Perhaps you’ll pepper in a couple of y’all’s after spending the weekend with your Texan mother-in-law. Or you might drop a few R’s after binge-watching a British period drama on Netflix.
Linguists call this phenomenon “linguistic convergence,” and it’s something you’ve likely done at some point, even if the shifts were so subtle you didn’t notice.
People tend to converge toward the language they observe around them, whether it’s copying word choices, mirroring sentence structures or mimicking pronunciations.
But as a doctoral student in linguistics, I wanted to know more about how readily this behaviour occurs: Would people converge based on evidence as flimsy as their own expectations of how someone might sound?
Three years of experimentation and an entire dissertation later, I had my answer, which was just published in the academic journal Language.
People do, in fact, converge toward speech sounds they expect to hear – even if they never actually hear them.
But before getting into the specifics, let’s talk about what convergence is and how it’s related to other speech adjustments like code-switching, which refers to alternating between language varieties, or style-shifting, which happens when a person uses different linguistic features in different situations.
Convergence refers to the shifts people make to their speech to approximate that of those around them. This is an intentionally broad definition meant to encompass all sorts of adjustments, whether intentional or inadvertent, prominent or subtle, or toward entire dialects or particular linguistic features.
You could imitate aspects of speech you actually observe. Or maybe you throw in some words you think kids these days use, only to have your use of “bae” and “lit” be met with teenage eye rolls.
Code-switching or style-shifting can also be examples of convergence, as long as the shift is toward an interlocutor – the person you’re talking to. But people can also shift away from an interlocutor, and this is called “divergence.”
Code-switching and style-shifting can occur for other reasons, too, like how you feel, what you’re talking about and how you want to be perceived. You might drop your G’s more and say things like “thinkin’” when reminiscing about a prank you played in high school – but switch to more formal speech when the conversation shifts to a new job you’re applying to.
To determine whether people converge toward particular pronunciations they expect but never actually encounter, I needed to start my investigation with a feature that people would have clear expectations about. I landed on the “I” vowel, as in “time,” which in much of the southern US is pronounced more like “Tom.” This is called “monophthongization,” and it is a hallmark of Southern speech.
I wanted to know whether people would produce a more Southern-like “I” vowel when they heard someone speak with a Southern accent – and here’s the crucial part – even if they never heard how that person actually pronounced “I.”
So I designed an experiment, disguised as a guessing game, in which I got more than 100 participants to say a bunch of “I” words.
In the first part of the game, they read a series of clues on their computer screen – things like, “this US coin is small, silver, and worth 10 cents.”
Then they named the word being described – “dime!” – and I recorded their speech.
In the second part of the game, I had participants listen to clues read by a noticeably Southern-accented talker and instructed them to respond in the same way. By comparing their speech before and after hearing a Southern accent, I could determine whether they converged.
Using acoustic analysis, which gives us precise measurements of how participants’ “I” vowels sound, I observed that Southerners and non-Southerners alike did, in fact, shift their “I” vowels toward a slightly more Southern-like pronunciation when listening to the Southern-accented talker.
They never actually heard how the Southerner produced this vowel, since none of the clues contained the “I” vowel. This means they were anticipating how this Southerner might say “I,” and then converging toward those expectations.
This was pretty clear evidence that people converge not just toward speech they observe but also toward speech they expect to hear.
What does this say about human behaviour?
For one, it means that people perceive accents as coherent collections of different linguistic features. Hearing accent features X and Y tells people to expect accent feature Z, because they know X, Y and Z go together.
But it’s not just that people passively know things about others’ accents. This knowledge can even shape your own speech.
So why does this happen? And how do those on the receiving end perceive it?
First, it’s important to point out that convergence is usually very subtle – and there’s a reason. Overly exaggerated convergence – sometimes called overaccommodation – can be perceived as mocking or patronising.
You’ve probably witnessed people switch to a slower, louder, simpler speech style when talking to an elderly person or a nonnative speaker. This type of over-the-top convergence is often based on assumptions about limited comprehension – and it can socially backfire.
“Why are they talking to me like I’m a child?” the listener might think. “I understand them just fine.”
For expectation-driven convergence – which, by definition, is not rooted in reality – such a faux pas might be even more likely. If you don’t have an actual speech target to converge toward, you might resort to inaccurate, simplistic or stereotyped ideas about how someone will speak.
However, subtler shifts – in what might be called the “sweet spot” of convergence – can have a number of benefits, from social approval to more efficient and successful communication.
Consider a toddler who calls their pacifier a “binky.” You’d probably be better off asking “where’s the binky?” and not “where’s the pacifier?”
Reusing the terms our interlocutors use is not just cognitively easier for us – since it takes less effort to come up with a word we just heard – but it often has the added benefit of making communication easier for our partner. The same could be said for using a more familiar pronunciation.
If people can anticipate how someone will speak even sooner – before they utter a word – and converge toward that expectation, communication could, in theory, be even more efficient. If expectations are accurate, expectation-driven convergence could be a social asset.
That’s not to say that people necessarily go around consciously making these sorts of calculations. In fact, some explanations for convergence suggest that it is an unintentional, automatic consequence of speech comprehension.
Regardless of why convergence happens, it’s clear that even beliefs about others play a major role in shaping the way people use language – for better or for worse.