Diglossia

Why do some communities keep two forms of one language and use each only in its own setting?

In much of the Arabic-speaking world a person may deliver a sermon or read the news in one variety of Arabic and then, walking out of the room, discuss lunch in a quite different one. A Swiss German speaker writes and is schooled in Standard German but converses in a dialect that a Berliner may struggle to follow. Neither situation feels unstable to the people in it, and both have lasted for centuries.

That should puzzle us. When two things do the same job, we expect one to win. Two currencies, two spellings, two rival standards — usually one drives out the other. So why do these two varieties not compete themselves down to one?

Begin by testing the assumption hidden in the question: do they do the same job?

Ask where each one appears. Charles Ferguson, who named the pattern in 1959, set out the division plainly. One variety — call it the high variety — carries sermons, formal writing, university lectures, political speeches, broadcast news. The other, the low variety, carries conversation at home, with friends, in the market, in folk stories and popular song. Now the crucial observation: the lists barely touch. There is almost no setting where a competent member of the community would be uncertain which to use, and using the wrong one is not merely odd but faintly ridiculous — a sermon in market speech, or a bargain struck in the language of scripture.

If that is right, then the premise of our puzzle is wrong. They are not two candidates for one job. They are two tools for two jobs, and a tool does not compete with a tool it never meets.

But hold on. Why should a language specialize in that way at all? Consider how each is acquired. The low variety is nobody's second language: children learn it at home, before school, without instruction. The high variety is learned formally, from teachers and texts, and is typically nobody's mother tongue — which is why speakers of it can be fluent yet still describe it as effortful. That difference alone would keep them apart, because a variety that must be taught cannot colonise the nursery, and a variety learned in the nursery arrives without the codified grammar and the inherited literature that formal settings demand.

Add the prestige asymmetry. Communities in this situation typically hold the high variety to be more beautiful, more logical, more correct — and will sometimes deny that the low variety exists as a language at all, while speaking it fluently all day. That belief is not incidental; it is what keeps the high variety attached to the settings that confer authority, and thus keeps the specialization from eroding.

Now we can see why the arrangement is stable rather than transitional. Each variety is protected in its own domain by something the other cannot supply: one has native acquisition and intimacy, the other has codification, prestige, and a written tradition. Neither can invade, because whatever advantage it has is irrelevant where the other one lives.

Which suggests a test. If specialization is what makes the arrangement stable, then blurring the domains should destabilise it. And that is what history shows. Greece maintained katharevousa against demotic Greek for well over a century, and the arrangement was formally abandoned in 1976 after mass education, mass media, and a nationalist politics of language had put both varieties in the same rooms doing the same work. Where domains stay separate, as in Swiss German, the arrangement persists comfortably.

Two honest caveats. First, the neat picture is Ferguson's idealisation; real communities show a continuum, with intermediate mixed varieties, and speakers move along it rather than flipping a switch. Second, the term has been stretched — Joshua Fishman extended it in 1967 to any two languages divided by domain, such as Spanish and Guarani in Paraguay, and there is a genuine and continuing disagreement about whether that broadening keeps the concept sharp.

The misconception worth dissolving is that the low variety is a degraded or careless version of the high one. It is not: it is a fully rule-governed system, usually the older line of descent in the community's ordinary speech, and it does things the high variety cannot — carry intimacy, humour, and everything a child learns before school. The hierarchy is one of assigned prestige, not of linguistic capacity.

It is also worth separating diglossia from two neighbours it is often confused with. Ordinary register variation — speaking more formally at an interview than at a barbecue — exists in every language and is a matter of degree within one variety. Bilingualism is command of two languages that may both be used anywhere. Diglossia is narrower than either: two varieties with a community-wide, conventional division of labour, one of them not natively acquired, sustained across generations.

And note what makes it stable is not tradition for its own sake but the absence of contested ground. Take the contest away and rivals coexist forever; create it, and one of them starts to lose.

• Ferguson's original four cases: Arabic, Swiss German, Haitian Creole and French, and the Greek pair.
• Fishman's extension to two distinct languages, and the argument over whether it overreaches.
• What ended the Greek case, and whether Arabic is drifting the same way under mass media.
• Code-switching: what happens at the seam, when speakers mix the two within a single conversation.

Every term the collection defines is gathered in the glossary.