Functions of Language 5/1, 1998, 87-98.
On ‘usefulness’ and ‘validity’ in the theory and practice of linguistics
A riposte to H.G. Widdowson
Robert de Beaugrande
your professional training as a linguist […] just doesn’t help you to be useful to other people
— Noam Chomsky (1991: 88)
Whether a linguistic theory is useful or not is not for linguists to decide.
— H.G. Widdowson (1997: 166)
A. Is ‘confusionism’ (still) going around?
1. Modern linguistics seems to have suffered from a malady we might call, a bit wryly, ‘confusionism’: a proclivity for diagnosing ‘confusion’ without proposing to resolve it by looking at real language in human interaction. Saussure (1966 : 99) already mused, with unintentional irony, that ‘the ideal, theoretical form of a science is not always the one imposed upon it by the exigencies of practice’, which ‘account’ ‘for the confusion that now predominates in linguistic research’. One of his prime examples was itself rather confused: ‘synchronic truth is so similar to diachronic truth’ that ‘linguistics has confused them for decades without realising that such a method is worthless’; but it is ‘absolutely impossible to study simultaneously relations in time and relations within the system’ (1966: 97, 81). Had linguistics been ‘confused’ into doing the ‘impossible for decades’? (See § 20 for fresh evidence.)
2. Some later moves were equally evasive. Hjelmslev (1969 : 6, 76f) deplored how ‘the theory of language’ had been ‘confused with the philosophy of language’ and had postulated a ‘universal’ ‘system’, or a ‘construction of grammar on speculative ontological systems’; such ‘projects are necessarily foredoomed to miscarry’, lacking any ‘possible contact with linguistic reality’ (cf. § 27). Yet he also issued ‘a warning’ ‘not to confuse the theory’ with any ‘application’ or ‘practical method’, and he went on to proclaim that ‘linguistic theory cannot be verified (confirmed or invalidated) by reference to any existing texts and languages’ (1969: 17f). Chomsky (1965: 9) in turn castigated the ‘persistent confusion’ about ‘generative grammar’ being ‘a model for a speaker or a hearer’; the ‘grammar’ merely ‘characterizes’ ‘the knowledge of the language that provides the basis for the actual use of language’ but ‘says nothing about how’ such use ‘might proceed’.
3. The historical pattern seems plain: influential theoretical linguists diagnosing ‘confusion’ but not committing themselves to dispel it by looking closely at the ‘actual use of language’. Apparently, their aspirations for an ‘ideal theoretical science’ (§ 1) led them to consider the use of language itself to be confusing and confused. For Saussure (1966: 14, 9, 11), ‘language is a well-defined object in the heterogeneous mass of speech facts’; ‘speech cannot be studied’, nor can it be ‘put in any category of human facts, for we cannot discover its unity’. Similarly, Chomsky (1965: 4, 201) vowed that the ‘observed use of language’ ‘surely cannot constitute the subject-matter of linguistics, if this is to be a serious discipline’; and that ‘much of the actual speech observed consists of fragments and deviant expressions of a variety of sorts’.
4. Such theoretical posturings imply a remarkable dualism: putting language into use abruptly moves from stable and integrative order over into unstable and disintegrative disorder. So to discover the ‘deeper’ or ‘underlying’ order of language (called ‘langue’, ‘competence’ etc.), linguistics must take it back out of use (called ‘parole/langage’, ‘performance’ etc.). In effect, doing so replaces rlll language with ideal language which exists nowhere except in some ‘linguistic theory’ (§ 10). (Beaugrande 1998a, 1998b). So we needn’t be too shocked when a linguist openly ‘speculates’ that ‘linguistics is not about language’ (Smith 1983: 4).
B. The ‘Widdowson fence’
5. Now comes H.G. Widdowson (1991, 1995, 1997), an applied linguist, to diagnose ‘confusion’ in the work of three of the currently most important theoretical linguists who have insisted upon examining ‘actual use of language’: John Sinclair, Norman Fairclough, and Michael Halliday. He seems anxious to keep his home domain of applied linguistics safely fenced off from corpus linguistics, critical discourse analysis, and systemic functional linguistics, respectively. All three papers warrant ‘critical discourse analysis’, but I shall only examine the latest, which seems most likely to provoke a counter-productive relapse into ‘confusionism’.
6. Here in Botswana, the whole eastern border of the Central Kalahari Game reserve is sealed off for hundreds of kilometres by a ‘veterinary cordon fence’ which prevents the spread of cattle disease but also blocks the natural migration of wild species. Similarly, the ‘Widdowson fence’ would seal off applied linguistics from unsettling neighbours, but might also block the entry of ideas vitally needed for enrichment and renewal.
7. Perhaps Widdowson is uneasy about systemic functional linguistics because, more than any other internationally acclaimed linguistic approach, it resolutely seeks to integrate theory with practice, and to place theoretical issues in the context of potential applications. Halliday (1994: xxixf) has expressly promoted a linguistics to ‘help’ people ‘learning their mother tongue’ or a ‘foreign language’; ‘training translators and interpreters’; composing ‘reference works (dictionaries, grammars)’; assisting ‘the diagnosis and treatment of language pathologies’ (‘tumours’, ‘autism’, ‘Down's syndrome’); ‘designing appliances’ for ‘the hard of hearing’; and so on.
8. But according to Widdowson, systemic functional linguistics (a) is not doing, and never can do, what it believes and says it does (§ 13, 24); and (b) sacrifices its ‘generality’ and ‘validity’ by ‘designing a linguistic theory’ ‘to suit a particular purpose’ (166)1 (§ 10f, 13). He seems to question but retains the ‘supposed opposition between theoretical validity and practical utility’, which nicely preserves ‘the distinction’ between ‘theoretical linguistics’ versus ‘applied linguistics’ (146). ‘Use and usefulness cannot be intrinsic design features of the description and cannot be adduced as measures of its validity’ (145) Yet this ‘opposition’ is irrelevant because in Halliday’s (so far) most definitive exposé (second edition 1994), no claims were lodged for ‘validity’ anyway.
9. The view expressed by Labov — a sociolinguist, please note, not a systemic linguist — that ‘the application of a theory’ ‘determines its value’ (146) is then claimed to render ‘science’ ‘indistinguishable from technology’ (147). This sweeping claim would degrade or disqualify all those portions of science that have been, consciously or not, attuned to potential applications, such as medical science and social science, plus the sociolinguistics of Labov or, for that matter, of Bernstein, Hasan and many others.
10. What would Widdowson have science do? He approvingly invokes the ‘argument’ that theoretical ‘linguistics’, ‘in its quest for understanding, cannot do otherwise than idealise reality and produce abstract models which bear no direct resemblance to the actual experience of language’ (146) (§ 15). ‘Linguistics, as the theoretical study of language has to be at a remove from reality as a condition on its being theoretical; its business is the intellectual quest for understanding through the devising of abstract systems which underlie reality’ (166). Conversely, ‘commitment to certain uses‘ is suggested (by rhetorical questions) to ‘reduce the scope of a theory’ and its ‘internal consistency’ and ‘cogency’ (147). These criteria would award ‘validity’ to linguistic theories exactly like Chomsky’s (see § 4), who recently announced that ‘your professional training as a linguist’ ‘just doesn’t help you to be useful to other people’ (1991: 88). The highest ‘validity’ should go to theories dealing only in ‘language universals’, where ‘generality’ and ‘scope’ are maximised at the price of empirical and social vacuity (cf. § 25).
11. A paradox caps Widdowson’s argument that ‘whether a linguistic theory is useful or not is not for linguists to decide’: ‘designing theories expediently to suit a particular purpose’ not only ‘loses their more general validity’ but even ‘correspondingly reduces’ ‘their range of possible usefulness’ (166)! This logic is perverse unless we accept the dubious premises that (a) specific applications are best when derived from general theories; and (b) the ‘validity’ of a theory can be inferred from the sheer number of diverse applications it leaves room for without anticipating them. Surely systemic functional linguistics is precisely the refutation: the range of applications enumerated by Halliday and cited in § 7 shows how anticipating them guides, supports, and justifies the development of the theory, for which ‘validity’ in the narrow sense of ‘idealising reality’ and ‘being theoretical’ is not claimed anyhow (§ 8, 27).
12. Widdowson’s rationale for ‘idealising reality’ sounds innocuous enough, though a bit patronising: ‘after all to understand anything means to see what underlies actual appearances; knowledge depends on distancing’ (146). But in systemic functional linguistics, this can only be one half of a dynamic dialectical process: already for J.R. Firth (1968 : 19), ‘abstract linguistics’ gets its ‘justification’ when ‘the linguist’ ‘finally proves his theory by a renewal of connection with the processes and patterns of life’ and ‘experience’ (cf. § 18). Because ‘a speech event’ is an ‘expression of the language system from which it arises and to which it is referred’, ‘we can only arrive at some understanding of how’ ‘language works’ if we ‘take our facts from speech sequences’ ‘operating in contexts of situation which are typical, recurrent, and repeatedly observable’ (Firth 1957 : 144; : 35).
13. For Widdowson, however, ‘contrary to the claims often made for it, systemic functional grammar cannot account for language use’ (166). Unlike Chomsky, whose ideas he is known to reject, he doesn’t dismiss ‘language use’ as a morass of ‘fragments and deviant expressions’ which no ‘serious discipline’ should investigate (§ 3). But he does characterise ‘language use’ in such terms that no ‘grammar can actually account’ for it (149) (§ 25), which reminds me of the Saussurian reservation about the ‘heterogeneity’ of ‘speech’ (§ 3). He roundly asserts that ‘there cannot be’ a ‘correspondence between categories of description and categories of behaviour’; and that ‘unresolved discrepancies’ ‘invalidate the claim that the validity of functional grammar is based on its usefulness’ (149) — forgetting again that this claim was not lodged by Halliday (§ 8, 11).
14. These opportunistic ‘discrepancies’ constitute the ‘signposts on the Widdowson fence’, so to speak, which appear in the 1997 paper in these terms:
theoretical linguistics applied linguistics
intellectual quest pragmatic activity
categories of description categories of behaviour
stay removed from reality explore conditions of relevance
theoretical validity practical utility/applicability
semantic potential pragmatic realisation
not depend on context depend on context
Table 1. ‘Signposts on the Widdowson fence’
15. The ‘business’ of ‘theoretical linguistics’ is to conduct an ‘intellectual quest’ by ‘establishing abstract, necessarily idealised generalities’, and by building ‘models’ that ‘exist only in the world of their own abstraction’ and ‘at a remove from reality’ (165) (§ 10). The ‘business’ of ‘applied linguistics’ is to conduct ‘pragmatic activity’ by ‘exploring different conditions of relevance and different possibilities of appropriation in practical domains, most obviously in language pedagogy’ (165f).
16. The obvious lesson is that these two businesses are ‘informed by different principles of enquiry’ (145). On one side is ‘analysis’, whilst its counterpart on the opposite side is usually called ‘interpretation’ or ‘understanding’ (145, 150, 151, 153, 154, 158, 163f), but also ‘language experience’ (155), and, at one point, even ‘language’ itself (157). Failing to distinguish ‘analysis’ from these counterparts is diagnosed as ‘confusion’ ‘with far-reaching consequences’ (151). By no coincidence, the same accusation was marshalled against ‘critical discourse analysis’ by Widdowson (1995), who misses no occasion to deplore it (here too, see § 23), namely by defining what Fairclough and others do as ‘interpretation’, not ‘analysis’. Widdowson has recently signed a contract for a book telling the world what ‘discourse analysis’ really should be (cf. § 23).
17. For Widdowson, ‘analysis’ must be exhaustive, ‘arbitrary’, and non-’selective’:
Analysis, in principle, takes everything into account that is encoded, and the decision to select certain features to attend to is essentially arbitrary, a matter of descriptive convenience. But […] when this system is actually exploited in use only a part of its potential is realised. […] This is not arbitrary. (153)
If this be ‘analysis’, I have never done any, nor can I recall a single example in anyone else’s work; perhaps Widdowson’s book will be the first. I see no way to determine, much less ‘account’ for, ‘everything that is encoded’, much of it probably tedious and irrelevant in any case. Instead, we all ‘decide to select certain features to attend to’; and we do so not in any ‘arbitrary’ fashion but in alliance with what I have termed the ‘cognitive interests’ of the discipline (translating German ‘Erkenntnisinteresse’, from Schmidt 1975) (Beaugrande 1980: 2). Yet Widdowson (1995) turns this very alliance into a rebuke against ‘critical discourse analysis’, whose ‘confusion is bred of ideological commitment’. Moreover, much of what we do when we undertake ‘analysis’ (though by Widdowson’s criteria we must always fail) is precisely to ‘regulate our attention and select what is significant’ in ways that, at least in early stages, are not terribly different from the ‘interpretation’ of ordinary discourse participants — what Fairclough (1989: 141f) has called ‘interpretation’ as distinct from ‘explanation’.
18. I would lodge the same objection against the interlocking claims that ‘understanding is not a function of analysis at all’; that ‘linguistic analysis, no matter how detailed, cannot result in understanding of how and why a text means what it does’; that ‘linguistic analysis’ can be ‘entirely derived from text’ and ‘not depend on context at all’; and that ‘the formal components which constitute the text as linguistic object’ can be ‘exemplified without reference to any context at all’ (152ff). Both ‘understanding’ and ‘context’ are omnipresent in every order of ‘linguistic analysis’. Whereas Chomskyan linguists hide them under the table, Hallidayan linguists seek to make them explicit and methodical, and now get rewarded with accusations of ‘confusion’. Indeed, interpretation and understanding, far from being separate from ‘analysis’ or ‘invalidating’ it (Widdowson 1995: xxx), are the front end of analysis, just as practice is the front end of theory in real life if not in linguistics. Besides, the functionalist principles already stated by J.R. Firth (1968 : 202, 176f, m.e. [= my emphasis]) stipulate that ‘a theory of analysis dispersed at a series of levels must require synthesis’ and ‘congruence of levels’; and that ‘the levels’ must be ‘congruent and complementary in synthesis on renewal of connection with experience’ (cf. § 12).
19. Widdowson proffers, as his supporting argument, the plausible speculation that ‘texts’ are ‘designed to make the most economical connection with context precisely to avoid unnecessary linguistic processing’ (154):
if the actual context provides sufficient information for your communicative needs, you do not have to pay much attention to how this contextual information has been encoded in the language. […] You regulate your attention and select what is significant. […] you activate whatever in the text seems to be contextually relevant and disregard the rest. Otherwise, language use would be an intolerably cumbersome process.
This argument provides no real support because analysis would also be ‘intolerably cumbersome’ if it did not focus upon what ‘seems contextually relevant’. Moreover, recent research has surprisingly disproven the plausible assumption that text processing operates by selective attention; instead, all meanings of a given expression are briefly activated, and the relevant ones are separated from the irrelevant ones by selectively exciting or inhibiting the strength of connections (Kintsch 1988). So ‘understanding’ and ‘interpretation’ are inescapable and largely instantaneous, self-organising processes far in advance of any conscious ‘analysis’. I cannot see how an ‘analysis’ could somehow ‘identify what semantic functions are manifested in a text’ apart from ‘discriminating which of these is pragmatically activated’ (the job of ‘interpretation’) (151), especially if such ‘analysis’ must ‘take everything into account’ (§ 17).
20. I would lodge the same reservation against the doctrine that ‘significance’ ‘cannot be equated with signification’ because ‘significance, the contextual function of language, depends on paying selective heed to the contextually derived signification inscribed in the code’ (152f). Just as ‘analysis’ cannot hope to ‘take everything into account’ (§ 17), neither can it capture the full ‘signification inscribed in the code’. Widdowson seems to ‘accept that significance is a function of signification in that it is historically derived from it’ (153). But he draws the startling consequence that to explain ‘language being as it is, as a system, because of the social functions it has evolved to serve’ is ‘a diachronic statement’ (152). We recall, with an eerie jolt, Saussure’s primordial complaint about ‘linguistics confusing’ ‘synchronic truth’ with ‘diachronic truth’ (§ 1). The whole of systemic linguistics plus its ‘grammar’ and its social semiotics gets miraculously whisked out of the ‘synchronic’ domain into the ‘diachronic’ recesses of history, where it supposedly ‘cannot be an account of how language is used’ (148, 152, 166) (§ 24).
21. A closely related ‘discrepancy’ is orchestrated between semantics versus pragmatics, even though Halliday (1994: xiv, m.e.) has expressly attributed the three-part separation of ‘syntax’, semantics’, and ‘pragmatics’ to ‘formal linguistics’ and to ‘philosophy of language’, in contradistinction to ‘functional linguistics’. ‘Pragmatics’ rarely figures in Halliday’s theory, and least of all off in a world neatly separated from ‘semantics’ or from ‘grammar’. Yet Widdowson first says that the ‘semantic functions’ ‘are specified in the separate components of systemic functional grammar’, then faces about to aver that ‘there is no separate’ ‘functional feature’ ‘in systemic functional grammar: it is incorporated into the semantic’ (145, 152, m.e.). And he objects to both arrangements.
22. Widdowson predictably argues, along lines we saw in § 17, that ‘in pragmatic use only certain parts of semantic potential are pragmatically realised’ (155, m.e.), and theb charges Halliday with a ‘confusion [!] of internal semantic and external pragmatic function’ (164). Widdowson protests that ‘semantic functions’ ‘combine pragmatically under variable conditions of interpretation which will always elude grammatical analysis’; and that ‘the functionally informed semantic resource which is systemically laid out in the grammatical model is quite different from the pragmatic functions which are achieved in the use of that resource’ (145, 164, m.e.). He goes a step further by contending that ‘once structures are put to use, they naturally contract textual and contextual relations with others and are pragmatically modified’ (163, m.e.), just as ‘all models of grammar’ and ‘all theories of language’, when ‘made operational’ get ‘inevitably altered’ (165). So what is ‘pragmatically’ generated is not just a ‘selection’ from the ‘semantic’ (§ 17) but a ‘modification’ or ‘alteration’ whose ‘differences’ and ‘variable conditions’ must ‘always elude’ any ‘grammar’, including Halliday’s.
23. Castigating the ‘mistaken assumption’ that ‘the manifestation of semantic categories is the same as their pragmatic realisation’ handily reaffirms Widdowson’s notoriously ‘strong reservations about much of the work done in critical discourse analysis’ (164f) (cf. § 16). That work is also accused of giving ‘interpretations’ that ‘invoke the authority of grammatical analysis on the assumption that textual meaning is a function of the grammar itself’ (165). No evidence is given, and the accusation was not made in these terms in the attack upon Fairclough (Widdowson 1995; riposte in Fairclough 1996). We must await Widdowson’s own book to put things right (§ 16f).
24. These, then, are some of the ‘unresolved discrepancies’ making up the Widdowson fence (§ 13f), and forming the groundwork for his denials that ‘a grammar can actually account for language use’ (148). The mantra-like protest-too-much suggests some degree of defensive anxiety (my emphasis throughout). Halliday’s is ‘a grammar of the system and not a grammar of the text at all’, and ‘it is misleading to claim’ otherwise (150, 165). ‘A text-based grammar is not at all the same as a grammar of text’ (151). ‘Which function is operative in a particular textual instance cannot be determined by the grammar’ (158). ‘The grammar’ ‘can never be an account of what people can mean’, because ‘the very account of the potential’ ‘can never tell you exhaustively what language users will make of the language resources at their disposal’ (155). ‘The interpretation of text as the realisation in discourse is a pragmatic matter beyond the scope of grammar’ (145). And so forth. My reply would be: systemic functional grammar moves further toward text and discourse than all others I know of. In real language use, the grammar — Halliday’s (1994: xiv) ‘shorter term for lexicogrammar — constrains text and discourse without fully determining them (Beaugrande 1997a). How this operates should be a central question in both theoretical and applied linguistics, and can never be answered by ‘abstract models which bear no direct resemblance to the actual experience of language’ (§ 10).
25. Fencing off ‘grammar’ from ‘text’ strategically runs in parallel with fencing off theoretical linguistics from applied linguistics. Temporal separation and unpredictability get added to the ‘variability’ we have seen (§ 22): ‘what theories do’ ‘is establish abstract, necessarily idealised generalities which are then actualised and made particular in practice in all manner of ways, very often in ways that cannot be anticipated in advance, and which cannot therefore, be factors in their design’ (165, m.e.). So the ‘appropriation of a model’ — this being, as we saw, the ‘business’ of ‘applied linguistics’ (§ 15) — ‘depends on it not being designed to be appropriate to any one purpose in particular’ (165). The perverse paradox cited in § 11 returns: ‘the more it [the model] strives to be useful, the less useful it is likely to be’ (165); and this runs in parallel with the even more perverse paradox that ‘the more it strives to account for actual use, the less valid it is likely to be as a model’ (165). Again, Chomskyan linguistics and ‘universal grammar’ would win the highest ‘validity’ (§ 10).
26. To borrow Widdowson’s own words, ‘I have some difficulty with these ideas’ (150). They imply that the design of a theory or model of language is frozen from the outset and for all time in respect both to what it can ‘account for’ and to how it can be ‘applied’. Yet Halliday’s is the most eminent paradigm case of a design in continual and dynamic evolution, not least because of its applications in ‘educational linguistics’, but also now in acknowledging the profound implications of corpus linguistics, another field Widdowson (1991) has tried to fence off against Sinclair. Moreover, Halliday’s theory has evolved most of all though its efforts to suggest how ‘the separate systems’ of a language ‘interrelate in dynamic fashion’, which Widdowson oddly vows to be ‘in principle precluded by the divisions built into the design of the model’ (158) (cf. § 21). For Halliday (1973: 49), ‘language’ ‘is a range of possibilities, an open-ended set of options in behaviour that are available to the individual in his existence as social man; the context of culture is the environment for the total set of these options, while the context of situation is the environment of any particular selection that is made from them’. No fences there.
27. Furthermore, I submit that a successful application can very well be one reliable indicator of the validity of a theory (Beaugrande 1997a, 1997b, 1998a). If you want to assist the teaching and learning of a languages effectively, your theory should provide a realistic and accurate model of the processes involved. In training most human skills, the principle of building directly upon the learners’ current level of skills is accepted and practised without question. But language educators are hard put to follow suit if ‘linguistic theories’ persist in ‘idealising reality and produce abstract models which bear no direct resemblance to the actual experience of language’ (§ 10); even Hjelmslev admitted these are ‘necessarily foredoomed to miscarry’ (§ 2). What is audaciously called a ‘quest for understanding’ (§ 10) is in fact a headlong flight from real language. ‘What indeed’, Widdowson asks, ‘is the good of scholars indulging in the play of intellect without regard for what goes on in “the real world”?’ (147). ‘It is hard to defend the case for useless ideas’. But whoever wants to defend it can find justifications in this same paper as it conjures up ‘troublesome issues’ supposedly ‘raised’ by ‘the idea that theories should be designed to be useful’.
1. To save space, I shall just give the page numbers when citing the 1997 paper.
Beaugrande, R. de (1980) Text, Discourse, and Process. Norwood, NJ: Ablex.
Beaugrande, R. de (1997a) New Foundations for a Science of Text and Discourse. Greenwich, CT: Ablex.
Beaugrande, R. de (1997b). Theory and practice in applied linguistics: Disconnection, conflict, or dialectic? Applied Linguistics 18, 3: 279-313.
Beaugrande, R. de (1998a) Society, education, linguistics, and language: Inclusion and exclusion in theory and practice. Linguistics and Education.
Beaugrande, R. de (1998b) Performative speech acts in linguistic theory: The rationality of Noam Chomsky. Journal of Pragmatics.
Chomsky, N. (1965) Aspects of the Theory of Syntax. Cambridge: M.I.T. Press.
Chomsky, N. (1991). Language, politics, and composition. In G. Olsen and I. Gales (eds.) Interviews: Cross-disciplinary perspectives on rhetoric and literacy. Carbondale: Southern Illinois UP, 61-95.
Fairclough, N. (1989) Language and Power. London: Longman.
Fairclough, N. (1996) A reply to H.G. Widdowson’s ‘Discourse analysis - a critical view’. Language and Literature 5, 1: 49-56.
Firth, J.R. (1957) Papers in Linguistics 1934-1951. London: Oxford UP.
Firth, J.R. (1968) Selected Papers of J.R. Firth 1952-1959, ed. F.R. Palmer. London: Longman.
Halliday, M.A.K. (1973) Explorations in the Function of Language. London: Arnold.
Halliday, M.A.K. (1994) An Introduction to Functional Grammar: Second Revised Edition. London: Arnold.
Hjelmslev, L. (1969 ) Prolegomena to a Theory of Language. Madison: University of Wisconsin Press.
Kintsch, W. (1988) The role of knowledge in discourse comprehension: A ‘construction-integration model’. Psychological Review 95, 2: 163-182.
Saussure, F. de (1966 ) Course in General Linguistics. New York: McGraw-Hill.
Schmidt, S.J. (1975) Literaturwissenschaft als argumentierende Wissenschaft. Munich: Fink.
Smith N.V. (1983) Speculative Linguistics: An Inaugural Lecture. London: University College.
Widdowson, H.G. (1991) ‘The description and prescription of language’. In J. Alatis (ed.) Georgetown University Round Table on Languages and Linguistics 1991. Washington, D.C.: Georgetown UP, 11-24.
Widdowson, H.G. (1995) Discourse analysis - a critical view. Language and Literature 4, 3: 157-172.
Widdowson, H.G. (1997) The use of grammar, the grammar of use. Functions of Language 4, 2: 145-168.