Theoretical linguistics , like theoretical physics , theoretical chemistry or theoretical biology , is , of itself , non-empirical .sx It is free to create its own theoretical constructs as it will .sx But , also like these other theoretical sciences to which I have just referred , it originates with the observation and systematization of identifiable phenomena which appear pre-theoretically to have something in common .sx In so far as it retains its internal coherence and distinctive identity - in so far as theoretical linguistics is to be distinguished from the theoretical branches of other sciences - it maintains , and must maintain , its connexion with what is pre-theoretically identifiable , across all societies and cultures , as the referent of Saussure's 'langage' .sx Theoretical linguistics - more precisely , theoretical general linguistics - is that branch of the subject which sets out to provide a non-trivial , intellectually satisfying , answer to what I referred to earlier as the central defining question of linguistics :sx 'What is language ?sx ' ( construed , as we shall see below , in a particular way) .sx Although this question contains the ontological presupposition that there is such a thing as language ( 'langage' ) , of itself it says nothing about its ontological status .sx It does not necessarily imply that language is empirically and pre-theoretically separable from non-language .sx Both general and descriptive linguistics have always operated , however , with the assumption that this is so .sx To quote W.S. Allen , on this point :sx " We presume that there is a particular mode of human behaviour which it is legitimate to isolate and to label as 'language' ; we assume also that this behaviour is such that systematic statements may be made about its various manifestations " ( 1957b :sx 13) .sx Once again , if we wish to be precise , we need to be clear about the distinction between process and product , and consequently about the different ways in which language 'manifests' itself to us in the physical world .sx Most branches of linguistics draw their data from the products of the process , not from the process itself ( various kinds of muscular and neurophysiological activity) .sx This is an important point which has been dealt with above ( see Chapter 2) :sx I will not elaborate upon it further .sx For present purposes , let us simply note that the two assumptions , or postulates , made explicit in the passage just quoted - the postulate of isolability and the postulate of systematicity - have proved their worth over the centuries ( the history of linguistics , in some of its branches at least , and of linguistic theory is measured in centuries ) and need not be justified in detail here .sx Nor is there any need to labour the point that the isolability of the " particular mode of human behaviour " that is pre-theoretically identifiable as language ( 'langage' ) rests , operationally , upon the relatively clear , empirically determinate and theory-neutral , difference between speech and non-speech .sx The question 'What is language ?sx ' can be addressed from several points of view and can be answered in several different , but equally legitimate , ways according to the point of view that is adopted .sx Theoretical linguistics , founded upon the Saussurean and post-Saussurean trichotomy of 'langage' , 'langue' and 'parole' , interprets the question as meaning 'What is a language ?sx ' ( 'Qu'est-ce qu'une langue ?sx ' ) .sx The different branches of theoretical linguistics adopt characteristically different points of view and consequently postulate different kinds of theoretical constructs in the answers they give to the question .sx Theoretical microlinguistics ( often called autonomous linguistics :sx see Chapter 2 ) adopts the point of view expressed by Saussure , or rather his editors , in the famous final sentence of the Cours :sx " la linguistique a pour unique et v e ritable objet la langue envisag e e en elle-m e-circ me et pour elle-m e-circ me " ( 1916 :sx 317) .sx It is the controversial " en elle-m e-circ me et pour elle-m e-circ me " , of course , which distinguishes theoretical microlinguistics from the various sub-branches of theoretical macrolinguistics .sx But they too , as I shall argue , have their own distinctive conceptions of the language-system .sx It must not be thought that sociolinguistics or psycholinguistics , or the other branches of macrolinguistics , can dispense with the distinction between the system and the process ( or its products) .sx I will come back to this point .sx But first let me make explicit something else which is not immediately obvious .sx This is the fact that 'la langue' in the famous passage from the Cours that I have just quoted can be interpreted either generically or specifically .sx Its generic interpretation defines the field of theoretical general linguistics ; its specific interpretation , that of theoretical descriptive linguistics .sx This is the gloss that I would add to Katz's formulation of the goals of what he calls linguistics tout court ( I would call it theoretical linguistics ) in his recent defence of Platonic realism ( as an alternative to both American structuralism and Chomskyan cognitivism) :sx " linguistics tries to construct theories to answer the questions , first , 'What is English , Urdu , and other natural languages ?sx ' and second , 'What is language in general ?sx ' " ( Katz , 1981 :sx 21) .sx Two further terminological comments may be made about this passage , by way of exegesis :sx ( i ) by " natural languages " Katz , like most philosophers and linguists , clearly means N-languages ; ( ii ) by " language in general " he means , in Saussurean terms , not 'langage' , but 'langue' ( construed generically) .sx His two questions are in fact post-Chomskyan reformulations of Saussure's " la linguistique a pour .sx .. objet .sx .. " ( Chapter 4) .sx Although theoretical general linguistics existed long before Chomsky published his seminal work in the mid-1950s , modern theoretical descriptive linguistics is very much his creation .sx A generative grammar of any N-language - English , Urdu , etc. - is a theory of that language :sx more specifically , a theory of the well-formedness of the sentences of the language .sx So far , I have been concerned , first of all , to point out that , although until recently there was no need to distinguish between 'general linguistics' and 'theoretical linguistics' , nowadays there is ; and , second , to prepare the ground for the distinction that I am drawing between theoretical linguistics and linguistic theory and for the necessarily brief presentation of my own approach to the definition of the field of theoretical linguistics , on the basis of alternative , equally legitimate , conceptions of language-systems .sx But why , it may now be asked , is it not possible to operate with a single notion of the language-system valid in all branches of linguistics , micro- and macro- , theoretical and non-theoretical ?sx This is a question that has been dealt with in some detail in the preceding chapter .sx The answer , as we have seen , derives partly from the apparently sui generis properties of N-languages and partly from the complexity and heterogeneity of the pre - theoretically isolable phenomena identifiable as 'language' ( i.e. , as Saussure's 'langage') .sx The multiplicity and heterogeneity of the connections that can be established between what are pre-theoretically classifiable as language-data ( 'des donn e es langagi e-grave res ' , if I may employ this useful post-Saussurean adjective ) and other data , natural and cultural , constituting the subject matter of other disciplines are such that , in my view at least , there is no immediate possibility , perhaps even no ultimate possibility , of constructing a unified theory of the natural and social sciences within which a unitary theory of language ( of 'langue' construed generically ) would find its place and be descriptively and explanatorily adequate to the data that it systematizes and accounts for .sx As to the apparently unique , or sui generis , character of what are commonly referred to as natural languages , this may well have been exaggerated at times by proponents of so-called autonomous linguistics .sx The discontinuity between language and non-language , on the one hand , and the determinacy , arbitrariness and closedness of grammatical structure , on the other , have certainly been greatly exaggerated by linguists of various schools , generativist and non-generativist .sx The fact remains that nothing remotely resembling a comprehensive , intellectually interesting and empirically satisfactory account of the grammatical structure of N-languages in terms of the theoretical concepts and explanatory principles of other disciplines has yet been provided by any of those who have challenged the sui generis character of languages .sx There is every reason therefore to continue to subscribe to a working hypothesis that has proved its heuristic value in the practical description of languages over the centuries and has been , more recently , the foundation-stone of what is so far the most sophisticated branch of theoretical linguistics , both general and descriptive :sx theoretical ( synchronic ) microlinguistics .sx This does not mean , however , that we should , as practitioners of microlinguistics , whether general or descriptive , close our minds to those aspects of language that are not , or do not appear to be , sui generis or deny the validity of alternative views of the nature and ontological status of language-systems .sx The ontological status of the language-system ( Saussure's 'langue' ) has been controversial ever since the publication of the Cours .sx Saussure's own views are unclear and perhaps contradictory .sx At one time , he says that they are supra-individual social facts ; at another time , that they are stored in the brains of individual members of the language-community ( 1916 :sx 23-32) .sx And each of these conflicting views is incompatible with the view , recently advocated by Katz ( 1981 ) , that language-systems are purely abstract , mathematical ( so-called Platonic ) objects ( see Chapter 4) .sx As will be obvious from what has been said earlier , Katz's view ( which is close to Hjelmslev's , 1943a ) , is the one that I accept for microlinguistics , though not for psycholinguistics , sociolinguistics or other branches of macrolinguistics .sx Some part of the confusion and controversy that has surrounded the Saussurean distinction of 'langue' and 'parole' over the last half-century or so is to be attributed to the fact that both words are used in the Cours non - technically ( i.e. , pre-theoretically ) as well as technically ; and , since the theoretical distinction ( or , as we shall see , distinctions ) that Saussure draws between them correlates with differences of meaning in everyday French , it is not always clear in what sense they are being employed in particular contexts .sx It must also be admitted that Saussure's own comments ( or those of his editors ) about the rough equivalence between French 'parole' and German 'Rede' ( and Latin 'sermo' in contrast with 'lingua' ) are less than helpful ( 1916 :sx 31) .sx They must have encouraged , even for those who have read the Cours in French , what has undoubtedly been , over the years , by far the most serious misunderstanding of the technical distinction between 'langue' and 'parole' :sx the view that it relates basically , or primarily , to the distinction between language and speech .sx It does indeed cover one dimension or one part of the semantic difference between 'language' and 'speech' ( between German 'Sprache' and 'Rede' , between Russian 'jazyk' and 'recj' , etc.): or rather , to be more precise , between 'language' construed as a count noun and 'speech' understood as referring to the product , rather than the process , of speaking .sx But it does so , as we have seen , only secondarily .sx The primary distinction is between a language and utterances ( spoken , written , or whatever :sx i.e. , products , not processes , inscribed in some appropriate physical substance or medium ) which , by virtue of their structure ( and independently of their physical manifestation ) , are identifiable as utterances of the language in question ( see Chapter 1) .sx It is unfortunate that the beginnings of theoretical linguistics should have coincided , for historically explicable reasons , with a period of extreme phonocentrism .sx But no more needs to be said on that score .sx Much of the controversy , if not confusion , that still attaches to the Saussurean , or post-Saussurean , distinction between 'langue' and 'parole' ( or the Chomskyan distinction of 'competence' and 'performance' , which is valid for psycholinguistics but not for microlinguistics ) must , however , be attributed to what in this and the previous chapter I have characterized as a false assumption :sx the assumption that there is only one kind of reality and that so-called natural languages , N-languages , must be either psychological or social entities , or , in terms of an alternative dichotomy , that they must be either physical or non-physical .sx It is my contention that microlinguistics , on the one hand , and the several branches of macro - linguistics , on the other , start from the same pre-theoretical notion of N-languages and that , according to their own viewpoint and the alliances that they forge with other disciplines ( mathematics , psychology , sociology , anthropology , etc ) , they each practise a particular kind of abstraction and idealization in the construction of the ontologically appropriate model of the underlying language-system .sx I have already referred to one common misunderstanding of Saussure's terminological distinction between 'langue' and 'parole' :sx a misunderstanding based on the view that it correlates directly with the distinction between language and speech .sx