Tip:
Highlight text to annotate it
X
Languages change over time, but how fast?
Let's say you fall into a coma hearing your language spoken all around you.
You snap awake in the same place but at a different time.
They're still speaking your language, but you can't understand a word.
How long were you out?
How long can a language last before it becomes unrecognizable?
Last time we met an eccentric Australian tongue called Dyirbal, a language where nouns can
be edible or harmful, where you have to talk completely differently around your in-laws,
and where speaking vaguely is considered a serious flaw.
And they don't even have a word for language!
At least, it used to be that way, because when linguists came back to visit the Dyirbalŋan
decades later, their language had changed.
Then we heard from Swadesh, who argued that, actually, no, all languages change at the
same pace, replacing core vocabulary at a constant rate, keeping over 80% of their most
basic words after 1000 years.
Swadesh then used his vocabulary lists and calculations to draw a line marking how long
we'd expect to wait before dialects of a single language evolve into separate tongues.
A line in the sand.
The elusive moment of change itself.
Spot any problems with this idea?
These researchers sure did.
They grabbed their Swadesh lists and pointed them at two sibling languages: Norwegian and
Icelandic.
The basic question they wanted Swadesh to answer was this: how many years had passed
between Old Norse and Norwegian, and how many between Old Norse and Icelandic?
Looking at word lists for replacement, plugging in that stable rate assumption, it looks like
Norwegian diverged 1000 years ago, not a bad estimate actually, while Modern Icelandic
is... 200 something years old.
In other words, no one should be speaking Icelandic right now.
Call up Iceland!
Your language is a mistake.
The numbers don't lie.
Today, thanks to debunkings like this, and the fact that Icelandic exists, historical
linguists accept that the rate of language change varies, and for various reasons.
According to this paper, one factor could be population size.
In a computer simulation, smaller populations borrowed and replaced words faster than bigger
communities.
The author of that Dyirbal grammar, Dixon himself, proposed another factor.
He called Swadesh's idea "the chimera of glottochronology" and argued instead that languages have long,
stable periods, but then change in punctuated bursts during moments of social upheaval.
So is that why Dyirbal changed so much so fast?
A small population and a quick punctuation?
Well, a careful look at the switch from Traditional Dyirbal to Young Dyirbal suggests another
variable, the biggest reason why the language changed so much in just decades:
language death.
A few decades before Dixon's grammar, about 30 Dyirbalŋan still lived around the upper
Tully River, hanging on to the old ways.
But these days, it seems like nobody is learning Dyirbal, threatening even Young Dyirbal with
extinction.
It's not just assimilating to Australian English or turning into a bare-bones pidgin or creole.
This researcher tells us it's not that simple.
Behind the steep change from Traditional Dyirbal to Young Dyirbal was a smooth oldest-to-youngest
dropoff.
Dyirbal has been dying its own death, and language death is its own variable.
With all these complexities, how long do we have to be out before we wake up to a brand
new language?
It depends.
Some tongues seem to keep their shape over 1000 years, others do indeed change quickly
within centuries.
But take another look at the question we started with: how long can a language last before
it becomes unrecognizable?
Not unintelligible to earlier speakers.
Utterly unrecognizable.
See, even if a time-traveling Dyirbal speaker couldn't understand the language, they might
be still able to pick out familiar words.
How do we know this will happen?
Because it has happened.
This isn't the first time Dyirbal has evolved.
We got a glimpse of its past changes in part one when we met a relative of Dyirbal where
the word for dog is "dog".
That word took linguistic know-how to prove it was actually related to Dyirbal, not English,
but other patterns among Australian languages are easier to recognize.
Whatever you think about Swadesh lists, here's a fun use for them.
Compare basic vocabulary from a bunch of Australian languages and look for cognates.
Comparing cognates and sound changes and grammar, linguists focusing on Australian languages
looked far back beyond Dyirbal to recover its entire family: the Pama-Nyungan languages.
There is dissent, Dixon in particular came to reject this grouping, but the consensus
is that Pama-Nyungan constitutes the largest family on the continent by far.
Using various lines of evidence, Pama-Nyungan is held to be around 4,000 to 5,000
years old.
Which is intriguing.
Because look what happens when we estimate the ages of other language families: we get
similar time depths.
The one aberration here seems to be Proto-Afro-Asiatic, a story for another time.
Some historical linguists warn that they don't do dates, but some that do see a limit here:
comparing languages can only take us several thousand years into the past.
Of course, in my Tower of Babel first language video, we did meet some mavericks who dare
to go even further.
So, how long should we keep that hypothetical linguistics patient under to make sure they
wake up to a changed language?
Well, within hundreds of years we could expect degraded intelligibility, with handwaving
on dates and how understandable it will be.
But within many thousands of years, even distant traces of the language may have dissolved.
Though keep your apologies handy in case this answer is too vague for the Dyirbalŋan.
Hah.
Stick around and subscribe for language.