It used to be assumed that structural linguistics, as exemplified in the work of the Prague and Copenhagen circles in Europe and Bloomfield and his immediate successors in the United States, represented a definite break with the immediate past, in particular with the views associated with the great historical linguists of the nineteenth century. The latter were depicted as interested in such problems as linguistic change, reconstruction, genetic relationships, and so forth: their cast of mind was primarily archeological. Interest in synchronic description and the theoretical foundations of linguistics did not, so the story went, come into the forefront of attention until the advent of structuralism.
At the same time, the structuralists themselves were not unaware that their theories had not come into being ex nihilo. Trubetzkoy acknowledged the pioneering work of Baudouin de Courtenay and Winteler in phonology, and Hjelmslev thought of himself as developing theoretical insights arrived at by Ferdinand de Saussure. So the structuralists identified a number of scholars active in the latter part of the nineteenth century as standing apart from the general run of scholarly activity in linguistics: they were the forerunners of structuralism, in other words.
This historical picture of the rise of structural linguistics was, as it were, tacked on to an already existing picture of the development of nineteenth-century historical linguistics from what was thought to be the dilettantism of preceding centuries. This picture has been painted by the founders of historical linguistics themselves and touched up and preserved for posterity by such historians of linguistics as Theodor Benfey and Holger Pedersen.
The composite picture then was a three-stage affair. First there was the prescientific period of linguistics up to 1800. Then came the scientific but predominantly historically-oriented linguistics of the nineteenth century, and finally structural linguistics incorporating and transcending the results of the previous century.
This comforting perspective began to fade soon after generative grammar appeared in the late 50's. First of all, the scientific validity of structural linguistics was called in question. The critical guns were directed first at structuralist theories of syntax, but eventually at what passed for the most secure and unshakeable portion of structuralist doctrine, namely its phonological theory. Postal makes one point there which I would like to draw your attention to. In addition to repeating Chomsky's arguments against the postulation of a phonemic level in phonology, he suggests that the phoneme is merely a different surface manifesta- | [p. 417] tion of the theory of sound change held by such archetypical neogrammarians as Hermann Paul. If Postal is right, this means that conceptually much of the synchronic approach of twentieth-century linguistics is based on theories having to do with diachrony held by nineteenth-century linguists. In other words, it makes twentieth-century linguistics seem much less novel, much less of a departure from nineteenth-century linguistics than we have been led to believe.
This raises the question of how much of what we like to think of as distinctively twentieth-century style linguistics really rests on nineteenth-century conceptual foundations. Let us consider, for example, the widespread notion that different languages have different basic grammatical structures, and that the analyst has to be careful to describe each language in its own terms. Bloomfield put the matter as follows:
"Our descriptions must be unprejudiced, if they are to give a sound basis for comparative work. The only useful generalizations about language are inductive generalizations. Features which we think ought to be universal may be absent from the very next language that becomes accessible. Some features, such as, for instance, the distinction of verb-like and noun-like words as separate parts of speech, are common to many languages, but lacking in others. The fact that some features are, at any rate, widespread, is worthy of notice and calls for an explanation; when we have adequate data about many languages, we shall have to return to the problem of general grammar and to explain these similarities and divergences, but this study, when it comes, will not be speculative but inductive."
Bloomfield is claiming that since it is not the case that all languages share the same set of grammatical categories the analyst must approach each language with an open mind and record what categories it actually possesses. After this same process has been carried out for all accessible languages, the results of these separate descriptive efforts can be compared and generalizations made on their basis. But the underlying assumption is that languages do not in fact agree in basic grammatical structure. For if they did, the same categorial structure would be discovered no matter which language was examined, and hence it would not matter how many were described. Significant generalizations would not have to wait until all accessible data were collected. In other words, the analyst could proceed immediately to the problem of explaining the one and only categorial structure. But this, I take it, Bloomfield did not believe to be the case, and it seems that most twentieth-century linguists would agree with him on this point.
But let's have a look now at a typical representative of nineteenth-century linguistics, namely August Schleicher. You will recall that Schleicher introduced the use of starred forms | [p. 418] into Indo-European linguistics and was responsible for what has since been referred to as the Stammbaumtheorie, the family-tree theory. His most influential publication was a compendious comparative grammar of the Indo-European languages, several editions of which appeared between 1861 and 1876. In 1865, three years before his death, Schleicher published a paper entitled "The distinction between noun and verb in phonetic form" (Die Unterscheidung von Nomen und Verbum in der lautlichen Form).
In his article, Schleicher defended the thesis that noun and verb are separate parts of speech only in the languages of the Indo-European family. The distinction between these two parts of speech is therefore not a universal feature in all human languages, but rather a peculiarity found only in certain ones. Schleicher bases this conclusion on the assumption that a grammatical category does not exist unless expressed by a phonetic form. This is so because phonetic form is the only observable embodiment of content or function. A linguist has therefore no right to posit functions in languages where their existence is not indicated by phonetic forms. Schleicher therefore lays down as a fundamental postulate the inseparability of form and function.
To defend this position he then resorts to what he takes to be a reductio ad absurdum argument. Deny the inseparability of form and function, says Schleicher, and you have to assume that all languages have the same functional system. But how can such a universal functional system be arrived at, how can a linguist infer what this system is like, given the fact that phonetic forms are all he can observe? Moreover, the observable systems of languages so far accessible to linguists are in fact incompatible with one another. Consequently, the ideal system could not be arrived at by simply adding together all the grammatical functions found to exist in the languages of the world. But if summation is rejected as a method of determining the ideal schema, then it will have to be set up completely a priori. That is, however, not possible because a priori procedures are ruled out in an empirical science such as linguistics.
To show that summation is not practicable as a method of inquiry in this area, Schleicher considers the category of gender. Some languages have the three genders familiar to classical scholars; others have three genders, but not the same ones as we find in Latin and Greek; others have only two genders, or a distinction between animate and inanimate, and there are languages which have a multiplicity of genders. From all these facts how can one possibly determine which gender distinctions belong to the essential nature of language and which are accidental?
Once you claim, for example, that the animate-inanimate distinction is an essential one, you are forced to assume that native speakers of German are just as aware of the distinction as, say, speakers of Slavic languages in which the distinction is phonetically represented. But in fact speakers of German feel that an animate noun like Sohn | [p. 419] 'son' and an inanimate noun like Baum 'tree' belong together because they are both masculine nouns. In other words, speakers of a language are unaware of linguistic categories which their language leaves unexpressed.
Schleicher's general position then is that categorial structure is directly inferrable from phonetic form and cannot be inferred from any other evidence. But phonetically marked categorial distinctions do not correspond from language to language. It follows therefore that there is no such thing as an ideal system inhering in all languages.
Though Schleicher does not allude in this article to philosophical grammar of the kind which had been in vogue up to the middle of the nineteenth century, it is clear that we have a veiled critique of that type of literature here. In any case, Schleicher had explicitly attacked philosophical grammar some years before in the introduction to his Lithuanian grammar. This is what he said:
"I know of nothing more unpalatable than attempts at philosophizing in grammar. Even syntax can and must be as comprehensive a description of linguistic usage as possible, arranged perspicuously, accessibly and in a fashion appropriate to the subject matter. More general points of view do not belong in grammar, if indeed so-called philosophical grammar has any justification whatever at the present time. In addition to the fact that such things are out of place in a grammar and merely interfere with the plan of arrangement, philosophical grammar is completely mistaken, the particular error involved consisting in deriving syntactic usage from general inner reasons (innere allgemeine Gründe), from the laws of human thought and the like. But very often, indeed in most cases, something demonstrated to be necessary in one language or in several turns out not to be the case in some other language."
Schleicher's position then is as follows: a grammatical category exists if and only if it is phonetically expressed. Different languages have different categorial structures. The kind of universal functional system posited by writers of philosophical grammar is to be rejected because it cannot be determined from observation and generalization on the basis of observation -- the one and only procedure permissible in natural science. Languages are organisms, natural objects, like plants and animals, and must be investigated with the same general method used in the other natural sciences: namely observation and generalization on the basis of observation.
The similarity between these notions of Schleicher's and those expressed by Bloomfield in his monograph Language is so striking that I shall dispense with a detailed comparison. I am aware that Schleicher is not customarily considered to have made any contribution in the area we have been considering. Historians of linguistics depict him in a different light, in fact in a rather negative way usually, but this is because they continue to see him through | [p. 420] the eyes of his immediate successors, the linguists of the Neogrammarian generation, who rejected so many of his ideas that they were unable to realize how indebted they were to him for many of their basic presuppositions about language. Bloomfield, as it well known, was trained by the Neogrammarians and thus must have received these ideas directly from them.
What I should like to establish then is a continuity of ideas from the mid-nineteenth century until the present. It is also easy to show that similar ideas are to be found in many of the major theoreticians of the first half of the nineteenth century, including Wilhelm von Humboldt. What seems to be novel in Schleicher's approach is the interest in what he conceives to be scientific methodology and the rather narrow conception of linguistic category which goes with it. In Humboldt, on the other hand, there is an emphasis on the individuality of each language, and the notion of an ideal categorial schema which particular languages strive to attain. But the basic tenet that the differences between languages constitute the main problem of linguistics is due to Humboldt, if not to his predecessors. But to show that this is the case will require another paper at least as long as this one.