History and historicity in linguistics

Functions of Language, 4/2, 1997, 169-213.

On history and historicity in modern linguistics

Formalism versus functionalism revisited¹

Robert de Beaugrande

the history of linguistic theory cannot be written and its evolution cannot be followed — it is too discontinuous.

— Louis Hjelmslev (1969 [1943]: 7)

Linguistic theory appears to offer a stunning variety and disparity of clashing doctrines, [and] striking divergences in terms, slogans, and technical contrivances.

— Roman Jakobson (1970:12)

A. Disconnecting language from history

1. Viewed from a general perspective, linguistics appears to have both a peculiar history, i.e., how linguists and their works have been situated with respect to each other in time; and a rather unreflective sense of its own historicity, i.e., how its theories, methods, statements, and so are situated in the history of science and the history of society. Many university courses and conference workshops reverently approach the ‘classic’ works of linguistics with no prominent consideration of the historical distance that separates them from ourselves and of the major shifts in historical contexts that have since occurred. Such is the symptomatic approach to two of the most widely-read books in linguistics courses, published in 1916 (Saussure) and in 1933 (Bloomfield) respectively. Neither work refers to its own historical setting: the international tensions leading into world-wide warfare and the downfall of multilingual empires in the one case, and the world-wide economic depression in the other case — as if the authors were expressly preparing their work for a non-historical reception.

2. Such a state of affairs is rather unique among scientific disciplines. In physics, chemistry, and biology, very little of the theoretical material that was current sixty or eighty years ago is approached today with a comparably unreflective sense of history and historicity. Indeed, many of the central conceptions of these sciences have just emerged fairly recently, ushering in revolutionary transformations. Not even mathematics, geometry, and formal logic are genuinely stable over time, though they are relatively detached from empirical objects and events (cf. Beaugrande 1991b).

3. In the concluding chapter of my survey on the discourse of linguistic theories (Beaugrande 1991a), I noted how many of the ‘classic’ works perform an emphatic break with the immediate past while reaching further back. The mentalist descriptive linguistics of early continental European structuralism spearheaded by Saussure (1966 [1916]:82) charged prior philology with holding a ‘hybrid and hesitating’ ‘conception of language’ and applauded ‘traditional grammar’ for being ‘absolutely above reproach’ in its ‘strictly synchronic’ ‘program’ for ‘describing language-states’ (cf. § 11). The physicalist and behaviourist descriptive linguistics of American structuralism inaugurated by Bloomfield (1933:32, 6) repudiated ‘mentalism’, which ‘still prevailed’ ‘among men of science’, and which he associated with the outlook of ‘grammarians’ in ‘our school tradition’; in exchange, philology was lauded as ‘one of the principal enterprises, and one of the most successful, of European science in the nineteenth century’. In turn, Chomsky (1965:67, 4f) rebuked Bloomfieldian descriptive structuralism and behaviourism as ‘fundamentally inadequate’ and gave a high estimate both of ‘mentalism’ and of ‘traditional grammar’, praising the latter for having given ‘a wealth of information concerning the structural descriptions of sentences’.

4. This pattern, which I have wryly called ‘ancestor-hopping’, suggests that linguistics does not seek to deny its historical roots and branches wholesale, but that each theory tends to define itself in programmatic opposition to the currently prevailing theory and proposes to consign it to the past. This tendency suggests a deep-lying troubled awareness of persistent and fundamental problems and an ambition to escape from them through energetic theoretical innovation.

5. A bit paradoxically, a triumphant theory has typically ushered in a phase of uncritical orthodoxy and dogmatic allegiance, attested by exclusive claims to scientific authority and by voluminous citations in professional discourse (§ 19). We thus find Newmeyer’s (1980:249f) Linguistic Theory in America presenting Chomsky’s as ‘the world’s principal linguistic theory’, for which ‘no viable alternative exists’; ‘the vast majority’ of ‘linguists’ ‘who take theory seriously² acknowledge (explicitly or implicitly) their adoption of Chomsky’s view of language’. At the book’s conclusion came the astonishing boast that ‘on the basis’ of Chomsky’s ‘idealization, more has been learned about the nature of language in the last 25 years than in the previous 2500’. I felt keenly reminded of Firth’s (1957 [1934-51]:139) remark: ‘to dismiss two thousand years of linguistic study in Asia and Europe’ ‘is just plain stupid’. But for the present paper, I would concede that such invitations for linguistics to forget its own history, though rarely couched in such hugely self-congratulatory terms, have a venerable history of their own (cf. § 11).

6. The understandable outcome has been a rather choppy evolution and a lingering uncertainty regarding what might count as long-range progress in answering basic questions (§ 17, 19f, 22, 45, 57), such as just what kind of a ‘system’ a language might be: a ‘system of pure differences’ (Saussure), a ‘system of voluntarily produced symbols’ (Sapir), a repertory of human behaviour (Bloomfield, Pike), a ‘self-subsistent, specific structure’ (Hjelmslev), an ‘infinite set of sentences’ (Chomsky), and so on (cf. Beaugrande 1991a for details). Conditions remain ripe for another theory to proclaim itself as the final solution and to rally a fresh swarm of adherents.

A. History in philology

7. Several factors may have contributed to the peculiar history of linguistics. Philology, the institutional and intellectual precursor of linguistics, was intensely ‘historical’ in its concern for the evolution of language. Yet this evolution was assumed to conform to ‘laws’ that are specific to language and thus relatively independent of the history of the language community. The foundational historical research of Jakob Grimm (1819) even applied the term ‘grammar’ to these evolutionary ‘laws’, such as the famous ‘Grimm’s law’ accounting for the sweeping sound-shifts that led toward the development of Modern German along different lines from Modern English and its near relatives. In this context, explanation takes the format of a global contrast between two stages and the most concise statement of the regular changes from earlier to later (cf. § 23). The shift, say, of the sound represented by the letter ‘t’ to the sound represented by the letter ‘s’ could thus be described wholly apart from the context of the actual historical events that occurred during the same time span. Philologists did not deny in theory the important effects of such events as the migration of a language community and its contact with other communities (cf. § 65), or the peremptory divisions of territory in Merovingian and Carolingian times. But they held it was not necessary in theory, and often not feasible in practice, to directly relate the events to the evolution of particular vowels or consonants, or of configurations of these.

8. The rather abstract and self-contained view of language history was thus less a matter of high-ranking theoretical principles, than a practical reaction to the widespread lack of documented evidence regarding the histories of language communities. Philology was quite happy to consult such evidence for periods with a relative plenitude of extant documents which both report on history and reveal major changes in the language, e.g., between Middle English versus Modern English. But philology had to develop methods for pursuing its work without such evidence. Only in this way could its histories of languages be extended far back into times where little or no documentation has survived, and where we are obliged to work with idealised constructions. ‘Indo-European’ is surely the most famous: because ‘the basic language is not reconstructable as a fact’, ‘the question “how did the ancient Indo-Europeans speak?” was deflected to the level of the language system’ — a ‘formal schema reconstructed’ from the ‘features’ ‘left after a process of abstracting, unifying, and rarefying’ (Hartmann 1963:74ff, m.t. [= my translation]).

9. The high level of historical abstractness was therefore justified by the proven ability of philology to overreach historical documentation. This abstract evolution of language was also quite congenial for Herder’s concept of language being central to the ‘spirit of the nation’,³ Darwin’s concept of evolution, and the organic conception of language propagated by the philosophy and science of German Romanticism: ‘language in its peculiarity as a changing and hence “living”, “organic” manifestation’ (Hartmann 1963:7f, m.t.). The abstractness was best maintained in ‘sound analyses’, which ‘reached far back, while syntax was postulated roughly as it is found in historically documented times’ (Hartmann 1963:79, m.t.) — a foretaste of the resolutely static and non-historical theories of syntax in modern linguistics (§ 42-47). Foregrounding language sounds, even to the point of bringing them under the heading of ‘grammar’ (§ 7), reassuringly focuses on the most compact and orderly aspect or ‘sub-system’ of language and the easiest one to disconnect from the concrete history of speakers or communities (§ 34). Yet the eminent philologist Hermann Paul (1891:6, my translation) warned that ‘making the study of sounds’ into the ‘essence of language science’ might lead into ‘isolation’ and ‘superficiality’.

10. Still, language sounds can connect to the concrete reality of speakers through a different channel: the human vocal tract and the operations of articulating sounds in various combinations. For example, we might readily explain the evolutionary choice of a prefix like ‘im-’ versus ‘in-’ in terms of the ease of pronouncing with the subsequent consonant, e.g., ‘impossible’ and ‘inclination’ rather than ‘inpossible’ and ‘imclination’. The long-range ‘macro-history’ of processes like assimilation is thereby connected to the very short-range ‘micro-history’ of articulating sounds in succession (cf. § 34). Yet articulation is also a highly stabilising factor insofar as the vocal tract has not undergone significant modifications within the time span usually addressed in historical philology.

B. The evolution of formalist linguistics

11. When modern linguistics resolved to declare its autonomy as a science, even the abstract conception of history in philology was suspended in order to found an entirely ‘static’ or ‘synchronic’ science of language. Some of the most widely-read books, such as Saussure’s Cours of 1916 expressly advocated a non-historical science, possibly as part of their claims be original and innovative (§ 4). In a section headed ‘Inner duality and the history of linguistics’, Saussure (1966 [1916]:81) supported his advocacy of a ‘static linguistics’ over an ‘evolutionary linguistics’ by declaring:

The first thing that strikes us when we study the facts of language is that their succession in time does not exist insofar as the speaker is concerned. [So] the linguist can enter the mind of speakers only by completely suppressing the past. The intervention of history can only falsify his judgement.

This frank invitation to forget the history of language suggests that, by placing the language system in ‘the mind of speakers’, mentalism was impelled to postulate the mind existing in an ideal timeless state in order to evade the prospect that any such mind is in reality continually evolving, especially during discourse (§ 53f) (cf. Beaugrande 1996a).

12. For Saussure, this placement was chiefly a programmatic step to secure the autonomy of ‘general linguistics’ and to mark the turn away from the historical studies of language in philology (R. Harris 1987) (cf. § 3). The latter would henceforth be relegated to the domain of ‘diachronic linguistics’, which, Saussure argued, again in highly mentalist terms, would ‘study relations that bind together successive terms not perceived by the collective mind but substituted for each other without forming a system’ (1966 [1916]:99f) (but cf. § 40).

13. Yet Saussure, who had begun his own career as a distinguished philologist, evidently remained undecided. He moved to include history, giving it much the same sense it had in philology, e.g.: ‘the scope of linguistics should be to describe and trace the history of all observable languages, which amounts to tracing the history of families of languages and reconstructing as far as possible the mother language of each family’ (1966:6). But his ‘definition of language presupposed the exclusion of everything that is outside its organism or system’, including ‘all the relations that link the history of a language and the history of a race or civilisation’ (1966:20). He compromised by proposing to relegate history to an ‘external linguistics’, without specifying how ‘internal linguistics’ might adopt or adapt its results. Over the years, formalist linguistics has insisted upon maintaining the border between inside versus outside (§ 18, 84).

14. This insistence led to the conspicuous resolve of the newly founded discipline of ‘general linguistics’ to defend its autonomy against neighbouring disciplines by promulgating ‘the fundamental idea’ that ‘the true and unique object of linguistics is language studied in and for itself’, i.e., ‘langue’ as opposed to ‘parole’, the use of language in ‘speaking’ (Saussure 1966:232, 14f, emphasis changed). Such a ‘linguistics must be carefully distinguished’ from sciences like ‘prehistory’, ‘political history’,⁴ ‘palaeontology’, ‘anthropology’, ‘sociology’, ‘ethnography’, and ‘psychology’ (Saussure 1966:6, 20, 102f, 224, 9, 147, emphasis changed). The promulgation was renewed and intensified by Chomsky (1965:3): ‘linguistic theory is concerned primarily with an ideal speaker-hearer in a completely homogeneous speech-community, who knows its language perfectly’ (but cf. § 46).

15. Obviously, such conceptions of ‘language’ and ‘linguistic theory’ disconnect language from history far more radically than philology had done. For philologists, disconnecting ordinary history had been done chiefly on practical grounds to mitigate the complex problematics of sparse and uneven historical documentation (§ 7f). For formal linguistics, in contrast, history was disconnected on emphatically theoretical grounds. This move could have contributed to the low sensitivity of linguists for the historicity of the discipline itself, as attested when Chomsky (1965:3f) attributed his ‘primary concern’ (just quoted) to ‘the founders of modern linguistics’ and saw ‘no cogent reason for modifying it’ despite the interval of some fifty years.

16. The major corollary of disconnecting language from history in linguistics was the ascendancy of formalism, because the formal aspects of a language are the ones that appear to hold still and to be the most readily disconnectable (cf. Beaugrande 1994a, 1996a, 1996b). In exchange, the disconnection has understandably led to some disquieting pressures and evasive moves. The relation between theory versus practice has lost its natural dialectic whereby the two sides should control and define each other as they evolve: between the explicit theory held by linguists versus their practices of ‘doing linguistics’; and between the implicit ‘theory of language’ attributed to speakers versus their practices of discourse and communication (cf. § 38) (Beaugrande in press a, in press b).

17. The non-dialectical relation between theory versus practice has naturally encouraged an emphasis in formal linguistics upon abstract theoretical arguments. These can serve as an effusive rhetorical compensation for the abstractness and intangibility of ‘language by itself’, for which direct evidence is difficult to adduce, and to pointedly demonstrate the superiority of the new theory over those already promulgated. Progress was expected to emerge from discursive theoretical bootstrapping rather than from the continuing accumulation and analysis of authentic language data.

18. These tendencies reflected the ambition to defend a separate academic territory for linguistics against its neighbours (§ 13f). Most of the methods for marshalling substantial bodies of empirical evidence in support a linguistic theory would have to adopt procedures like those practised in anthropology, sociology, or psychology. The theory might then be implicated in a dependency on these neighbouring disciplines — just what so many linguistic theoreticians since Saussure were keenly anxious to avoid, as we saw (§ 14). Constructing a theory without such evidence naturally requires theoretical bootstrapping to define language and all its aspects and sub-systems in ‘purely linguistic’ terms, e.g., by postulating ‘laws of basic clause structure’ (Perlmutter and Postal 1978) as compared to the ‘laws’ of Grimm’s ‘grammar’ (§ 7).

19. Theoretical bootstrapping by argumentative rhetoric predictably tends to erode the discipline’s sense of its own history and historicity. Each ‘theory of language’ is represented by a classic work and a school of adherents, who devoutly accept its conception of ‘language’ and reject all others on grounds of theoretical conviction rather than empirical demonstration. A sense of continuity and common purpose become hard to maintain. Moreover, the non-dialectical relation to empirical data about the practices of language in use impedes routine progress through the accumulation of mutually compatible insights or regularities (cf. § 17, 19f. 22, 46); more often, handfuls of data get casually enlisted in elaborating theoretical disquisitions, as when as Chomsky presented his Aspects of the Theory of Syntax, in which just 24 invented sentences were analysed or ‘transformed’.

20. As far as I know, Chomsky’s was the first widely accepted theory that expressly dismissed the discovery of data. He announced that the ‘observed use of language’ ‘surely cannot constitute the subject-matter of linguistics, if this is to be a serious discipline’; and that for attaining ‘a new and deeper understanding of linguistic structure’, ‘sharpening the data by objective test is a matter of small importance’ (1965:4, 20f). So he judged it an unduly ‘strong requirement’ ‘that a theory must provide a practical and mechanical method for actually constructing the grammar, given a corpus of utterances’ — ‘a discovery procedure’ (Chomsky 1957:50f) (cf. § 82). His official reason was not that his own theory happened to be incapable of providing such a method, but that ‘knowledge of the language, like most facts of interest and importance, is neither presented for direct observation nor extractable from data by inductive procedures of any known sort’ (1965:18). The dismissal was justified with the arguments that ‘no adequate formalisable techniques are known for obtaining reliable information concerning the facts of linguistic structure’; and that ‘there is no reason to expect that reliable operational criteria for the deeper and more important notions of linguistic theory’ ‘will ever be forthcoming’ (1965:19). These arguments hinge on the terms ‘formalisable techniques’ as opposed to descriptive or expository techniques; and ‘linguistic theory’ in Chomsky’s sense as opposed to other theories. But the sound conclusion, I submit, should be just the opposite of the one he advocated. Instead of dismissing ‘observation’ and ‘induction’ of ‘data’, we should reject the demand for ‘formalisable techniques’ and for a ‘linguistic theory’ of this kind, precisely because they do not make adequate provisions for observation and induction and thereby cloud the prospects for cumulative progress (§ 6).

21. Replacing ‘observation’ and ‘induction’ led to a severe but little-noticed contradiction On the one hand, linguistics was assigned ‘the problem’ of ‘constructing a description, and, where possible, an explanation, for the enormous mass of unquestionable data concerning the linguistic intuition of the native speaker, often himself’ (Chomsky 1965:20, my emphasis). On the other hand, the same linguistics expressly denied that the ‘speaker of a language’, who has ‘mastered and internalised a generative grammar, is aware of the rules of the grammar or even’ ‘can become aware of them’, and that ‘his statements about his intuitive knowledge are necessarily accurate’; since ‘a speaker’s reports and viewpoints about his behaviour and competence may be in error’, ‘a generative grammar attempts to specify what a speaker actually knows, not what he may report about his knowledge’ (Chomsky, 1965:8). If these denials are valid, they must also hold whenever linguists assume the role of ‘speakers of the language’ whose ‘intuition’ and ‘competence’ linguistic theory is supposed to account for: they too could not ‘be or become aware of rules’, and their ‘reports and viewpoints about competence’ would be equally subject to ‘error’. So linguists were implicitly claiming a miraculous superhuman capacity to ‘become aware of and report’ an ‘enormous mass of unquestionable data’ that the speakers themselves cannot — indeed, to become aware of and report data about the ‘competence’ of the ‘ideal speaker-hearer’ ‘who knows the language perfectly’ (§ 14); so of course ‘discovery procedures’ cannot be ‘reasonably demanded’.

22.WASSS21. We can hardly be surprised if these moves diluted and deflected the already clouded sense of linguistics toward its own history. In a ‘normal science’, the observation and induction of data contribute the most to the day-to-day progress and the soundest basis for the long-range accumulations of data required to make major new discoveries, especially ones that can refute a prevailing theory. Chomsky may have hoped that dismissing ‘discovery procedures’ and instating the linguists as the privileged reporters of ‘competence’ would render his own theory irrefutable and immune to evolution. But with rich poetic justice, the outcome has been rather the contrary. The new freedom to produce theories without a firm basis in observation and induction from authentic data of real speakers has been briskly exploited by other formalists. The history of formal linguistics over the past four decades therefore projects an image not so much of a ‘normal science’ achieving steady routine progress, but rather of a perplexing miscellany of alternative theories, formalisms, and notations, as diagnosed by José Luis Escribano’s (1993:229f) survey of ‘formal theories’:

there seem to be a great many approaches ‘on the market’ whose interrelationships remain as poorly understood as ever. In fact, it is not easy to even determine which of the thirty-odd major syntactic frameworks that have appeared over the last forty years continue ‘alive’. [Some might] not have been ‘theories’ at all, but just ‘formalisms’ built in such a minimalistic way from the very beginning that practically no progress was possible in principle!

The proliferation of ‘minimalistic formalisms’ has been a historical adaptation to the academic decorum that tends to evolve when language as it historically and empirically encountered is replaced by a timeless and idealised formal abstraction. Theories are then advocated on such self-determining criteria as formality, abstractness, and rigour. Such ‘theories’ are virtually immune from any definitive refutation or falsification because they claim the right to legislate their own degree and mode of connection to, or disconnection from, authentic data produced and recorded in historical settings. In return, they offer no sound basis for making progress or indeed for determining what would count as progress in principle (cf. § 6, 19f, 46, 57).

23. Here, explanation takes the format of aligning small sets of invented sentences with a set of ‘formal rules’ that ‘assign’ to each of them a ‘structural description’ (cf. § 43). In contrast to the concise global contrast between two historical stages of a language, as formulated in philology (§ 7), this transaction is pre-eminently tortuous and local insofar as the gap between the sentences and the formalism is handled in ad hoc or intuitive ways, given the enduring lack of ‘formalisable techniques’ for bridging the gap (§ 20). The formality of the rules is thus largely arbitrary and purchased at the heavy cost of marginalising or suppressing those aspects of the data which resist formalisation but which might be crucial for a genuine explanation (Beaugrande 1994a, 1996a).

24. A conception of a language as a timeless system also forecloses the prospect of explaining a language system in terms of how it has evolved. Instead. formal linguistics has implicitly postulated an ideal underlying order corresponding precisely to the current arrangement of the system, whence the famous conception of language forming ‘un système très délicate et très compliqué où tout se tient rigoureusement et qui n’admet pas de modifications arbitraires et capricieuses’ (Antoine Meillet 1903-04:641).⁵In such a system, historical change would seem disruptive, and Saussure (1966:230f, 87, 152ff) did associate it with ‘deteriorations’, ‘vicissitudes’, ‘damage’, ‘disturbance’, ‘breaking’, and ‘effacement’. Indeed, he darkly envisioned a ‘great mass of forces that constantly threaten the analysis adopted in a particular language-state’ (1966:169). This dark vision is the counterpart to his bright vision of a ‘system’ ‘that always comes to the rescue’, e.g., in the ‘preservation of a form’ menaced by change through ‘analogy’ (1966:173) (cf. § 26).⁶

25. A similar dilemma impends in respect to the origins of language. What could have been the origin of such a ‘delicate and complicated system, where everything is rigorously held in place’ (to translate Meillet programmatically)? The two logically possible scenarios are frankly problematic:

(1) the system originated with precisely this design, and has remained so ever since; or

(2) a ‘non-delicate’ and simple system originated and then gradually assumed this design.

Scenario (1) raises the problem of explaining how the ‘delicacy’ and ‘complication’ of the system could have remained essentially constant over an enormous span of pre-historical and historical time — hardly plausible, especially in regard to the lexicon (cf. § 40). Scenario (2) raises the problem of explaining how a language could function without the ‘delicacy’ and ‘complication’ Meillet claimed it must have, and how these factors could have later been incorporated into the design. Saussure’s ‘static linguistics’ evaded all such problems at the price of renouncing all evolutionary explanations.

26. At most, the history of a language would be conceived as merely a succession of ‘systems’, each of them having an equally rigorous design but with modified forms. If so, language changes would not, after all, entail ‘deteriorations’, ‘damage’, and so on, as Saussure fretted (§ 24), but rather shifts that themselves had a systematic quality (Beaugrande 1996a; Halliday in press). Alternatively, the shifts were not systematic but were kept under control by Saussure’s principles that ‘rescue’ the design against the ‘great mass of forces’ (§ 24, 29).

27. His central principles turned out to be quite austere: difference or opposition, and arbitrariness. For Saussure 1966:121, 107), ‘in any semiological system, whatever distinguishes one sign from the others constitutes it’; so ‘language is characterised as a system based entirely on the opposition of its concrete units’. His often-quoted declaration that ‘in language there are only differences without positive terms’ was tied to his seldom-quoted reservation that ‘the signified and the signifier be considered separately’; and doing so would be pointless because ‘their combination’ is ‘the sole type of facts that language has’: ‘maintaining the parallelism between two classes of differences is the distinctive function of the linguistic institution’ (1966:120f).

28. The principle of difference was chosen presumably because it is the most elementary and abstract one for underwriting the ideal organisation of the language system. Yet due to the idealised and disconnected conception of a language, the conception of the ‘sign’ being the ‘combination’ of a ‘signified’ and a ‘signifier’ called for further theoretical manoeuvring, leading Saussure to the implication that language originated by emerging from chaos (cf. Beaugrande 1996a, 1996c). ‘Thought, chaotic by nature, has to become ordered in the process of its decomposition; language takes shape between two shapeless masses’, namely, ‘the indefinite plane of jumbled ideas and the equally vague plane of sounds’ (Saussure 1966:112).

29. Now we face the question of how such a radical evolution from chaos into order could have occurred over time. As if to foreclose the question and to keep history at bay once again, Saussure juxtaposed ‘differential’ with ‘arbitrary’ as ‘two correlative qualities’, remarking that ‘a segment of language can never in the final analysis be based on anything but its noncoincidence with the rest’ (1966:118). So ‘the arbitrary nature of the sign is a ‘principle’ that ‘dominates all the linguistics of language’ (1966:67f). Saussure’s well-known justification was that the ‘signifier’ ‘has no natural connection with the signified’, which he showed by pointing both to different words for the same thing in different languages (§ 32), and — an interesting reflex of his earlier training in philology — to different forms of the ‘same’ word in different stages of language history (1966:69).

30. I find it striking how reverently these terms and conceptions have often been invoked in linguistics ever since, as if they were timeless truths rather than products of one provisional historical stage in linguistic theorising. The reason seems to be that they were deliberately stated at high degrees of generality and abstraction, inviting future linguists to be content with working out the more specific and concrete aspects by inquiring not whether but only how the terms and conceptions apply to each language. As an emphatically non-historical and non-evolutionary explanatory principle, ‘difference’ or ‘opposition’ was being massively overloaded. Quite apart from the familiar problems of absent differences, as in cases of homonyms among signifiers and synonymy among signifieds, differences first have to be generated by differentiation. And differentiation is largely an opportunistic transaction undertaken once some similarities have been registered and judged problematic or unproductive; it cannot be a transcendent, antecedent, and universal transaction that abruptly extracted language out of chaos. At the very least, some prior system, and not just two ‘shapeless masses’, must have been available to undergo differentiation (§ 28).

31. In addition, the conception of ‘arbitrariness’ got enlisted in contradictory ways: as an organisational principle allowing or enabling differences to constitute the system, and as a principle that threatens or resists organisation. On the one hand, we read: ‘signs that are wholly arbitrary realise better than others the ideal of the semiological process’; if ‘the choice of a given slice of sound to name a given idea’ were not ‘completely arbitrary’, ‘the notion of value would be compromised’ (1966:68, 113). On the other hand, we read:

the irrational principle of the arbitrariness of the sign […] would lead to the worst sort of complication if applied without restriction. But the mind contrives to introduce a principle of order and regularity into certain parts of the mass of signs. [Even so], the mechanism of language is but a partial correction of a system that is by nature chaotic. (1966:133)

So ‘everything that relates to language as a system’ serves ‘the limiting of arbitrariness’ (1966:133). Significantly, Saussure went on to suggest that languages wherein arbitrariness dominates most over ‘motivation’ would be ‘more lexicological’, whereas those in which arbitrariness is most limited would be ‘more grammatical’ (1966:133) (cf. § 40). And ‘within a given language, all evolutionary movement may be characterised by continual passage from motivation to arbitrariness’ and vice versa (1966:134); here, the ‘evolution’ of a language becomes a Manichean contest between order and disorder, even though the principle of disorder had been linked to ‘the ideal of the semiological process’.

32. These complex manoeuvres greatly overreach the simplistic argument that language must be ‘constituted by arbitrariness’ merely because otherwise language sounds would have to be inherently meaningful apart from their position within the sound system. The famous example of French ‘arbre’ versus Latin ‘arbor’ (Saussure 1966:67) accidentally made the wrong point by signalling what we lose by disconnecting language from history. Historically, ‘arbre’ is plainly not arbitrary but a regular derivation of ‘arbor’. After disconnecting history as an explanatory principle and rendering the intuitively obvious linkage between the two terms officially inadmissible, the shaky line of argument maintains that ‘arbre’ is what it is because it differs from or stands in an (‘arbre-trary’!) opposition to all other signs of the French language.

C. Formality versus history and evolution in the ‘levels’ of language

33. In the foregoing sections, I have rearranged some major constitutive conceptions and arguments in early modern linguistics around the programmes of first disconnecting language evolution from the history of language communities in philology, and then disconnecting language from both evolution and history altogether in formalist linguistics. I have also suggested that these disconnections were chiefly motivated by practical problems in philology but by theoretical ones in formalist linguistics (§ 7f, 15); and ironically, the outcome in linguistics has been a fresh crop of largely gratuitous and self-made problems that has led to complicated, strained, or contradictory lines of argument.

34. I would also suggest that the widespread acceptance, indeed acclaim of these conceptions and arguments is reciprocally linked to the low sense of historicity in linguistics. They were resoundingly vindicated in the early stages of linguistics by the study of language sounds — the very domain long favoured in philology, as I remarked, and, by no coincidence, the most readily disconnectable sub-system of language (§ 9f). On the ‘level’ of phonology, and only there, do we indeed find a system that genuinely approximates Meillet’s ideal of a ‘delicate and complicated system, where everything is rigorously held in place’ (§ 24f). As prefigured in philology, the ideal abstract order obligingly connects back to the realistic concrete order of the human vocal tract and the operations of articulating (§ 10). Here we have our ideal differences, oppositions, and features that hold each minimal unit — each ‘phoneme’ — in place despite the variability and continuity of actual contours of spoken sounds. Evolution enters into the short-range perspective wherein some articulatory events influence others in the vicinity, e.g., when a final /‑s/ in English is voiced after voiced stops (e.g. ‘beads’) and unvoiced after unvoiced stops (e.g. ‘beats’); and into the somewhat longer-range perspective of developmental studies on language learning among children. And history can either be kept on a very long-range scale of whole periods (e.g., Middle English versus Modern English) or else left aside.

35. The early success of phonology has massively affected the historicity of linguistics insofar as later projects attempted to reapply phonological conceptions such as ‘opposition’, ‘minimal unit’, and ‘distinctive feature’ to other ‘levels’ of language. The paradox resulted that progress (looking forward) in the discipline would be attempted by looking backward and reapplying old tools to new tasks with a minimum of remodelling, whence Halliday’s (1994a: xxxiv) wry commentary: ‘twentieth-century linguistics has produced an abundance of new theories, but it has tended to wrap old descriptions up inside them’. Such an outlook has not been widely acknowledged, due to the emphasis upon innovation (§ 4, 11), and has naturally has complicated the prospects for a coherent historical sequencing among theories and methods.

36. The most direct reapplication occurred of course on the ‘level’ of morphology, the description of minimal meaningful forms duly christened ‘morphemes’. On that level, the disconnection of history worked best for languages meeting four special conditions:

(a) their present historical stage has not been previously described;

(b) their prior historical stages are undocumented;

(d) they are remote enough to have been relatively unaffected by other languages.

In such languages, history appears fairly stable simply because we know so little about it.⁷ Such languages were precisely the ones where the early and most impressive successes of morphology were achieved. Linguists were obliged to engage in extensive fieldwork and thus to observe short-range stretches in the personal, social, and cultural history of speakers and their language community, and to make inquires about past stretches when certain practices or customs developed. Irrespective of any official ‘linguistic theory’, disconnections and idealisations of the types promulgated by Saussure and Chomsky cannot dominate the practices of investigation and description. Moreover, many of the languages described by morphological fieldwork were spoken by ‘pre-modern’ cultures with a rich historical continuity, where the past is still very much present, thanks mainly to the language (see especially Colby and Colby 1981).

37. For languages where the four special conditions are plainly not met, morphology is likely to suffer more severely under the disconnection from history. Such is undeniably the case for English, which has a long and ponderous history of descriptions, documentations, reform projects, and outside borrowings or intrusions. The natural consequence has been a disconcerting range of varieties and views of the language depending on such parameters as how much we emphasise tradition or ‘correctness’, and how much we know about the etymology and formation and words, especially ones whose morphological repertory is overlaid by importations from Greek, Latin, and French.

38. In such a case, morphology introduces an uncomfortable split that phonology does not. Linguistics can safely claim that the sound system described by phonology corresponds in a reasonably straightforward way to the implicit ‘theory’ of the general community of speakers’ knowledge about the sounds of their language, e.g., about when the sound of final /‑s/ should be voiced or not (§ 34). But the knowledge of speakers of English about morphology depends critically upon their awareness of the history of English, e.g., the choice between the suffixes ‘‑able’ versus ‘‑ible’ in the meaning ‘being capable of’ depending on whether the word was borrowed from Latin and if so from which position in the Latin conjugational system (say, infinitive versus past participle).

39. The arid principles of ‘difference’ and ‘arbitrariness’ are hardly helpful here. Morphemes need not ‘differ’ in their forms, witness English ‘‑s’ for both plural of nouns and third person present of verbs. And they are ‘arbitrary’ only in a trivial sense, since nobody would seriously assert that the phoneme /‑s/ naturally means plurality; and their motivation is largely historical. So we might feel inclined to reconnect history for morphology at least in languages where, like English, the conception of the ‘synchronic state’ applies in a different sense than to languages where, like Aymara of Bolivia and Peru, the four conditions I have cited might apply fairly well. Yet, as the extensive fieldwork of Martha J. Hardman (1993, 1994) has shown, a language like Aymara and its relatives in the Jaqi family may well inject history into their morphology by other means. Due to the history of their close-knit and non-patriarchal societies prior to the Spanish invasion, those languages do not take the masculine gender as the basic form and the feminine as the derivation; and they require the use of morphological signals for data sources: when and how in the speaker’s personal history the content of the message became known, as in these utterances meaning ‘she gave him bread’ (from Hardman 1996).

(1a) t'ant' churiw [the speaker saw her do it]

(1b) t'ant' churi siw [someone else said so]

(1c) t'ant' churpacha [judging from the evidence]

(1d) t'ant' churatayna [no direct evidence]

(1e) inas t'ant' churchixaya [no direct evidence; speaker takes no responsibility]

The morphology thus distinguishes among personal knowledge (1a), knowledge through language (1b), inferential knowledge (1c), non-personal knowledge (1d), and non-involvement (1e). I suspect the lesson would be similar, no matter which language we selected: the morphology still connects up to history, though in a wide variety of ways.

40. Within the history of linguistics as a discipline, we might expect that, moving from smaller toward larger units, the study of word-parts as ‘morphemes’ on the ‘level’ of morphology would be logically followed by the study of whole words as ‘lexemes’ on the ‘level’ of lexicology. But this ‘level’ has received far less consideration than phonology and morphology, mainly due to its obtrusive mismatch against such ideals as Meillet’s ‘system where everything is rigorously held in place’ (§ 24f). Instead, the lexicon is replete with diversities, isolated elements, curiosities, and antiquarian detritus — in short, replete with histories, both macro and micro. Here, ‘difference’ and ‘arbitrariness’ totally break down as explanatory principles by virtue of their sheer overabundance within a strictly ‘synchronic’ perspective, whence Saussure’s (1966:133) proposed label of ‘lexicological’ for ‘languages in which there is least motivation’ (§ 31). In exchange, words can often be explained from their histories in the sense of etymologies. The problem there is the staggering quantity of such histories, including micro-histories accounting for just a single word or ‘word-family’. As a one-time philologist, Saussure envisioned ‘precise laws governing the evolution’ of ‘the form of a word’ during ‘etymology’; yet he irritably castigated ‘folk etymology’ for ‘working somewhat haphazardly and resulting only in absurdities’, ‘mistakes’, and ‘deformations’ (1966:31, 173ff, e.a.). Ironically, Saussure the philologist was faulting ordinary speakers for not applying historical knowledge which, as we saw, Saussure the linguist claimed cannot be ‘perceived by the collective mind’ (§ 12). And he was envisioning ‘precise laws’⁸ despite his contention that the ‘relations that bind together successive terms governing the evolution’ do not ‘form a system’ (§ 12, 65).

41. The complex and intense historicity of the lexicon, especially for a language like English, has surely been a main reason why lexicology failed to achieve an equal status with the other ‘levels’, to the point that it is often slighted when discussing the total scheme of linguistics (cf. Bolinger 1970). It demands emphatically evolutionary theories for situating the current state of any lexeme or lexeme-group within a continual process of motivations and changes that are not merely linguistic, but cognitive and social (cf. Beaugrande 1996a; Halliday, in press).

42. So, formal linguistics was largely content to pass on from morphology to the ‘level’ of syntax. Now, the disconnection of history attained new extremes and produced new problems. Whereas phonology and morphology can point to minimal units actually in the data and corresponding to the ‘phonemes’ and ‘morphemes’, syntax postulates operations and rules which do not correspond to the sequence of units but constitute the underlying order (cf. § 23). This key difference impeded the prospect of reapplying the tools of phonology and morphology over to syntax. Two courses of action were open: (1) to construct a different theoretical framework for disconnecting language from language use and language history; or (2) to reconnect language and inquire how the arrangement of words, phrases, and utterances are determined by and social and historical conditions, including authentic contexts of situation, and speakers’ and hearers’ status plans and goals. These two courses were adopted by formalism and functionalism, respectively.

43. The centrepiece of formalist syntax was the so-called ‘generative’ approach that postulates an elaborate system of formal rules capable of ‘assigning a structural description’ to every sentence of the language and to no non-sentences (cf. § 23). Officially, the theory was resolutely non-historical, and the sentence was a strictly non-historical unit. The theory expressly rejected the intent to describe ‘the process of producing utterances’ or the ‘synthesis and analysis’ ‘speaker and hearer must perform’ (Chomsky 1957:48). The timelessness of such a ‘grammar’ was perhaps most conspicuous when Chomsky, after declaring that ‘problems of linguistic theory’ and ‘language learning’ can be ‘formulated as questions about the construction of a hypothetical language acquisition device’, fell back on an ‘“instantaneous” model’ wherein ‘successful language acquisition’ happens in a single 'moment’ (1965:47, 36). And to keep language safely idealised and disconnected from the life-histories of speakers, he doubted, rather absurdly, that the ‘use of language in real-life situations’ ‘plays any role in how language is acquired, once the mechanism is put to work’ (1965:33; cf. Beaugrande in press a).

44. Officially, generative theory did allow for one specialised type of history with a patently theoretical status and disconnected from parameters of real time. The ‘history of derivation’ of a sentence subsumes the ‘sequence of transformational rules that carry strings with phrase structures into new strings’ (Chomsky 1957:107). This ‘history’ merely reflected the design of the ‘grammar’ and its requirement that ‘rules’ be applied to special kinds of ‘strings’ in a certain order. It certainly did not indicate what a speaker or hearer does in real time.

45. Unofficially, however, ordinary human history persisted in a highly anecdotal status. I have observed that when formalists explain or discuss the structure of a sentence, they regularly construct a plausible but fictional ‘mini-history’ for it about when or why it might be said. When Chomsky asked us to ‘consider such a sentence as “I had a book stolen”’, he diagnosed ‘three structural descriptions’ and explained: ‘(i) I had a book stolen when I stupidly left the window open’ in ‘my car’; (ii) ‘I had a book stolen from his library by a professional thief who I hired’; (iii) ‘I almost had a book stolen, but they caught me leaving the library with it’ (1965:21f). Chomsky claimed to be just ‘slightly elaborating the sentence’ and ‘arranging matters in such a way’ that the ‘hearer’s’ ‘linguistic intuition, previously obscured, becomes evident to him’. Evidently, Chomsky and similar formalists were unaware of constantly inventing these fictional mini-histories that implicitly determined their explicit formal analyses and results. Similarly, the disputes that regularly follow formalist presentations at conferences have been due to the disparities that naturally arise because formalist theory nowhere spells out the operations for inventing such mini-histories — again, no ‘formalisable techniques’ (§ 20, 23). Everybody follows their own imagination and ingenuity while claiming to be discovering ‘syntactic rules’ in the ‘grammar’ of the English language. Such disputes are unlikely to lead to progress in the discipline; every fresh mini-history only elicits another ‘rule’ or two, and the ‘grammar’ grows more diffuse and unwieldy.

46. It might be argued here that ordinary speakers also invent fictional mini-histories if given language samples to judge, so that linguists are merely representing the activities of the language community. But two counter-arguments can be readily adduced. The first counter-argument would return to the contradiction noted in § 21 between making the linguists represent the ‘speakers of a language’ versus denying that speakers can ‘become aware of rules’ or give ‘reports and viewpoints about competence’. By implication, the linguists were claiming to be highly exceptional super-speakers who can somehow access ‘rules’ and ‘competence’ after all. Yet their own ‘completely homogeneous speech-community’ expressly disallows any such super-speakers. So the claim would presumably be that the linguists’ own personal, social, and cultural histories have been transformed by their academic training, during which getting disconnected from the ‘observed use of language’ created connections to the ‘deeper and more important notions of linguistic theory’ about which our ‘techniques’ cannot be required to supply ‘reliable information’ (§ 20).

47. The second counter-argument would be that ordinary speakers very seldom have occasion to pronounce judgement upon the formal structure of isolated sentences, but frequently consider what would be sensible, helpful, relevant, polite, and so on, to say during the ‘micro-history’ of a typical situation. Two conclusions impend: either the formalist linguists are indeed representing the activities of the language community and therefore not discovering the rules of formal syntax; or else the linguists are discovering the rules of formal syntax and therefore not representing the activities of the language community. Both of these conclusions may hold in fluctuating degrees that need to be clarified instead of waved aside by attributing invented language samples to an ‘ideal speaker-hearer in a completely homogeneous speech-community’ (§ 14).

48. If this impasse is acute in syntax, semantics suffers the worst of all when language is disconnected from history. Some formalists had hoped to escape by simply disconnecting language from meaning as well and investigating only the forms of language and their distributions (e.g. Z.S. Harris 1951); but then we are left with no semantics at all. Within the framework of conventional formalist linguistics, the preferred solutions have been to reapply old tools to new tasks though claiming to innovate (§ 35), this time by postulating theories in which meanings are either composed of minimal semantic units (e.g. ‘sememes’) in analogy to phonology, or else determined by formal rules in analogy to syntax — or both. The two analogies symptomatically break down when specific descriptions are proposed. ‘Phonemes’ and ‘morphemes’ correspond to units actually in the data, whereas ‘sememes’ do not; ‘syntactic rules’ visibly build or transform phrase and sentence patterns, whereas ‘semantic rules’ merely ‘interpret’ them. If we declare, as Bloomfield (1933:264) did, that a ‘sememe’ is just the ‘meaning of a morpheme’, then we might consign meaning to morphology and again get by without semantics. However, doing so would force upon us the flimsy assumption that the meaning of a whole sentence or utterance is just the sum of the meanings of its morphemes. And morphology would need a far more expansive and elaborated theory of meanings than it has if it is to attack the main issues in philosophical semantics, such as intention, reference, intension/extension, denotation/connotation, and so on.

49. At all events, the history of semantics has certainly been uneven and uncertain. As I have argued in previous papers (Beaugrande 1984, 1988, 1991c), the major obstacle to genuine progress has been the general and premature assumption that the meanings of a language should be described within yet another ideal, abstract, and static system outside of time, just as the cited analogies to other linguistic levels would suggest. If, as I would argue here, meanings are the linguistic domain furthest from constituting such a system, then the assumption is fundamentally counter-productive. Linguists are in effect forced into the dire position of trying to describe a system that has none of the main properties their theories project upon it, especially ‘static’. The actual achievements will be limited to isolating tiny corners of such a system and ‘freezing’ them for purposes of analysis; and the semantics literature indicates that the results seldom lead either to a convergence of data or to a consensus among analysts.

50. In their efforts to establish clear-cut distinctions between what can versus cannot be ‘meant’, semanticists are exceptionally prone to invent exuberant fictional mini-histories, e.g., to assess the status of ‘rules for subject-raising’ and ‘predicate-lifting’ by adducing the sentence ‘John reminds me of a gorilla with no teeth though I don’t perceive any similarity’ (Lakoff 1971:270f, responding to Paul Postal). This kind of activity strays well beyond inventing sober mini-histories for sentences like ‘I had a book stolen’, and clearly leaves behind the activities of the language community (§ 45ff).

51. For the purposes of the present paper, I would also stress that meaning is the most historical domain of language, combining the rich and complex historicity of the lexicon with its own intense and multi-functional dynamics. By struggling to disconnect meaning from history, semanticists are propelled into the abstruse mental gymnastics of imagining what words or sentences ‘mean’ when no one is actually uttering them in the contexts of everyday social interaction. Yet even if these gymnastics succeeded, the results would be inapplicable to what those same words or sentences ‘mean’ when someone is uttering them, because it is precisely those contexts — and not some minimal units or formal rules — that most precisely determine what the meanings are to be.

52. Meaning is therefore historical along the most detailed and differentiated set of parameters of all language domains. The most familiar parameter and the longest in range appears in etymology, where the history of word-meanings is retraced through characteristic shifts, such as when general becomes special, e.g., ‘groom’ for a man and later for a bridegroom or horse-groom. Less familiar but still reasonably straightforward are evolutions of meaning in the wake of social change, e.g., when derogatory or discriminatory meanings are suspended by shifting to other expressions, such as from ‘queer’ to ‘gay’. Much less familiar except in studies of text coherence are the evolutions in meaning that occur among the various uses of the same expression during a single discourse. A memorable instance uncovered during my own work on ‘language for special purposes’ (Beaugrande 1995) is the meaning of ‘logic’ and ‘logical’ in the discourse of Piaget on the evolution of children, e.g. (my italics):

(2) The total structures intervening in the field of logical and mathematical operations are, contrary to the perceptive structures, characterised by their reversibility in the form of inversion or reciprocity and by their additive composition. (1976:128)

(3) a large number of logical, mathematical and physical operations develop [...] spontaneously in the child [and are] completed by propositional and formal operations, making the adolescent’s hypothetico-deductive deduction [sic] possible. (1976:76)

(4) The linguistic distinction of nouns and adjectives corresponds in general to the logical distinction of categories and predicates. (1976:110)

(5) [in the] period of formal operations [there arises] the logic of propositions, the capacity to study statements and no longer only objects placed on the table or immediately represented (1976:61)

(6) the period of concrete operations is characterised by a series of structures on the point of completion; [...] on the logical level, [they are] classifications, successions, […] simple or successive correspondences, multiplicative operations (1976:58f)

(7) the period of concrete operations is [...] a logic which is not based on verbal statements but on the objects themselves, manipulatable objects. This will be a logic of classifications because objects can be collected all together [...]; or else it will be a logic of relations because objects can be combined [...] or else it will be a logic of numbers, because objects can be materially counted by manipulating them. [...] (1976:20f)

In (2) and (3), the meaning is closest to the technical one in the discourse of formal philosophy and mathematics. Less formal and more ‘linguistic’ but still technical but is the meaning in respect to ‘distinguishing categories and predicates‘ (4) and ‘studying statements in terms of ‘propositions’ (5). The meaning then expands and dilutes across ‘classifications, successions’, ‘correspondences’ (6), and finally includes ‘manipulatable objects’ — which were expressly contrasted with ‘logic’ in (5) — merely because they can be ‘collected’, ‘combined’, or ‘counted’ (7). This evolution of meanings moves far away from the formal and technical sense and subsumes an encompassing range of linguistic operations, cognitions, and ‘manipulations’. The evolution is strongly motivated within Piagetian discourse. His ‘stage theory’ assigned the central role to the evolution of ‘logic’ in young people, but he found the formal and technical sense too limited; instead of replacing the term and sacrificing the prestige associated with ‘logic’ as the ideal and most rigorous mode of human thought, he situated the term in a range of contexts with evolving meanings.

53. The most radical parameter in the evolution of meanings obtains in the complex operations and processes required for the human mind to entertain any meaning at all. Despite the wishful thinking among many linguists and logicians, meanings cannot be stored away in the human mind like so many file-cards in pigeonholes, waiting to be pulled out and plugged into sequences (as in the ‘language’ machine’ critiqued in R. Harris 1987). Instead, they emerge on-line, much like the arrays of a complex system when it goes to critical mass and gains emergent properties that are not just the sum of previous properties nor readily predictable from these (Beaugrande 1996a).

54. My line of argument and retrospection so far ought to indicate why this radical on-line mode of evolution has been thoroughly ignored in the past by conventional linguistics and semantics, and why it may well be resisted in the future. It definitively dooms all the projects of formalist linguistics that imply language to be holding still while we describe it (§ 16). Besides, it overturns formalist semantics by ruling out the prospect of investigating and analysing meanings without changing them in the process and thereby getting results that are partly artefacts of our own work (§ 49). And it roundly calls for a new generation of theories quite unlike most of those contending for supremacy so far (see Beaugrande 1996a for details). In that sense, on-line evolution enforces an inrush and onrush of history and historicity for which formalist linguistics is blankly unprepared.

D. On the ‘evolution’ of this paper

55. In this paper, I have been offering assessments of ‘modern linguistics’ that might well appear unduly general in view of my own observation that the discipline has been contentious and fragmented. Or, I may seem to have been unduly specific by isolating the conventional ‘levels’ of language in a hypostatic fashion as if each one had some sustained identity independently of the linguistic theories that have described it.

56. To justify my mode of presentation, I might claim to be taking my cues from influential writings in modern linguistics. There too, we find sweeping generalisations, such as Chomsky’s much-quoted pronouncement upon the ‘primary concern’ of ‘linguistic theory’ (§ 14f). And we also find the frequent isolation of levels, e.g., in the divisions within introductions or surveys, or within linguistic conferences and congresses, or within academic course offerings, where ‘phonology’ and ‘syntax’ are treated like separate worlds. So generalising and isolating are common strategies within the rhetoric of linguistics itself, even when official theories may suggest the contrary by emphasising the diversity of issues and the interaction among the ‘levels’ without showing how we might actually fit them all together (Beaugrande 1994a, 1994b).

57. More importantly, my motivation has been to construct a ‘thought-experiment’ by treating formalist linguistics with its theories and levels as ‘constants’ in order to make history and historicity stand out as ‘variables’. As we move forward in the history of the discipline, we can diagnose an overall strategy to turn familiar tools to new tasks, notably to build morphology upon phonology, and semantics upon syntax, while neglecting lexicology because it resists this strategy (§ 35-42, 48). This strategy makes the discipline appear conservative despite its claims to innovation and places a historical drag upon theories and methods and upon their prospects for progress (cf. § 6, 19, 22, 45). It also instils a premature sense of security and continuity, implying that all the major issues and problems are under control, when in fact many of them have hardly been recognised, let alone resolved, due to the conventions adopted for idealising and disconnecting language.

58. Admittedly, the sequence ‘phonology - morphology - syntax - semantics’ is at most a general drift in the history of modern linguistics. Still, that history would have been significantly different not merely if the sequence had been just the reverse — hard to conceive except as a thought-experiment — but also if linguists had drawn up a co-ordinated research programme for working on all four domains over roughly the same span of time. In particular, if semantics had been a major concern from the start, the grand hypotheses and dichotomies tailored to phonology would surely have not gained such wide and uncritical adherence (cf. § 35).

59. Furthermore, the sequence ‘phonology - morphology - syntax - semantics’ does represent a progression in the steady disconnection of language from history, quite apart from the detailed chronology of research. What I have undertaken to suggest, in return, is that history avenges itself in precise proportion to the degree we try to disconnect it. Moving away from the micro-history of articulation in phonology and the interactional history of fieldwork in morphology, syntax and semantics have become more deeply entrained in inventing fictional mini-histories for artificial data and ideal speakers, whilst actual data and the real activities of speakers and language communities slip out of view (cf. § 45-51).

60. The most serious limitation in my thought-experiment has been to illustrate the history of conventional linguistics directly with a few well-known projects of formalism. Here, my exoneration might simply be that formalism in general and these projects in particular have usually proceeded as if all other methodologies were in principle incompatible with science, discipline, rigour, etc. and could be safely disregarded or dismissed (a striking example was cited back in § 5). Such gestures again deny history in hopes of exempting formalism from its own historicity and awarding it the status of a universal and timelessly valid methodology — whereas in fact it appears timeless only insofar as it adopts a view of language (as ‘langue’, ‘competence’, ‘universal grammar’, etc.) which is safely shielded against change under pressure from authentic data and empirical discovery (cf. § 19-22). Whatever might contradict such a theory can be soothingly consigned to domains like ‘parole’ and ‘performance’, which the theory doesn’t purport to account for. In such an ambience, it is not the relation between signifier and signified that is ‘arbitrary’ (cf. § 27-31), but the relation between linguistic data and any one theory about the timeless ideal ‘underlying’ system.

E. Functionality versus history and evolution

61. The chief motive for my ‘thought-experiment’ has been to prepare the background for the prospect that the methodology of functionalism has had a different history within modern linguistics, and largely insofar as it has not undertaken to isolate ‘language by itself’, to split it into free-standing ‘levels’, and to disconnect it from history and evolution. I shall fittingly illustrate this prospect with the functionalist approach having the longest continuous history,⁹ namely the ‘Prague School’, which has antedated and anticipated recent functional linguistics in rich and extensive ways that have only occasionally been properly acknowledged (e.g. Halliday 1974; Davidse 1987; Beaugrande 1992). The name must be understood quite loosely, since the school had both roots and offshoots elsewhere, such as Kazan (e.g. Kruszewski, Baudouin de Courtenay) Leningrad (e.g. Vološinov), Brno (e.g. Jakobson for a time, Firbas), Bratislava (e.g. Skalička), and Vienna (e.g. Trubetzkoy for a time, Bühler).

62. These linguists were familiar with Saussure’s distinction between synchronic versus diachronic methods (e.g. Mathesius 1975 [1935]:11), but did not accept his claim that modern linguistics must be entirely synchronic. On theoretical grounds, Jakobson (1986 [orig. 1928]:166) objected to the dark vision of historical change noted back in § 24:

Saussure’s teaching that sound changes are destructive factors, fortuitous and blind, limits the active role of the community to sensing each given stage of deviations from the customary linguistic pattern as an orderly system. This antinomy between synchronic and diachronic linguistic studies should be overcome by a transformation of historical phonetics into the history of the phonemic system

This viewpoint he shared with Valentin Nikolaievich Vološinov, whose work he praised in letter of 1931 to Trubetzkoy for his ‘superb interpretation of linguistic problems’ and his ‘dialectical method’ (quoted by Matejka and Titunik in Vološinov 1986:vii). For Vološinov (1986 [1929]:61), who was among the first to approach language from an explicitly dialectical standpoint ‘Saussure’s views on history are extremely characteristic for the spirit of rationalism that continues to hold sway in the philosophy of language and that regards history as an irrational force distorting the logical purity of the linguistic system’. However, Vološinov (1986 [1929]:59) regarded the phonology emanating from ‘the so-called Kazan school (Kruszewski, Baudouin de Courtenay)’, which would presumably carry over to Jakobson (and Trubetzkoy), as a ‘vivid expression of linguistic formalism’. Also, in a letter of June 1996, Daneš writes me that Jakobson ‘tended to highly abstract and general schemas’ and cites a 1974 paper by Miroslav Renský:

Where Jakobson tended to reduce Mathesius’ interplay of tendencies to a set of hierarchically ordered rules (modern parlance) and saw the necessity of a rigorous and schematic analysis of one subsystem before moving to integration, Mathesius’ first concern was to maintain the structural and functional complexity of language. While Jakobson (and Trubetzkoy) pulled the Prague doctrine toward scientism and incipient formalism, Mathesius steered closer to the humanist approach in his quest for crucial problems rather than exclusive solutions

In the same spirit, Daneš’ presidential address to the Societas Linguistica Europaea in September 1994 ends with a call for ‘taking up a humanistic stance to which’ ‘any human issue should be of deep concern’ (1995a: 193) .

63. On practical grounds, many Prague school linguists were concerned with language history because they had had extensive training in philology, an established part of university language and literature programmes in central Europe, in some places even today. Historical outlooks were also encouraged by the rich documentation for the major languages the School studied, such as English, German, and French, as well as the Slavic (or ‘Slavonic’) group, mainly Czech, Slovak, and Russian.

64_.Moreover, the region has been emphatically multilingual for centuries, so that Czech and Slovak scholars are accustomed to approach languages from a communicative standpoint and from the outside as well as the inside: how the organisation of one language appears from the viewpoint of the organisation of another language. In this tendency, they were close to philology, witness Hermann Paul’s (1891:6) observation that ‘die Vergleichung ganz unverwandter Sprachen untereinander, wodurch die Kraft des Einzelnen noch mehr absorbiert wird, wichtige Aufschlüsse über das allgemeine Wesen der Sprache zu liefern vermag’ [in my translation: ‘the comparison of entirely unrelated languages with each other, whereby the power of each single one is even more absorbed, can provide important insights into the general essence of language’].

65. These special qualities of Prague School linguistics can be retraced in the foundational work of Vilém Mathesius from 1907 onward up through the 1920s and 1930s. His ‘linguistic characterology’ (Mathesius 1928) both anticipated by decades what would later be called ‘contrastive linguistics’¹⁰ and exceeded it in scope by combining the linguistic aspects with historical and literary aspects. As I view his work, his central idea was both quite plausible and quite different from the central idea of formalism to treat every language by itself and on its own terms as an abstract formal system. Instead, Mathesius reasoned that to understand the organisation of a language in any one stage, we can profitably compare and contrast it with other languages or to the same language in another stage, so that we are less likely to take things for granted as the ‘normal’ way a language should be. So Mathesius’ posthumous volume purporting to be A Functional Analysis of Present Day English on a General Linguistic Basis (published in Czech in 1961 as Obsahový rozbor současné angličtiny na základé obecné lingvistikém and in English in 1975)¹¹ was in fact an explication of English in terms of how its organisation compares to that of Czech, German, and French, and how it evolved through Old and Middle English up into the ‘present day’. For example, his ‘formal analysis of the sentence’ proffers an explanation that would be inconceivable, say, in a Chomskyan formal syntax of the sentence: ‘when French verbs were taken over into English they preserved their constructions, which often contained the accusative and thus differed from the older Germanic constructions, which frequently contained the dative and thus agreed with Czech’; examples would be ‘aider quelqu’un’ or ‘flatter quelqu’un’ as contrasted with German ‘jemandem helfen’ or ‘jemandem schmeicheln’ (1975:124). Such analyses also seriously challenge Saussure’s evasive claim, cited in § 12 and § 40, that ‘diachronic linguistics’ is concerned with ‘relations’ ‘substituted for each other without forming a system’.

66. The comparative viewpoint of the Prague School has naturally resisted formalism, which highlights how each language distributes its forms on purely internal grounds, and welcomed functionalism, which highlights how the several languages exploit their respective formal means to perform similar functions. Quite apart from any official linguistic theory, the functions of language inevitably offer the most productive basis for comparing them.

67. In addition, the grammars of Czech and Slovak are resolutely ‘functional’ in the sense that they do not have repertories of fairly frozen sentence patterns of the kind English offers, like ‘the man hit the ball’ or ‘the bill is large’. Instead, word order depends so closely on the current context that a single sentence can hardly be formulated without considering when and where it would be uttered. So Czech and Slovak linguists are ‘natively’ sensitive to functionality when they describe other languages. Mathesius himself was consciously aware of this nativeness factor, e.g., when he explained why ‘verbal aspect in Germanic languages’ had been ‘tackled in an involved manner’ by ‘German linguists’ who ‘lacked insight’ because ‘verbal aspect is by no means so developed in German as it is in Slavonic languages’ (1975:68).

68. With this special vision, the Prague School was strategically positioned to notice an aspect of functionality that had previously eluded synchronic descriptions of non-Slavic languages,¹² a principle they came to call ‘Functional Sentence Perspective’ or ‘FSP’ for short. The English term is not altogether felicitous in that the Czech term ‘aktualní’ is not ‘functional’ but closer to ‘current’ or ‘being of immediate interest or concern’ (compare German ‘aktuell’) (Firbas 1992:22); and the ‘sentence’ is certainly not the sole unit (§ 74-79). Still, the term underscores the highly dynamic conception of the ‘functional’ in Prague School research. And for my own line of argument here, the term neatly points to a deeply non-formalist sensitivity for the evolution of linguistic sequences in discourse — whence Trnka’s (1966 [orig. 1940]:163) conception of ‘experiencing language’, which Daneš (1994:252) glosses with a ‘literal translation’ from the Czech (‘prožívání jazyka’) as ‘actual living through one’s language’ (cf. § 78).

69. A leading factor in this evolution is the ratio between well-known or insignificant content versus new or significant content: the parameter Firbas (1971) has christened with another deeply non-formalist and evolutionary term: ‘communicative dynamism’. With their elaborate inflections and flexible word order, Czech and Slovak have much more leeway than languages like English, French, and German to control the curve of dynamism, usually by beginning an utterance or sentence low with the ‘theme’ and ending it high with the ‘rheme’.

70. This factor can be illustrated with parallel passages, these taken from the second chapter of Luke in the New Testament (after Firbas 1995):¹³

(8a) In the countryside there were shepherds out in the fields keeping guard over their sheep during the watches of the night. An angel of the Lord stood over them and the glory of the Lord shone round them.

(8b) Or il y avait dans la même contrée des bergers, qui couchaient dans les champs et gardaient leurs troupeaux pendant les veilles de la nuit. Un ange du Seigneur se présenta á eux; la gloire du Seigneur resplendit autour d’eux.

(8c) In der Gegend dort hielten sich Hirten auf. Sie waren in der Nacht auf dem Feld und bewachteten ihre Herde. Da kam ein Engel des Herrn zu ihnen und die Herrlichkeit des Herrn umstrahlte sie.

(8d) V té krajině nocovali pod širým nebem pastýři a střídali se na hlídce u svého stáda. Najednou u nich stál anděl Páně a sláva Páně se kolem nich rozzářila

(8e) V tom istom kraji boli pastieri, ktorí v noch bdeli a strážkili svoje stádo. Tu zastal pri nich Pánoj anjel a ožiarila ich Pánova sláva.

In each version, the shepherds, who hadn’t been mentioned in the text before, are introduced and then become thematic for the new arrivals, the angel and the glory. The English, French, and German versions (8a-c) all postpone the mention of the shepherds by giving the location (‘countryside, contrée, Gegend’), which had been mentioned before; and (8a) and (8b) also postpone with an ‘expletive’ or ‘existential’ construction to fill in the early stretch of the first clause (‘there were, il y avait’), while (8c) uses the standard German inversion for clauses with an initial non-subject element. All three versions then present the two new arrivals (angel, glory) before the thematic pronouns (‘them, eux, ihnen-sie’) for the shepherds: (8a) and (8b) at the very start of the clauses, and (8c) with one minimal postponement (‘Da kam ein Engel’). The Czech