The multivalence of interjections
« previous post | next post »
Feast your eyes on a small segment of the total number of interjections in English: Aha, Hurray, Oh, Ah, Aw, Ouch, Wow, Alas, Boo, Hey, Oh my God, Ahem, Bah, Cheers, Hmm, Huh, Huzzah, Oops, Yay, Agreement, Amen, Argh, Awesome, Boy….
Interjections may be exclamatory, quizzical, condemnatory, laudatory, and so forth and so on. They may convey pain, delight, dismay; approbation, disapproval.
Some interjections are neutral: uh, um.
Some are positive: woo-hoo, bravo.
Some are negative: duh, which is disdainful, demeaning, and even insulting.
"Huh? The Valuable Role of Interjections." Holmes, Bob. Knowable Magazine | Annual Reviews, February 24, 2025.
Listen carefully to a spoken conversation and you’ll notice that the speakers use a lot of little quasi-words — mm-hmm, um, huh? and the like — that don’t convey any information about the topic of the conversation itself. For many decades, linguists regarded such utterances as largely irrelevant noise, the flotsam and jetsam that accumulate on the margins of language when speakers aren’t as articulate as they’d like to be.
But these little words may be much more important than that. A few linguists now think that far from being detritus, they may be crucial traffic signals to regulate the flow of conversation as well as tools to negotiate mutual understanding. That puts them at the heart of language itself — and they may be the hardest part of language for artificial intelligence to master.
“Here is this phenomenon that lives right under our nose, that we barely noticed,” says Mark Dingemanse, a linguist at Radboud University in the Netherlands, “that turns out to upend our ideas of what makes complex language even possible in the first place.”
When we come to the next paragraph of Bob Holmes' thought-provoking article, the question of spoken vs. written language, this is something that lies at the heart of my own preoccupation with written vernacular, which is closer to spoken language than classical and literary language.
For most of the history of linguistics, scholars have tended to focus on written language, in large part because that’s what they had records of. But once recordings of conversation became available, they could begin to analyze spoken language the same way as writing.
When they did, they observed that interjections — that is, short utterances of just a word or two that are not part of a larger sentence — were ubiquitous in everyday speech. “One in every seven utterances are one of these things,” says Dingemanse, who explores the use of interjections in the 2024 Annual Review of Linguistics. “You’re going to find one of those little guys flying by every 12 seconds. Apparently, we need them.”
Many of these interjections serve to regulate the flow of conversation. “Think of it as a tool kit for conducting interactions,” says Dingemanse. “If you want to have streamlined conversations, these are the tools you need.” An um or uh from the speaker, for example, signals that they’re about to pause, but aren’t finished speaking. A quick huh? or what? from the listener, on the other hand, can signal a failure of communication that the speaker needs to repair.
On the other hand, interjections are often used as filler words to give the speaker time to think of the next words he wants to say, and these come in a withering variety of personal favorites. Believe it or not, a classics professor of mine at Dartmouth would say "oink, oink" when he was thinking of what to say next. The first few times I heard him say "oink, oink" in the middle of a sentence about Socrates or Plato, I almost couldn't believe my ears and had to stifle a burst of laughter.
I've also heard the following used as idiosyncratic interjections: "OK" (by the speaker or listener), "fantastic" (by the listener), "precisely" (by the listener), the first sotto voce the latter two with exaggerated emphasis.
Unbelievably, I know a professor who is so addicted to the "f" word that they use it as an interjection (as well as a verb and an adjective ["f—in"]) even in faculty meetings.
The article delves into many more aspects of interjections, such as "continuers" (mm-hmm) and "feedback" (wow, yeah, eh?).
Holmes concludes with a biggy:
Can AI learn to use interjections?
To make artificial intelligence sound more natural, developers are building interjections into its responses. Google’s NotebookLM, for example, offers the option of summarizing information — say, one or more scientific papers — in the form of a podcast hosted by two AI-generated hosts.
On first hearing, the program does a pretty good job: The hosts joke, laugh and insert Mm-hmm and Wow! at superficially appropriate times. But to the ears of a trained linguist, there’s something amiss. (Listen to an example.) [VHM: don't miss it!]
“They almost work, but not quite,” says theoretical linguist Martina Wiltschko of the Catalan Institution for Research and Advanced Studies. “They kind of remind me of AI art, where there’s too many fingers. You don’t see it at first, but if you look carefully, you see that something’s wrong.”
One tell is that when the listener mm-hmms or laughs, the speaker pauses while they do so, lending a slightly creepy note to the simulated conversation. “To me, it’s almost like the uncanny valley,” says Wiltschko. “It’s close, but it’s not quite close enough.”
The biggest deficiency, though, is a robust sense of who knows what in the conversation. The AI hosts seem to flip back and forth on which one knows which pieces of information.
“It’s not just what they’re saying, it’s who’s talking in what context, and who knows what,” says Wiltschko. “I would be really surprised if the AI could ever handle that — and human beings handle that with ease.”
All of which leads me to give vent to my favorite Nepali expression, "bāphre bāph!" and to resort to my habitual Cantonese exclamation-continuer "wah!.
Selected readings
- "The evolution of verbal interpolations" (7/6/24) — with valuable bibliography
- "Slang and fillers not allowed" (10/7/21)
- "I'm (like)" (8/30/21)
- "'I feel like'" (8/24/13)
- "I feel like 'I feel like'" (9/2/21)
- "Like thanks" (11/26/15)
- "'Like' youth and sex" (6/28/11)
- "Like, a Christmas gift card" (12/28/06)
- "That, that, that…" (1/24/16) — the (in)famous Mandarin demonstrative filler, "neige, neige, neige"
- "'That, that, that…', part 2" (8/28/20)
- "'That, that, that…', part 3" (9/14/20)
- "A Chinese analog to English 'you know'" (11/22/19)
- "'Bāphre bāph!' — my favorite Nepali expression" (8/12/18)
- "The wonder of Cantonese particles" (5/14/18)
- "New Singaporean and Hong Kong terms in the OED" (5/12/16) — Cantonese "wah"
[Thanks to Ted McClure]
David Marjanović said,
April 19, 2025 @ 8:43 am
It's a very widespread intensifier. Russian puts же after the word that is to be emphasized, English puts a reference to vertical gene transfer in front of, most commonly, the stressed syllable of that word: "unbefuckinglievably". There's a whole series of papers about this.
Jerry Packard said,
April 19, 2025 @ 9:34 am
In linguistics the markers used by the listener to show they are paying attention are called backchannel markers. Substantial work in the field of Conversation Analysis over the past few decades has studied their use in regulating the flow of discourse.
Yves Rehbein said,
April 19, 2025 @ 12:01 pm
MYL's namesake Anatoli Liberman, Oxford Etymologist, has referred to ouch!
Well, let me give an example. When something goes wrong, someone falls down, children especially say OUCH. It seems that this OUCH is a natural sound that one makes. No, it’s extremely artificial. […]
https://becauselanguage.com/98-origin-uncertain/
Oy vey! ROFL!
Philip Taylor said,
April 19, 2025 @ 2:28 pm
Not convinced that "um" is neutral — when drawn out and uttered with a downward intonation, I would say that it is intended to signal "maybe", "not convinced" or similar.
J Greely said,
April 19, 2025 @ 3:51 pm
I was very disappointed that the list of interjections didn't lead off with "Hurray! Aw! Eeek! Rats! Wow! Hey!". Everything's better with Schoolhouse Rock.
-j
Chas Belov said,
April 19, 2025 @ 5:24 pm
I have constant conflicts with my BFF because she takes my uh-hm as an indication that I'm not listening or am planning what to say in response. But I've also had people in phone conversations ask me if I'm still there if I don't say uh-hm now and then.
Francis Deblauwe said,
April 19, 2025 @ 5:24 pm
A recent Youtube video by Dogen, an American in Japan who teaches Japanese, explained how in Japan people expect you to continuously and explicitly react to what they're're saying to a much larger extent than in the West. If they're talking and you don't say anything, they'll actually think that you don't care for what they are saying and they'll stop talking…
https://youtu.be/r3ZjVYSMThM?t=706
Robert T McQuaid said,
April 20, 2025 @ 7:18 am
At an early stage in international telephony, communication between North America and Australia was routed through synchronous satellites. This introduced a half-second delay between speech and the responsive interjections. Normal conversation was impossible. After a few bad experiences, I instructed clients to avoid all interruptions and switch speakers only when one of us ended with "over".
Chris Button said,
April 20, 2025 @ 7:56 am
I appreciate it varies in degree between different languages or dialects or even people, but I wonder if he is also just more aware of it in Japanese because the interjections sound Japanese as opposed to the English sounding interjections that don't jump out to him in the same way?
Victor Mair said,
April 20, 2025 @ 8:03 am
@Robert T McQuaid
That reminds me of a philosopher friend of mine named Roger who closes his e-mails with "Roger, over and out".
"Roger" stands for "received" in military communications.
"Over" indicates that the speaker is finished and is handing over the communications channel to the other party and is awaiting a response.
"Out" signifies that the speaker has finished what he wanted to say and is not expecting a reply.
This leads to a related question. Namely, in news telecasts, why is there always a slight delay from when the anchor hands the camera / mic over to the reporter in the field. What is actually happening / going on during that pause?
Chris Button said,
April 20, 2025 @ 9:32 am
I have a somewhat related question. TV news anchors in the US (and perhaps Canada, although I don't really recall from when I lived there) seem to be trained to use a truly exaggerated form of intonation. It is suggestive of days gone by when the audio quality was poor.
The result is countless more intonation/phrase breaks than would ever occur in natural speech. Those numerous breaks then result in many more focus words across a sentence, whose stressed syllables then receive prominent contour accents instead of just the usual pitch change if anything.
Nowadays, it just sounds really unnatural and stilted (at least to my ear). Why are news anchors still trained to speak this way?
Jerry Packard said,
April 20, 2025 @ 10:25 am
@VHM: “…in news telecasts, why is there always a slight delay…”
As the anchor with the mic is broadcasting, that feed is what the reporter in the field sees on her monitor. So when the mic is handed over to the reporter, she doesn’t begin speaking until she sees the handoff, which is at the same time that we at home see it. That results in a 1.5 sec delay because of the time it takes for the reporter to begin speaking. If the reporter is overseas the delay is even longer. Note that this is not the same as the 7-second delay used in live broadcasts like the Emmys which allows them to edit out unacceptable language.
Philip Taylor said,
April 20, 2025 @ 11:16 am
Francis D — my Vietnamese wife does exactly the same as your Japanese informant when speaking on the telephone to a member of her family — I have joked on more than one occasion that they cannot possibly require the number of (falling-rising) interjections that she utters !
Victor Mair said,
April 20, 2025 @ 11:25 am
@Chris Button
Excellent observations and questions!
@Jerry Packard
Thanks for the clarification of something that had bothered me for years.
Josh R. said,
April 20, 2025 @ 7:42 pm
Chris Button mused:
"I appreciate it varies in degree between different languages or dialects or even people, but I wonder if [Dogen] is also just more aware of it in Japanese because the interjections sound Japanese as opposed to the English sounding interjections that don't jump out to him in the same way?"
Of course, we have to consider Dogen's audience of novice Japanese learners, and he's making a point about something that isn't covered very much in most popular textbooks. But for what it's worth, I actually did a study on this for an undergrad linguistics class, recording conversations between native speakers of English and native speakers of Japanese, and counting the frequency of back-channel utterances. The Japanese sample had a higher frequency by far.
My hypothesis on this (which my professor wanted me to delay graduation to turn into an application for the Monbusho scholarship) was that the Japanese discourse was marked by speakers looking at their listeners' faces less frequently, and it was possible that the participants in the English discourse made up for a lack of back-channel utterances with non-verbal expressive communication (e.g., nods, raised eyebrows, smiles, frowns).
Alas, with my time in college already extended due to switching majors, I was ready to graduate and get out to the "real world," and my hypothesis remained untested. But Dogen's assertion that aizuchi are *generally* more frequent certainly matches my own experience, as well as my penny-ante undergrad research.
Julian said,
April 21, 2025 @ 2:03 am
@robert t McQuaid
I recall the half second delay. It's surprising how disruptive it is
Jerry Packard said,
April 21, 2025 @ 6:22 am
For back channeling work in CA, I recommend the work of the field founder Emanuel Schegloff, and for CA in Japanese my former colleague Makoto Hayashi.