The beginnings of full-fledged market society in the early part of the nineteenth century coincided with and were so to speak emblematized by a new systematization of linguistic knowledge. Wilhelm von Humboldt’s posthumous 1836 Ueber die Verschiedenheit des menschlichen Sprachbaus und seinen Einfluß auf die geistige Entwicklung des Menschengeschlechts [The Diversity of Human Language-Structure and Its Influence on the Mental Development of Mankind], published most recently in translation as On Language (Cambridge, 1988; all references here are to this edition), was intended as a theoretical introduction to Humboldt’s incomplete magnum opus on the ancient Kawilanguage of Java. Humboldt’s work grounds comparative language study in a pre-Darwinian linguistic structuralism.

Parallels between various Indo-European languages, including Sanskrit, had been noted as far back as the 16th century (see Daniel Droixhe, De l’origine du langage aux langues du monde : études sur les XVIIe et XVIIIe siècles, Tübingen: G. Narr, 1987); in 1786, William Jones’s third “Anniversary Discourse” to the Asiatic Society sketched a first picture of the Indo-European family tree as a whole. Languages could no longer be cast into vague categories such as Rousseau’s “northern/southern”; their historical relationships were beginning to be analyzed in detail. The establishment of the Indo-European tree in turn inspired interest in languages outside it: what other trees would it be necessary to construct? how many families, how many types of languages were there? Of the growing number of students of language in the first half of the nineteenth century, Humboldt was the one most deeply concerned with these questions. Fluent in a dozen languages, he is said to have studied over three hundred. Although Humboldt’s philosophy of language is filled with romantic mist, his work in classifying languages into families and above all in creating a general linguistic typology (agglutinating, incorporating, isolating, inflected) shows him to be no longer a philosophe and already a professional linguist.

In the later part of the nineteenth century, under the influence of Darwin’s Origin of Species (1859), such disciplined social scientists as Durkheim and Tylor (not to speak ofMarx and Engels) would classify cultural phenomena in terms of their closeness to the “origin,” albeit generally in the absence of hypotheses of origin as such. Written at an earlier time, Humboldt’s work exhibits a flowering of synchronic dominance that would be reasserted by Saussure only well after the turn of the twentieth century. This post-Darwinian turn away from historicism, suspected of racism or at any rate of “Eurocentrism” avant la lettre, produced a more rigorous structuralism than Humboldt’s, one purged of the value-judgments that had made his variety, as we shall see, the intellectual basis of the Aryan “race” theory proposed by Max Müller whose virulent offshoots contributed to the discredit of historicism.

Humboldt’s structuralism is the result of the vast accumulation of empirical knowledge that separates him from his eighteenth-century precursors. The more one knows about languages in the concrete, the more obvious is the barrier between the origin of language and its present state, and the more obvious as well, the difference in kind between concrete descriptions of language forms and families and unverifiable speculations about language origin. It is this dichotomy that underlies Humboldt’s statement, itself wholly speculative, that:

Language, indeed, arises from a depth of human nature which everywhere forbids us to regard it as a true product and creation of peoples. It possesses an autonomy that visibly declares itself to us, though inexplicable in its nature, and, seen from this aspect, is no production of activity, but an involuntary emanation of the spirit, no work of nations, but a gift fallen to them by their inner destiny. . . . It is no empty play upon words if we speak of language as arising in autonomy solely from itself and divinely free, but of languages as bound and dependent on the nations to which they belong. (sec 2, p. 24)

Rather than seek the path from language to languages, from the “involuntary emanation of the spirit” to the empirically known “work of nations,” Humboldt insists on their incommensurability, not because has less insight than Condillac but because he has much more knowledge, which he can articulate in a way qualitatively more concrete and verifiable than any discussion of language’s roots in the “depth of human nature” could possibly be. Humboldt’s refusal to speculate on the origin of this “spirit” is the beginning of the negative form of minimalism that will be reflected by the Académie des Sciences’ famous 1866 ban on such speculations.


The chief difficulty encountered by the modern reader of Humboldt’s treatise is in following its train of thought. Where a contemporary comparatist such as Joseph Greenberg is content with outlining diverse linguistic structures in order to argue for one or another configuration of one or another linguistic tree, Humboldt wants his typology to tell a “story” that is nevertheless not strictly chronological. The breadth of linguistic knowledge that takes him far beyond the Indo-European and the related Semitic family makes him–quite rightly–wary of postulating any kind of historical progression from one family of languages to another:

One might certainly suppose . . . a gradual progression [from the Chinese to the Sanskrit/Indo-European languages]. But if we truly feel the nature of language as such, and of these two in particular, if we reach the point of fusion between thought and sound in both, we discover there the outgoing creative principle of their differing organization. At that stage, abandoning the possibility of a gradual development of one from the other, we shall accord to each its own basis in the spirit of the race, and only within the general trend of linguistic evolution, and thus ideally only, will regard them as stages in a successful construction of language. (sec. 4, p. 32)

Humboldt posits an “evolution,” a set of “stages in a successful construction of language,” but this evolution is “ideal” rather than historical. The relation between Chinese and Sanskrit is not one of linear filiation but of differentially successful offshoots from a common stem, like that between Homo neanderthalensis and Homo sapiens. If Chinese were an earlier stage of Sanskrit, we could assume its speakers could evolve into speakers of the “more successfully” articulated tongue. But they are not; they incarnate a different, and inferior, creative principle, one that has its own unique virtues and capacities but can never acquire the higher ones of the superior linguistic-national stock.

The clarity and precision of language gain through the habit of expressing enlarged and refined ideas . . . But this whole progress of improved language-making can only go on within the limits prescribed to it by the original design of the language. A nation can make a more imperfect language into a tool for the production of ideas to which it would not have given the original incentive, but cannot remove the inner restrictions which have once been deeply embedded therein. To that extent even the highest elaboration remains ineffective. (sec. 4, p. 34; emphasis here and elsewhere is the author’s)

Thus Humboldt’s justified disinclination to present a scheme of linear evolution leads directly (via Müller’s “Aryan invasion” theory) to the linguistically grounded racialism that would become so important later in the century.


Given his comparativist preoccupations, it is unsurprising that Humboldt has little to say about the ultimate origin of language. He associates language with “sociality” by means of a causal link reminiscent of Condillac, to which he adds the Herderian idea of language as free reflection (“mental cultivation”):

The individual man is always connected with a whole, with that of his nation, of the race to which the latter belongs . . . His life is necessarily tied tosociality . . . In the merely vegetative existence, as it were, of man on the soil, the individual’s need for assistance drives him to combine with others, and calls for understanding through language, so that common undertakings may be possible. But mental cultivation, even in the loneliest seclusion of temperament, is equally possible only through language, and the latter requires to be directed to an external being that understands it. . . . Man thereby at once discovers that around him there are beings having the same inner needs . . . For the intimation of a totality, and the endeavour towards it, are given immediately with the sense of individuality . . . (sec. 6, p. 41)

Language for Humboldt is, famously, not a product but an activity, not Ergon but Energeia (sec. 8, p. 49), whence its affinity with sound, which “streams outward from the heart’s depths (sec. 9, p. 55).” Language is essentially social, and linguistic exchange requires the reproduction in the interlocutor of the thought-processes encoded in the speaker’s utterance. But once this has been established, Humboldt finds himself obliged to deny the very words he had previously used to explain the origin of this exchange. The following passage is clearly aimed at Condillac:

Even the beginning of language should not be thought restricted to so meagre a stock of words as is commonly supposed when, instead of seeking its inception in the original summons to free human sociality, we attribute it primarily to the need for mutual assistance, and project mankind into an imagined state of nature. . . Man is not so needy, and to render assistance, unarticulated sounds would have sufficed. Even in its beginnings, language is human throughout. . . Words well up freely from the breast, without necessity or intent, and there may well have been no wandering horde in any desert that did not already have its own songs. (sec. 9, p. 60)

The same “mutual assistance” that produced sociality in the earlier passage is now shown to be insufficient to determine human language; calls for help could be handled at a pre-human level. Language (again echoing Herder) is a “free” activity; yet to define this freedom, Humboldt finds himself obliged to deny not merely “necessity” but “intent.” No doubt these are familiar quandaries that the previous century’s speculations could not resolve; what is new here is the author’s lack of real concern with resolving them. The originary hypotheses of the past are inadequate because they construct only a protolanguage, whereas, as Humboldt observes, “even the languages of so-called savages, who would have . . . to come closer to such a state of nature (ibid.)” are always already articulated. As for the origin of the “sound-form” within which this articulation takes place, it is dismissed in a gesture that Durkheim will repeat at the beginning of the following century with respect to religion:

The creation [of a sound-form], if it is to be a true and complete one, could hold good only of the original invention of language, and thus of a situation that we do not know about, but only presuppose as a necessary hypothesis. (sec. 10, p. 76)

No conceivable hypothesis can take us from the “original invention of language” to the creation of a (specific) sound-form. Humboldt the structuralist sees language as a “totality”; but there is no point that can serve as the origin of all the dynamic totalities that humanity has generated from the struggle between innere Sprachform and external sound-substance. (The only way of reconciling the specific and the general is to postulate an articulated Ursprache from which all languages would be derived. This ancient idea, revived in more recent times notably by Morris Swadesh [The Origin and Diversification of Language, 1971] and some tenants of the “Nostratic” hypothesis, is incompatible with Humboldt’s empirically driven but “ideal”–i.e., Platonic–hierarchical classification system.)

Articulated language is always already differentiating and differentiated. To go beyond the variety of “sound-forms” to the common human fact of language itself can be done only by hypothesizing a moment prior to articulation that is the origin of the difference that constitutes it. This hypothetical moment had haunted the speculations of the previous century and would return near the end of the following one; Humboldt’s still inchoate linguistic science derives its professional self-consciousness from its expulsion.