Picking Up An Accent Is Less Automatic Than Many Assume

Told Not to Copy an Australian Accent, Americans Largely Stopped

Spend a few days around someone with a different accent, and some of their pronunciations can rub off. Speech science often treats that as something the brain does on its own, whether anyone wants it to or not. A new study of 136 American adults found that a simple instruction can stop it. People who heard an Australian woman and were then told to say words their own way showed no clear shift toward her accent.

One sound still came through. Participants given no instructions at all drifted toward the Australian habit of dropping R’s in words like “near.” That change was one nobody asked for, and participants may not have realized they were making it.

Kathryn Campbell-Kibler and Cynthia Clopper of Ohio State University reported the work in Laboratory Phonology: Journal of the Association for Laboratory Phonology. Their study put a basic question to an old assumption: is copying an accent a reflex, or something people can steer? The answer was mostly the second.

Accent Mimicry Switched On and Off With Simple Instructions

Volunteers were visitors to an American science museum. Of 236 who signed up, 136 made the final analysis after exclusions for speech or hearing disorders, unfinished questionnaires, recording problems, and a few other reasons. The group included 68 men, 64 women, 1 nonbinary person, and 3 who did not report a gender, with an average age of 33. Of the 136, 115 identified as white.

Each person first read a list of words off a screen, giving researchers a baseline of their natural speech. Next, each heard a college-educated Australian woman in her 20s say the same words and repeated them right after her, a setup called word shadowing. Researchers picked Australian English partly because Americans generally view it favorably.

Only the instructions changed. Forty-two people were told to “try to imitate the speaker as closely as possible.” Fifty-one were told to say each word the way they normally would, not the way she did. Forty-three got no guidance.

Researchers then measured the exact sounds in nearly 11,000 recordings: four vowel sounds, found in words like “bath,” “trap,” “kit,” and “price,” plus the R in words like “near.” Australian English tends to drop that R, while American English typically pronounces it.

When asked to imitate, participants shifted toward the Australian pronunciation on all five features, most sharply on the R and the “bath” vowel. Those told to avoid copying her showed no clear shift toward her on any feature. On the “kit” vowel, they moved a bit away from her pronunciation, a result the authors consider possibly a fluke.

That outcome conflicts with an earlier study that used the same recording, in which volunteers still copied the speaker when asked not to. The authors could not pin down why. One clue is that the earlier volunteers were told they were in a dialect identification study and heard several speakers, which may have focused their attention on accent features.

Among those told to imitate, individual choices varied widely on the R. Some participants copied the dropped sound much more than others did.

Accent Mimicry Crept In Through the Dropped R Sound

Among the 43 people given no instructions, only one feature clearly changed: the dropped R. The vowels held steady.

Afterward, participants were asked what they noticed about how the woman spoke. The R and the “bath” vowel were named most often, by 22 people (16%) and 21 people (15%). Written comments included “she didn’t pronounce r’s.” In an analysis added after the main results, and in need of confirmation, those who named one of these sounds tended to copy it more, whether told to imitate or given no instructions.

Well-Known Sounds Drew the Most Accent Copying

Why would the R slip through when the vowels did not? Both the R and the “bath” vowel are well-known markers of how American listeners hear other dialects, the authors note. That recognition may make a sound likelier to get copied, though only the R crossed that line when nobody asked.

A second explanation is possible. Because shadowing involves no conversation partner, some participants may have copied the R on purpose, for their own amusement, in what the authors describe as a deliberate but unrequested choice to play with language. The study could not tell the two explanations apart.

If recognizable sounds do the most sneaking, the accent someone brings home from a long trip could depend partly on which pronunciations already stand out to them. That is speculation. The study tested single words read aloud, not conversations.

A separate look at gender, also run after the main results, found women shifted their speech more than men did. The authors said what it means is unclear and noted that broad gender categories do not translate neatly to speech behavior.

In this lab task, people held more control over accent copying than speech scientists often assume. A reflex does not take orders, and this one did, with a single recognizable sound still creeping in when nobody said a word.

---