Typology – a new task of linguistics

James McElvenny
University of Edinburgh

In lieu of an introduction

Hypologie – title pageBelow I offer an English translation of the last essay Georg von der Gabelentz (1840–1893) personally submitted for publication, “Hypologie [Typologie] der Sprachen, eine neue Aufgabe der Linguistik”.

This essay is interesting for a number of reasons, not least of all because it would seem to be the first published use of the term Typologie (typology) in a linguistic sense. The term is in fact rendered throughout the published article as Hypologie, but from other sources we can be certain that this is a typographical error, which Gabelentz was unable to correct because he passed away before the final proofs of the article reached him. That the editors of the journal did not notice the error attests to the novelty of Gabelentz’ term.

As in so many of his writings, the program Gabelentz outlines in this essay combines traditional, in many ways outmoded, concerns with innovative, hyper-modern proposals. On the one hand, Gabelentz revisits key themes from nineteenth-century language classification in a conceptual framework that appeals to notions of “race”. On the other hand, he seemingly anticipates ideas and methods that would only come to the fore in twentieth-century efforts at language typology.

Since this is a preliminary blog post, the translation is given here without a detailed introduction or extended notes. These will come with a later publication. There is, however, no shortage of existing papers that treat Gabelentz’ essay in context. I would recommend:

McElvenny, James. 2017a. Georg von der Gabelentz. Oxford Research Encyclopedia of Linguistics. (Green open access: Humanities Commons, academia.edu)

McElvenny, James. 2017b. Grammar, typology and the Humboldtian tradition in the work of Georg von der Gabelentz. Language and History, 60:1, 1-20. (Green open access: Humanities Commons, academia.edu)

Plank, Frans. 1991. Hypology, Typology: the Gabelentz puzzle. Folia Linguistica, 25: 421-458. (Green open access: author’s website)

The reference for Gabelentz’ original article is as follows:

Gabelentz, Georg von der. 1894. Hypologie [Typologie] der Sprachen, eine neue Aufgabe der Linguistik. Indogermanische Forschungen, 4: 1-7. (Available at archive.org)

I would like to thank Manfred Ringmacher for his comments on my translation, which have helped me to improve it.

– James McElvenny

Typology of languages, a new task of linguistics

If linguistics is to answer the question of where the differences in human language structure come from, then the first thing that it will do, and which it has already been doing for some time, is this: it selects the most striking structural types – structural styles – and analyses, characterizes them. It explains the meaning of every peculiarity noticed according to the principle of idem per idem [1] – that is, it translates the linguistic phenomena back into the psychological, and in this way attains overviews of races and nations and tests the validity of these overviews against what is known from ethnology and history. This method would contain the greatest guarantee of certainty if it did not have to reckon with so many disturbing forces, which so often escape all observation and exact consideration. No one has made keener and deeper use of this method than Byrne (Principles of the Structure of Language).[2] But in my opinion he has paid too little attention to the disturbing factors, set the unknown values in his equations to zero, and at times he works with concepts that are a little too loose. Such a thinker is particularly susceptible to all these dangers. I hoped to be able to avoid them in my Sprachwissenschaft by working at a less fine-grained level, by sticking to the most tangible details: on the side of languages, to some of their most conspicuous physiognomic traits and, on the side of nations and races, to the broadest masses and to those who have presumably been subject to the same living conditions for the longest time and so have developed certain lopsided mental and temperamental dispositions to the greatest degree. It was only ever an attempt, but I still believe today that I did not go far enough rather than too far. I knew from the beginning that there were a lot of details to add; a substantial book could have been made out of the one chapter in my Sprachwissenschaft,[3] but not an exhaustive one like that which Byrne sought to deliver.

Was the path already taken really the only one that is passable? And if it is not: are there no others that are just as certain, perhaps even more certain?

The forces through which languages are shaped are of two kinds, which are distinguished by their origin: native and foreign. Under the native forces I understand all those and only those that are rooted in the language community itself, in the nation. I know that many would like to further distinguish here between inherited disposition and the upbringing in which the native soil itself plays the role of teacher, but these can rarely be separated from one another.

Nevertheless, this issue still deserves consideration, since mass migrations occur frequently in human history. Here we have a disturbing factor, which in addition will in most cases be uncontrollable and prehistoric in origin. But I do not want to overestimate the influence of this factor, since in nations and languages the distance between the inherited dispositions and the new tasks of life, when these are achievable by other means, seems to balance out in not too long a time. Both the nation and language would have to be in bad shape if they did not contain hidden forces in themselves that only await stimulation in order to awake from their slumber. But in both nation and language fully developed forces can also fall asleep if they have not had the opportunity to be exercised for a long time. We know of language families of very uniform type and then others with an astounding diversity of structural forms, and in both cases we believe we can discern the reason for this. Running parallel to the unity or diversity in human language structure is, on the ethnological-historical side, the essential similarity or difference of a nation’s living conditions. The ancient languages are not always the best witnesses here, and that kind of research which drills down to the proto-languages does not bring the most usable material to the surface. This research may try to reconstruct an image of the proto-nation from the proto-language, but it would simply go around in circles if it would seek to elucide the proto-language on the basis of the proto-nation. On the other hand, this approach is correct when it measures the racial disposition against the proto-language; the racial disposition has developed in different directions in the individual nations and yet it has done so through basically the same driving and pulling forces. But even this may be limited; the common characteristics of a family are perhaps only significant when they have been retained in living form. When Latin gives up the Indo-European vowel grades except for a few scanty remnants, when its daughters lose their case endings, and when Modern Persian and, in strangely different ways, the Slavic languages have altered the old accent system, when in the Indo-Chinese family polysyllabic and monosyllabic languages,[4] isolation, agglutination and inflection, and the most diverse forms of sentence structure have been developed, then we know that on this point the linguistic and racial type has shown itself to be flexible. We may then ask further what has caused this flexibility, and if history does not answer this question, if it does not reveal how much is to be attributed to the new homeland or to the disturbing or encouraging influences of the neighbouring nations, then we have at least gained negative knowledge.

Since our science, above all thanks to Hugo Schuchardt and Lucien Adam, has turned its attention to those new-born mixed languages, we can also make use of language mixtures for our purposes. Of course, least suited to our purposes are those deformed products of international business communication in which no nation has made itself mentally at home or lived its mental life to the full. I must exclude from my research whole groups of languages that seem to me suspicious in this respect, such as those of the Melanesians, those of the Gold Coast of Africa, and others. But we also know that this was the beginning of some languages that today count among the best. Here the bastard was made legitimate through subsequent marriage, and in and with the new nation grew a new language. And so we have once again that which is important to us: free structures that have grown on native ground.

Consequently, the material we have for undertaking our research is actually richer and more reliable than it appeared at first sight. The languages of civilized nations, however, with their power developing in all directions, are less suitable objects of research than the languages of lower races, with a lopsided upbringing and disposition. That’s enough for initial orientation.

If I disregard those foreign disturbing influences – as long as they remain foreign and a source of disturbance – and disregard as well those languages in which I suspect such influences, then I may say the following of the languages left over: they are free organic structures, and because they are, and inasmuch as they are, all of their parts stand together in a necessary mutual configuration. This makes sense a priori, and cannot be otherwise, and yet a lot is said by it. From the same mental disposition, the same historical conditions emerges everything that a language is and has: its phonology as much as the structure of its words and forms, its sentence structure and the national style, the grammar and the vocabulary.

We may put this forward as a thesis and be sure of general agreement. It also makes sense that certain traits in the physiognomy of languages, above all lexical, stylistic and syntactic features, are especially characteristic. If we go further, if we want to imitate the gypsy fortune teller, who sees in the palm of the hand the whole person, or Cuvier, who in his mind reconstructed from a single bone the whole animal – if we measure the theory against the facts, then it soon seems as if we had only the sad choice between either immediately declaring bankruptcy or artificially inflating the value of our stocks through trickery until bankruptcy results by itself.

But we cannot renounce a necessary idea that must ultimately be correct simply because the first attempt to realize it was unsuccessful. We must clothe the idea in a controllable form, and no form is more controllable than the statistical.[5] It is here that I would like to see the work begin. If above I spoke of cases where languages of one and the same family have taken on a very different character, then I would now like to turn my attention to those other cases, where languages of different stocks have converged to exhibit related traits. In this way, I was able in my Sprachwissenschaft to compare the Malay languages with the Semitic languages in terms of their syntax[6] – I could have also mentioned some correspondences between them in word formation and morphology. That neighbouring peoples of different stocks often share features in their phonologies has been observed frequently, and even by me, and may be amenable to explanation. But it seems rather mystical that in China and in the Trans-Gangean Peninsula (i.e. continental South-East Asia) three otherwise polysyllabic and agglutinating language families, Indo-Chinese, Kolarian and Malay, have monosyllabic-isolating members, while Chinese, Burmese and Siamese (Thai) with its group manifest such a fundamentally different mental character from other languages in the Indo-Chinese family. And Siamese, for its part, is so similar to the unrelated Annamite (Vietnamese) that it is as if the same building plan were here carried out in granite and there in sandstone.

Hardly less astonishing is another phenomenon that can be observed, when two physiognomic traits that seem to have absolutely nothing to do with each other re-appear together at different places in the world of languages. Let me offer the most compelling example of this kind that I know. The Basque language in Europe, Tibetan in Asia, Greenlandic and its relatives in America and the languages of black Aborigines in Australia are different enough from one another in their grammatical structure. But they have in common two otherwise rare traits: first of all, instead of a subject and object case, they have an active-instrumental and a neutral-passive case, so that – to put it in our terms – the object of an active (i.e. transitive) verb and the subject of a neutral (intransitive) verb on the one hand and the subject and the instrument of an active (transitive) verb on the other always appear in the same case.[7] Secondly, they all distinguish sharply between the two kinds of adnominal attribute in that they put the genitive before its head and the adjective after the head. Expressed statistically: A goes with B in ¾ of cases — B with A in perhaps ⅗ or ½ of cases; the correlation is not necessary, but it is more frequent than we would want to attribute to chance alone. We may suppose that we are on the trail of two sympathetic nerves, which do not work together completely regularly, and now we would like to know the place where and how they are connected as well as the reason why this connection is sometimes disturbed.

We should probably be prepared for such occasional disturbances everywhere, and certainly for numerous formulae that say: the phenomenon A appears with such and such a probability with B, C, D, etc, but rarely with E, never with F. And these are the statistics that I am looking for. The question arises: can this already be achieved? and what is achieved by it?

That this can be achieved with the means available to us today is something that I can safely guarantee. Drafting the plan should not be too difficult at all, and there is a rich supply of literary aids; the only effort required is to pick them out. The study of languages conquers one province after another, and marks its rule through ever more competent textbooks. Indeed, the edifice must be erected on a broad base; every important type must show to its best advantage, and for every language a well legitimated representative must be available for questioning. Little would be achieved in this respect with schematic-mechanical extracts from miscellaneous – at times even mediocre – grammars. But as large a number as possible of the most diverse languages must be represented, more than a single person can command. This work requires a commision, and the commission requires a program that covers the smallest details, and this program requires more self-effacing obedience than we can expect from most scholars. But such difficulties are to be overcome.

Under this program I imagine a kind of questionnaire that exhaustively surveys category by category all grammatical possibilites, so that every question can be answered with either “yes” or “no”. Posing such questions is difficult for the questioner, and it is sometimes also difficult for the answerer to respond, but neither of the two is being asked for the impossible.

The second step is purely mechanical, the statistics of correlations, which I briefly demonstrated above. Through this technique we will arrive in an irreproachably exact way at the knowledge of truly typical traits. What we have long suspected, what I myself have tried to show in isolated examples in my book: those predominating tendencies that manifest themselves in the most diverse facets of the life of language, they will now really be calibrated according to their content and value – in numbers, as if they were weights and measures. The ground would now be cleared for a truly valuable calculation of probability: from a dozen known properties of a language we would be able to conclude with certainty a hundred others; the typical traits, the ruling tendencies would lie clear before our eyes.

Clear but still raw, as long as we can only speak of a correlation and not of a connection. Working out the connection is the third, highest task. And here linguistics will once again depend on ethnology and history; it will reach out from them and also towards them – it is the construction of a tunnel, undertaken simultaneously from both sides of the mountain. From one side will be explained: this is the character of the language, therefore this is the character of the national mind. From the other side will be concluded: these are the constant living conditions, these are the historical experiences, these the habits and cultural achievements of the nation, so its mental type must be like this. Here the picks from the other side become audible, unless our tunnels are off course. Once again that subjectivity of which general linguistics and its representatives is so often accused may manifest itself. But how far it is pushed back, how far the most objective data that we could demand, data that can be determined in calculable numbers, now reaches! If the work were to progress to the point of incontestable statistics then general linguistics would no longer need to envy the solid foundations of historical-comparative research. And if the work gets further, then we may see realized at the beginning of the twentieth century what the beginning of the nineteenth vainly tried to work out through speculation: a truly general grammar, completely philosophical and yet completely inductive.

Berlin. Georg von der Gabelentz.

This article had been typeset and was to be sent to its author for proofreading when the distressing news of his passing arrived. Dr. Hans Georg Conon von der Gabelentz, member of the Academy and professor at the University of Berlin, died on the 10th of December 1893.
– The Editors


[1] Idem per idem is generally known as the name of a logical fallacy involving a circular explanation. On this point, see Klaas Willems, “Phenomenological aspects of Georg von der Gabelentz’s Die Sprachwissenschaft“, in Gabelentz and the Science of Language, ed. James McElvenny (Amsterdam: University of Amsterdam Press, forthcoming). Here, however, Gabelentz uses idem per idem to refer to the doublesidedness of the linguistic sign. See Albrecht von der Schulenburg’s additions to Die Sprachwissenschaft: Georg von der Gabelentz, Die Sprachwissenschaft, ihre Aufgaben, Methoden und bisherigen Ergebnisse, eds. Manfred Ringmacher & James McElvenny (Berlin: Language Science Press, 2016 [1891]), p. 420.

[2] James Byrne, General Principles of the Structure of Language, 2 vols. (London: Trübner & Co., 1885). (Available at archive.org: vol. 1, vol. 2)

[3] Gabelentz is referring to his chapter on Sprachwürderung (evaluating language): Gabelentz, Sprachwissenschaft, pp. 409-502.

[4] The “Indo-Chinese family” Gabelentz refers to corresponds approximately to the Sino-Tibetan family as it is understood today.

[5] On Gabelentz and statistics, see Lia Formigari, “Wilhelm Wundt and the Lautgesetze controversy” (2018): https://hiphilangsci.net/2018/01/17/wundt-lautgesetze/

[6] See Gabelentz, Sprachwissenschaft, pp. 434-438.

[7] Gabelentz is of course describing what would today be called “ergativity”.

How to cite this post

McElvenny, James. 2018. Typology – a new task of linguistics. History and Philosophy of the Language Sciences. https://hiphilangsci.net/2018/09/05/typology/

Tagged with: ,
Posted in 19th century, Article, Europe, History, Linguistics, Typology
5 comments on “Typology – a new task of linguistics
  1. David Nash says:

    Note that 1886 is the first citation in the OED’s entry for the relevant sense (3) of typology (and the next is 1929), so GvdG’s use in 1893 (or before) was early if it had been in English. Is the word attested much earlier in German (in the sense of ‘The study of classes with common characteristics, …’)?

    • James says:

      I believe that the most common sense for Typologie in German when Gabelentz was writing was theological (i.e. equivalent to OED sense 1). Manfred Ringmacher has said this to me before. The 1895 Brockhaus entries for Typologie and Typus would seem to support that view. Here at archive.org.

      For good measure, here are some entirely scientific and representative Google n-gram searches: 1800–1900 and 1800–1950.

      There’s a sudden jump in the use of Typologie, in diverse senses, from the 1920s on. If you click through for the 19th century, though, you’ll be assaulted by theological texts.

  2. Joan Leopold says:

    I wouldn’t consider Google n-gram searches as “entirely scientific and representative”. When I have used them for words related to “culture”, I found that one had to research every entry because the dates of publication apparently get misrepresented.

    Joan Leopold

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: