1093. Why do we say “prob’ly” instead of “probably”? This week, we look at elision in everyday speech. Then, we look at a wild study showing that the way people talk is being influenced by AI.
1093. Why do we say “prob’ly” instead of “probably”? This week, we look at elision in everyday speech. Then, we look at a wild study showing that the way people talk is being influenced by AI.
The elision segment was by Susan K. Herman, a retired multidisciplined language analyst, editor, and instructor for the federal government.
🔗 Share your familect recording in a WhatsApp chat.
🔗 Watch my LinkedIn Learning writing courses.
🔗 Subscribe to the newsletter.
🔗 Take our advertising survey.
🔗 Get the edited transcript.
🔗 Get Grammar Girl books.
🔗 Join Grammarpalooza. Get ad-free and bonus episodes at Apple Podcasts or Subtext. Learn more about the difference.
| HOST: Mignon Fogarty
| VOICEMAIL: 833-214-GIRL (833-214-4475).
| Grammar Girl is part of the Quick and Dirty Tips podcast network.
| Theme music by Catherine Rannus.
| Grammar Girl Social Media: YouTube. TikTok. Facebook.Threads. Instagram. LinkedIn. Mastodon. Bluesky.
Grammar Girl here. I’m Mignon Fogarty, your friendly guide to the English language. Today, we're going to talk about the different ways we leave sounds out of words, and then, we're going to look at a wild study showing that the way people talk is being influenced by AI.
But before we start, I have two quick notes.
First, David wrote in with a correction on one little bit of Latin in episode 1087 about why Latin died. We said that in a famous scene in "Life of Brian" a Roman soldier chastises Brian for conjugating "Romanus" to "Romani," but David says that these words are nouns, so they aren't conjugated—they're declined. And he added, "Yes, my teacher grabbed me by the ear when I made an especially egregious mistake, just like in "Life of Brian"!"
Thanks, David.
And then the second note is that Katy from West Seattle called in and said the domain name of John Kelly's blog wasn't clear in episode 1043. So that is MashedRadish.com. MashedRadish.com. M-A-S-H-E-D radish, like the vegetable, dot com. I actually talked to John just last week, and it's a great blog.
by Susan Herman
This segment is by Susan Herman.
In episode 1039, we talked about how to use ellipsis – initial pronoun deletion, not the punctuation mark – in informal language to make communication flow more easily and efficiently (like the well-known advertising slogan "Got milk?"). Ellipsis is the omission of whole words, but today we're going to talk about a similar phenomenon called elision, which is the omission of sounds or syllables, usually in casual or informal speech. We use it when we're sure the meaning will be clear, and our amazing brains fill in the missing sounds. We'll mostly talk about what we do when we're talking, but elision also happens in writing, especially dialogue, or in certain kinds of literature and poetry. In writing, we normally show elision with an apostrophe.
We skip over certain sounds or syllables – or even blend words together – when we talk to make our speech sound more fluid and rhythmic. That's why teachers of English as a second language teach elision to help their students sound more natural. In an article on the language-learning website Lingopie on how to sound like a native speaker, an re naturally." After all, it would sound "stuffy" to pronounce every single sound in a word in a casual conversation. We've all said something like "Do you s'pose we could go out t' eat t'nite?" and have been understood perfectly.
Vowels, consonants, and whole syllables can be elided (that's the verb form of "elision"). Almost every syllable in English contains a vowel, and we often refer to syllables as phonemes or units of sound.
First, let's look at how elision happens. Vowel elision occurs only with unstressed vowels, sometimes called "weak" or "reduced" vowels, like in the pronunciations "prob'ly," "choc'lat," and "vet'ran." Each of these words technically has three syllables — "probably," "chocolate," and "veteran" — but we often shorten them to two in everyday speech.
In each case, the unstressed vowel is dropped. When a syllable contains an unstressed vowel, we describe it as unstressed too.
For the linguistically curious, "stress" means emphasizing a sound by making it longer, louder, or higher in pitch, as in "gram-mar," where the first syllable is stressed, or "e-li-sion," where the second syllable is. Stressed vowels are never elided; that would sound strange and could cause confusion.
We can also give consonants the boot, usually when they come at the end of words or syllables or before another consonant. Some examples are "nex' time," "col' drink," and "goo' boi" (which the author of this segment admits to gushing frequently to her cats). Notice that the first two examples involve words that end in the same consonants the following ones begin with. It might sound a little stilted to pronounce both consonants, like "nexT Time" or "colD Drink." The last example, "goo' boi," is "baby talk," which can involve omitting sounds to make words shorter or more digestible.
"Nex'time" and "col'drink" are also examples of word boundary elision, which is when two words are blended to make communication smoother. This includes contractions, like "I'm" or "let's." Sure, we could say "I am" or "let us," but it may sound unnecessarily formal. The author of this segment remembers a language instructor writing "D'jeet d'jet?" (D'-J-E-E-T D'-J-E-T) on the whiteboard and asking if they knew what it meant. When they sounded it out, they realized it was how "Did you eat yet?" is often rendered in informal, rapid speech. There's a lot going on with that example, but you get the idea.
Now, we can break down elision further by where it happens, or which part of the word is deleted. First, we have aphaeresis, which is the deletion of sounds or syllables at the beginning of words, like "lone" instead of "alone"; "'tis" instead of "it is"; and "'em" instead of "them" (as in "Did ya see 'em?"). Merriam-Webster doesn't even label "'em" as "informal," since it's so common, but notes that it dates back to Middle English and was used before the 12th century. (For more on dictionary labels, see episode 1079.) So elision isn't a new phenomenon and doesn't happen suddenly; words and pronunciations change over time, just like all the other aspects of language.
And we have a special name for aphaeresis when the vowel or syllable deleted at the beginning of the word is unstressed: aphesis. It is very common and can be heard in words like "'neath" for "beneath," and "'fess" for "confess."
The next type of elision is syncope, which is when sounds or syllables are deleted from the middle of words. Think "li'bry," "cam'ra," or "prob'ly." With syncope, the unstressed syllable comes after a stressed one. Another interesting example is, well … the word "interesting," which, as University of Michigan professor Anne Curzan points out in an NPR interview, can be pronounced with either three or four syllables. And while the three-syllable and four-syllable camps may curse each other, according to Curzan, "if you look in modern standard dictionaries from the last ten years, they will show multiple pronunciations, three and four syllables." (And you can hear more from Anne Curzan in episode 980, when she was on to talk about her book "Says Who?")
The final type of elision is apocope (funny word, right?), apocope is the omission of sounds or syllables at the end of a word. For example, most of us would say "photo" instead of "photograph" in normal conversation. Or "auto parts" instead of "automobile parts." Or maybe "obit" instead of "obituary." It's just more efficient.
And Renaissance faires are an interesting example. First, we often use apocope to shorten "Renaissance" to "Ren," and yet we often spell "faire" F-A-I-R-E, with the E on the end that we don't pronounce to give it an old-timey feel, like how some businesses spell "olde" and "shoppe" with an E on the end for the same reason. These final E's were once pronounced but then dropped, and eventually, we stopped using them in modern spelling.
Finally, as I mentioned at the beginning, elision can be used in writing, especially in literary or poetic works. One notable example is the "Star Spangled Banner," whose lyrics includes "O'er the ramparts we watched," in which "over" is pronounced and written without the V. Even Shakespeare used elision in Sonnet 116, published in 1609, in the verse "It is the star to every wand’ring bark." These are both examples of syncope.
Now g'on and share your newfoun' knowledge of elision! And 'member: Grammar's fun!
That segment was by Susan K. Herman, a retired multidisciplined language analyst, editor, and instructor for the federal government.
by Mignon Fogarty
One of the things I've wondered ever since I encountered AI writing was whether the patterns we see there would change how we actually use language ourselves. For example, I've wondered whether ChatGPT correctly using idioms that people often confuse, like "champ at the bit" versus "chomp at the bit" — hypothetically, I don't know if that's actually happening — but if it did, would that cause more people to use the right phrase?
Well, we don't have the answer to that yet, but there was just a really interesting study that showed people are actually starting to talk like ChatGPT.
You might remember from the show last week that ChatGPT seems to use certain lofty words more often, especially when writing scientific papers. Words like "delve," "meticulous," "intricate," "captivate," and "tapestry."
Well, researchers at the Max-Planck Institute for Human Development looked for those top 20 words in the transcripts of almost 280,000 academic YouTube videos and found that since the release of ChatGPT, these YouTubers are also using some of these words more often.
I wondered whether the channels were just reading scientific papers out loud since the words are already known to appear there more often, but the researchers said that they looked at videos that included both scripted content and people just talking, and they found the increases in both!
People in these types of videos used words including "prowess," "tapestry," "reveled," and yes, "delve" far more often after ChatGPT came out in 2023 than they did in earlier videos. "Prowess" showed the biggest increase — showing up about 60% more often — but quite a few others showed up at least 30% more often.
One interesting thing is that it isn't happening with all the words though. For reasons they don't understand, the use of some words popular with ChatGPT, such as "groundbreaking" and "underscore," actually fell, and it'll be interesting to see if they can find other factors that are influencing people's word choices.
One thing I wish the study would have looked at is how many of the science YouTubers using the ChatGPT words had English as their second or third language. Another thing AI writing does is flatten out dialect quirks. It is highly likely to write in standard English, so when a researcher from Brazil whose first language is Portuguese, for example, puts a paper they wrote in English into ChatGPT and asks it to "clean up the language," it will most often strip out any phrases that don't seem like standard English.
I don't know enough about science in other languages to give you a relevant example, but I'm thinking about episode 969 from 2024 when Susan Herman wrote about calques, like when people in South Florida say they "make a party" instead of "throw a party" because "make" is the verb they'd use in Spanish. So if a scientific paper had a similar kind of nonstandard phrasing, I'd expect ChatGPT to change it. And then as we've seen, it could also put in a few "delves" or "realms." You can imagine that those writers, in particular, who are under pressure to publish and sound professional, could absorb the idea that these words represent "good English" more than a native speaker might if they're looking at the changes ChatGPT made for feedback on their writing. And if that happens, they may also be more likely to incorporate it into their speech.
There's still a lot to tease out here, and the researchers also say that this isn't the first time computers have influenced the way people behave. They point to people changing the way they play chess or the game Go based on strategies developed by machines.
Finally, an article about the study in The Verge points out that the existence of AI can also influence how we feel about messages we receive. For example, people have been using suggested replies in email and texts for a while. A Nature paper by researchers at Cornell says billions of these "smart messages" are sent every day. Ironically, people who receive these messages actually like them more than messages written by humans, and they describe the supposed people behind these messages as "closer and more cooperative." But once the people getting the messages become suspicious that their conversation partner is using AI, they change their tune and rate them more negatively.
This sets up a catch-22 highlighted by the researchers. They say, "perceived smart reply use is judged negatively, [but] actual use by the partner resulted in more positive attitudes." So it's not AI use itself that's turning people off — that's actually making people more satisfied — it's attitudes about AI use — the feeling that the messages are insincere, inauthentic, or that someone doesn't care enough to take the time to write a message — that's turning people off. Communication is about much more than the words we use.
I've been interested in language change for a long time, and I suppose it shouldn't be surprising that these new tools — that we interact with through conversational language — are also a new driver of language change. It's definitely interesting, and I'd say the takeaway message is that it might not be all bad to take some writing advice from AI, but when you want people to feel like you care about them, make sure they know that you're actually writing the messages you send.
Finally, I have another answer to a question in a past podcast.
Hi, Grammar Girl. This is Pete Atkinson calling. I'm on my commute. I just listened to episode 1063, I believe, and the familect was a question about a VW bug. They were asking about the space behind the seat that they -- excuse me -- they called the "cubby." I've heard it called a "kombi," K-O-M-B-I. And in fact, if you look up the lyrics of the old song by Men at Work, "Down Under," the first line says something about traveling in a burned out kombie. So I think that's the Australian version of the cubby. So it might come from there. I don't know. But that's my story, and I'm sticking to it. Great work. Later."
Thanks, Pete! What a great blast from the past it is to remember those lyrics!
Wikipedia tells me that a "Kombi" is another name for the Volkswagen Transporter, which looks like a type of Volkswagen Bus, and the name comes from from German: "Kombinationskraftwagen (combination motor vehicle)" because with side windows and removable rear seats, it was both a passenger and a cargo vehicle combined. But that doesn't mean that it couldn't also be a name for the cubby in a VW bug. Thanks for the call and listening to the podcast on your commute.
If you want to share the story of your familect, a word or phrase that you only use with your friends or family, leave a message on the voicemail line at 83-321-4-GIRL or leave a voice message on WhatsApp. Be sure to call from a nice, quiet place, and if you want that number or link later, you can always find them in the show notes.
Grammar Girl is a Quick and Dirty Tips podcast, and we have lots of other shows. This week, the Nutrition Diva responds to concerned listeners who worry they might not be getting all the nutritional benefit out of their favorite breakfast foods.
Thanks to Morgan Christianson in advertising; Dan Feierabend in audio; Nat Hoopes in Marketing; and Holly Hutchings, director of podcasts, who loves pilates, and her current favorite move is single leg bridges.
And I'm Mignon Fogarty, better known as Grammar Girl and author of the tip-a-day book "The Grammar Daily." That's all. Thanks for listening.