomies was not unanimous across the different schools of European structural linguistics: but it became impossible to avoid distinguishing between an abstract and a concrete aspect of language (langue vs parole) and between the analysis of a language at a given chronological moment and of the changes it underwent during its history (synchrony vs diachrony). This is therefore the most enduring heritage of Saussure's reflections about language and linguistics.