Posted: Dec 20, 2010 10:40 pm
by katja z
»The purest Sillian is spoken in the region of Dunts.«

»The purest English is spoken in the Appalachians/Inverness/on the BBC.«

»The purest French is spoken in the Loire valley.«

»The purest Spanish is spoken in Salamanca/Lima.«

»The purest Portuguese is spoken in Coimbra.«

This is a small selection from the collective wisdom of the internet. The list could go on ad nauseam, until we have run out of natural languages spoken today – and given that there are about 5,000 to 6,000 of them (the exact count depends on the definition of language, which is not nearly as clear-cut as most people imagine), this might take quite some time. Before undertaking such a mammoth task, then, it is worth asking whether this kind of common-sense statements actually mean anything. It would appear that the common sense says yes. I will argue that it may be common, but sense it is not.

Defining the myth

»Purity«, in fact, is one of those mythical beasts which everybody sort of knows about, but which nobody can really put a finger on. Possibly for fear of leaving dirty fingerprints? Be that as it may, as irreverent rationally inclined thinkers we must insist on prodding it a bit and see what happens. First, we need to ask what people might refer to when they make judgements about language purity, even though they are often vague on this themselves. Secondly, we will turn to the reality of language use and check these judgements against it.

Some years ago, the following question was asked at the WordReference forum: »I would also like to know what region is considered to be the zone in which the 'purest' English is spoken.« When challenged to explain his meaning, the poster added: »Every single language has a place where the language is supposed to be spoken best. (I'm not saying that I agree with this statement, but it's common knowledge, and it almost always coincides with the place where the language was born.)«

The two senses of the word »pure« put forward in this explanation, »purity« as referring to quality and »purity« as referring to origin, are not unrelated, but I find it useful to try and consider them separately. The first one is straightforward enough. My Collins dictionary says that »pure« can refer to »a perfect example of its type« or something that is »produced or practiced according to a standard or form that is expected of it«. In this sense, the purest Sillian is the most perfect form of Sillian, characterised by correct grammar, appropriate vocabulary and, very importantly, good pronunciation. The opposite of »pure« in this case would be »bad«, »broken« or »corrupted«. The second sense is somewhat akin to the concept of racial purity: pure as purebred, unhybridised, unmixed: the Sillian spoken in Dunts is pure in so far as it is true to the original Sillian, to its spirit and its true nature. The opposite of »pure« in this sense would be »hybrid«, »bastardised«. In both cases, the assumption is that there is one way to speak Sillian properly; the forms of the language that most people use are deviations from or degradations of this one proper template: a linguistic treason.

I will first try to explain just why this notion is silly, and not only for my newly invented language, Sillian, by drawing on sociolinguistics and historical linguistics; in conclusion, I will point out why you do not have to be a dunce from Dunts to fall for this particular myth.

The bogeyman of good language and linguistic variation

For any language, at any point in its history, there are a number of ways of using it. In sociolinguistics, we speak of lects: regional (dialects) and social (sociolects) linguistic forms used by specific speech communities, defined by factors such as social status, age etc. Given the definition, it comes as no surprise that lects are not discrete homogeneous forms either. They could be better described as only partially-overlapping clusters of linguistic features that shade into other lects: linguistic variation is often clinal. Lects may differ in any or all of the following: pronunciation (»accent«), choice of vocabulary, grammatical features, idiomatic expressions, pragmatic conventions etc. These language varieties are constantly-changing products of the speech communities' communication needs, which may include their identity strategies.

Well and good, but what is the relation between lects and what I will, for want of a better word, continue to call a language? How do we get from what is spoken in a Yorkshire village, in inner-city Lagos, or by members of the Queen's English Society, to »English«; from the speech of youngsters in a Paris banlieue to »French«? For analysing the emergent global system of world languages, a gravitational model has been suggested, in which languages are linked among themselves into regional constellations of influence by bilingual speakers (De Swaan, in Calvet 59). The same model can be used for micro-linguistic relations: lects are interlinked through »multilectal« speakers, whose communication practices make up the dynamic network that is their language. For depending on factors such as age, gender, profession, social background and social trajectory, each speaker has their own repertoire of a number of linguistic varieties that they use in different communication contexts (as well as some unique features, their own signature style, which is referred to as idiolect). In this sense, almost no-one speaks only one language, and no two »native speakers« speak the same language; and what is typical of a certain language across its varieties is a statistical effect of concrete language practices, not an effect of an underlying One True (or Pure) Form.

What is codified in grammar books and general dictionaries is not the whole spectrum of language practices, but the standard language. From the sociolinguistic point of view, this is just one lect among others. It is not even anybody's mother tongue (even though some sociolects are much closer to it than others); it is the variety we typically acquire through formal education, and use it in formal contexts: when applying for a job, writing a newspaper article or a novel, presenting our work at a conference, speaking on the television. But it is hardly the most appropriate form in all communication situations. If Eliza Doolittle went for a beer with her childhood friends and insisted on using her hard-won standard English, the effect would not be one of language purity or correctness, but of artificiality – and, very probably, arrogance and snobbery as well.

How a standard language is born is a complex story, one that, for reasons of space, I can only begin to tell. But the beginning, even simplified, is instructive. A standard language is typically based on a prestigious sociolect such as the language of a court (»the Queen's English«, »le français du roi«), or other influential social groups, typically city dwellers. It begins as just one way of speaking among others in a continuum of mutually intelligible dialects, but as it gathers influence, it begins to influence this continuum. Cities and/or courts, as centres of economic activity, political power and social prestige, act as gravitation centres for populations; and their dominant linguistic forms, therefore, act as gravitation centres for their plethora of lects, becoming the standard of »good« language practice, with far-reaching consequences for linguistic practices and representations (that is, ways how people think about different language varieties) within their spheres of influence. Depending on the geographic distribution of such centres, and on the identity politics of populations, such a dialectal continuum can be cut up into several chunks which are then labelled as distinct »languages«. Take, for instance, the Western-Iberian dialectal continuum, which shades into Portuguese on one end, Castillian on the other (and which is itself part of a wider continuum of Western Romance languages): nowadays, several other languages (Galician, Mirandese, Leonese, Asturian) have been recognised in the region, and the emergence of their new standards has restructured the perception of which dialect »belongs to« (and conversely, »deviates from«) which language. These are extremely difficult questions if you insist on the notion of languages as discrete entities, each with its own ideal type – but they are artefacts of a flawed perspective. In reality, there is no one inherently »correct« way of cutting up the continuum of real-life linguistic practices into »languages«, only more or less practical ones. But this more precise model of language variation means that we have just thrown any notion of pure Sillian out of the window.

We are left with the concept that it is extralinguistics factors that determine which central, prestigious variety the speakers of a particular dialect will gravitate towards, subject to shifting social and political landscapes. It is these factors, then, that account for the integration of speech communities into a wider sociolinguistic community which comes to be dominated by a common standard variety. It is also these extralinguistic factors that influence language representations. There is a considerable body of research showing that social attitudes towards particular social groups are translated into judgements of the correctness and/or aesthetic qualities of their language (cf. Preston and Giles & Niedzelski). We learn how to think about the various social dialects at the same time as we learn to use them – or even before and independently of that: many people who are unable to use the variety socially accepted as the standard, or »best« form of their language, will still use it (or their idea of it) as the yardstick to evaluate their own linguistic performance, a situation referred to as linguistic insecurity.

The spectre of authenticity and language evolution

By now it should be at least partly clear that the supposed original purity of Sillian – which I will refer to as »authenticity« – cannot fare any better under critical scrutiny as the purity of its form. I have already touched upon the subject of language evolution, but now is the time to take a more decided plunge into the subject.

On the face of it, language evolution seems straightforward enough. Modern English comes from Old English (with a messy episode of Norman French influence that complicates the story) and that, in turn, comes from the Proto-Germanic. French and Spanish and the rest of the Romance languages come from Latin. (The FSM knows where Sillian comes from.) In fact, it is not as simple as that; if it was, many a historical linguist would be out of their job. To start with, there is no single point in space and time at which a new language is born. The sudden emergence of languages in recorded history is an artifact of sparse evidence. This is not surprising, since spoken language does not fossilise, and until very recently, the use of writing (itself a relatively recent invention) was extremely limited. This also meant that only the most privileged language varieties ever made it into writing: the written record, far from reflecting the whole variability of language use, is heavily biased.

The metaphor of a branching tree that has often been used in historical linguistics is misleading. It disregards the basic fact that has already been stated: in any language, at any point in its history, there are a number of ways of using it. These linguistic practices keep changing as populations come into contact with each other, blend, shift to another lect (of the same language or of another one, in so far as the distinction makes sense), selecting or rejecting competing linguistic features in the process and thus gradually restructuring their language (cf. Mufwene). When an innovation appears in the written language, we can be almost certain of two things: that this feature had existed for some time in the language in competition with other features, spreading slowly and gradually becoming accepted in the prestigious dialect; and conversely, that the inclusion of this innovation into the standard language does not mean it has won acceptance in all of the lects within its sphere of influence.

So the evolution of any »one« language is not unilinear. This is not even true of its standard dialect. Thus, within the range of linguistic forms that we call Old English there were at least four prestigious varieties, corresponding to the centres of independent kingdoms. West Saxon became the dominant written form following the political unification under Alfred the Great, but lost this status after the Norman Conquest. The new standard written form which emerged in the 15th century, the so-called Chancery Standard, was not directly related to the late OE standard; rather, it was based on the language spoken in London (and specifically, at the royal court, where the use of English had begun to predominate at the time of King Henry V).

What, then, might be considered the historically pure origin of English, its authentic original form? Shakespeare is often credited with writing the »purest« English, but is that a logical choice? His Early Modern English, after all, is the result of a strong admixture of Norman French. Surely the Late West Saxon, which was used to record Beowulf, is the purer form? Or even some earlier stage in the continuous process of linguistic change, one uninfluenced by the Old Norse – but which? Or maybe we should look, not (only) further back in time, but (also) elsewhere, to nonstandard varieties, many of which have conserved traits of Old English that have disappeared from the standard dialect? Thus, when trying to define a »truly authentic« form of a language, we get caught in a truly nightmarish form of endless recursion. With language, it is not only turtles all the way down; it is also turtles all the way sideways.

Once again, the seemingly self-evident judgement of linguistic purity is based on purely extralinguistic factors. The two most important ones in this case are cultural prestige and the constructed national history. This is why no-one today would seriously argue that proto-Indo-European was the golden age of their language – as opposed to the Elizabethan era. But the emergence of modern nations as imagined communities (Benedict Anderson), and the role played in this process by the teaching of standard language and of literary canon, makes for a longer and more complicated story than can be told here.


Both senses of the word »purity« that I have identified rest on the notion that languages are discrete homoneneous entities. The continued hold of this notion is not simply the result of linguists' failure to communicate their findings to the general public (as pointed out by Bauer and Trudgill). For one thing, a large part of 20th-century linguistics built on just this assumption: this is true of both Saussurean linguistics, which treats language as a relational system, and of Chomskian linguistics, whose subject is »the ideal speaker-listener in a completely homogeneous speech-community« (Chomsky 3). These abstract approaches do have their uses (Haugen called this model a »useful fiction«; Calvet 10). They cannot, however, even begin to account for the complexities (or messiness, if you prefer) of real-world communication in natural languages; you might as well try to explain evolutionary adaptation without taking into account the ecology of a given species. What they get wrong is the very simple fact that »language« is an abstraction extracted from human communication practices – and not vice versa (cf. Calvet 6).

It is, however, true that for most people, their early encounter with prescriptive grammar in school remains their only contact with any kind of a formal approach to language – despite the fact that modern linguistics is overwhelmingly descriptive (that is, it attempts to describe how language functions, not legislate on how it should function). The prescriptivism informing language teaching in school naturally feeds into the purity myth: children are taught the correct way to use the language, with the implication that what they have learned to speak natively and continue to use in informal everyday communication is an imperfect form of this language, fallen from the grace of its original purity. The use of words such as »corrupted« to describe nonstandard dialects only serves to strengthen this perception.

And finally, it must be admitted that the kind of statements quoted at the beginning do refer to something very real – to an important aspect of the speakers' social reality. It is true that in every linguistic community, there are language varieties which are socially more desirable than others, for as we have seen, language representations are intimately linked with social attitudes. This is what makes myths about language so incredibly hard to shake off. On one level, they may be based on a misconception. But on another, as social constructs, they are part of the social and linguistic reality. As such, linguistic representations and attitudes have very real effects indeed, both on the speakers' (self-)perception and (self-)evaluation, and on linguistic and education policies.


Bauer, Laurie and Peter Trudgill (eds). Language Myths. ePenguin, 1998. N. pag. E-book.

Calvet, Louis-Jean. Towards an ecology of world languages (tr. Andrew Brown). Cambridge, UK, and Malden, US: Polity, 2006. Print.

Chomsky, Noam. Aspects of the Theory of Syntax. Cambridge: The MIT Press, 1965. Print.

»The Cradle of English«. WordReference. Web. 16 Dec. 2010.

»Dialect Continuum«. Wikipedia. Web. 16 Dec. 2010.

Giles, Howard and Nancy Niedzielski. »Italian is Beautiful, German is Ugly«. Language Myths. (Bauer and Trudgill, eds). E-book.

»Middle English.« Wikipedia. Web. 20 Dec. 2010.

Mufwene, Salikoko. »Population Movements and Contacts in Language Evolution«. Journal of Language Contact, THEMA 1 (2007). 63–92. Web. 3 Dec. 2010.

Preston, Dennis R. »They Speak Really Bad English Down South and in New York City«. Language Myths (Bauer and Trudgill, eds.). E-book.

»Pure«. Collins Cobuild English Dictionary. London: HarperCollins, 1995. Print.

»Variety (Linguistics)«. Wikipedia. Web. 16 Dec. 2010.