Grammar Girl Quick and Dirty Tips for Better Writing

How linguists solve crimes, with Natalie Schilling

Episode Summary

1027. This week, I talk with forensic linguist Natalie Schilling about how people's language gives them away — in manifestos, ransom notes, text messages, and more.

Episode Notes

1027. This week, I talk with forensic linguist Natalie Schilling about how people's language gives them away — in manifestos, ransom notes, text messages, and more.

Natalie Schilling is a professor emerita of linguistics at Georgetown University in Washington, DC, and runs a forensic linguistics consulting firm. You can find her on LinkedIn.

🔗 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 Links: YouTube. TikTok. Facebook.Threads. Instagram. LinkedIn. Mastodon.

Episode Transcription

MIGNON: Grammar Girl here. I'm Mignon Fogarty, and I'm here today with Natalie Schilling, forensic linguist and professor emerita of linguistics at Georgetown University. Natalie, welcome to the Grammar Girl podcast.

NATALIE: Thanks so much for having me. Excited to be here.

MIGNON: Yeah, you bet. So, you know, I understand that you have been portrayed on TV for your work and, you know, how did that come about, and what is it that forensic linguists do that is so exciting?

NATALIE: Yeah. So I have been, I have never been a TV star, but I've been played on TV and not just once, but twice. So, let me think of how to phrase this. So the first me on TV was a person on "Criminal Minds" who is called Dr. Alex Blake, and she's actually a composite, not the real me. She's made up of a couple of different people, and she happens to be part linguist at Georgetown University. So she teaches at Georgetown University, and she also happens to be an FBI agent at the same time. So we don't have too many people who teach linguistics at Georgetown who are also FBI agents. So they took a little bit of artistic license with that character, but that's me.

And then the other time I've been played on TV is there's a character in a scripted series called "Manhunt Unabomber," that's a Discovery Channel series, and there's a person on that show called Natalie Rogers, and she is a composite of me, Natalie Schilling, and Professor Roger Shuy, professor emeritus of Georgetown University, who is one of what we consider to be the founders of forensic linguistics.

So that's my other alter ego. Go ahead.

MIGNON: It sounds like Georgetown is sort of the place to be if you're a forensic linguist.

NATALIE: It can be, we've done our share. Professor Shuy was instrumental in starting the field of forensic linguistics, forming the field, getting it to be known at least in a limited way. And he worked a number of cases in forensic linguistics as well.

So what a forensic linguist does, what forensic linguistics is, is forensic linguistics very broadly speaking involves anything having to do with language and the law. So you're bringing training in the scientific study of language, which is linguistics, into the legal arena. This could involve anything from analyzing language evidence.

Maybe there's a crime where part of what you need to look at in connection with the crime involves language. So you have DNA evidence, but maybe there's also language evidence. Other areas that linguists might work with the law could involve issues in courtroom interactions. So, for example, courtroom interpreting, that sort of thing.

I've mostly been involved in the area of analyzing language evidence. So looking at, say for example, a criminal case, maybe there's an anonymous communication of some type, maybe it's an anonymous threat. And so you might look at that, either document or listen to a phone call and be able to say something about the characteristics, maybe the demographics of the person who's responsible for the communication.

So that would be an example of something that a forensic linguist would do.

MIGNON: Okay. What are some examples of the kind of things that you would detect in your work, like in a phone call or in some of the written correspondence that you're analyzing?

NATALIE: Yeah. So a linguist working in forensic linguistics would look at details of language on every level of language. So say, for example, there's an anonymous phone call. That's a threat of some type. And the investigators just want some help trying to figure out. Can we narrow down our search?

Who could this be? So they might contact a linguist to do what we would call speaker profiling. So can you listen to this audio and tell us anything about, for example, is this a native speaker of the language? So is it a native speaker of English? Is it somebody from a particular region or even country.

Can you tell, is this an American? Is this a Canadian? Is this a British person? If it's an American, anything about regionality, is it maybe somebody from the south? Is it somebody from the midwest? Are there indicators of ethnicity? Indicators of gender, we often think, oh, that shouldn't be any big deal to tell the gender of a voice, but sometimes it can be. Age.

So those would be some of the factors. So demographic factors that we might look for in that type of forensic linguistic work.

MIGNON: How much can you narrow it down? I had someone who listened to the podcast once ask me if I grew up in the actual city where I grew up, and they said it was because of the way I speak, not because they knew where I was from or anything. And I still think they were messing with me. Like, can you really narrow it down to an actual city?

NATALIE: I was, if I was really amazing right now, I'd say, well, I can tell you exactly where you're from right now, but I actually can't at the moment. But if sometimes, believe it or not, there actually can be like a linguistic tell, and not everywhere, not every city would have that. I don't want to give too many away because we like to keep them in our little bag of tricks for amazing you and making you feel like, wait, she's messing with me, but I can tell you a few.

So let me think about something that would be an example of where somebody is from. Oh, so this is a super easy example, but so for example, if I'm talking to you, and you say, “Hey, let's go out to lunch. Let's go get a hoagie.” Which is the term a very localized term for that long thin sandwich with cold cuts and cheese and all that stuff.

So it's what many people in the U.S. would call a sub. That's kind of an obvious example. It's very confined to the Philadelphia region and a little bit beyond that. Otherwise in the U.S. people don't use that term. So then I would say, “Oh, you must be from the Philadelphia area.” And you'd say, “Whoa, how would you know?”

I think that's kind of an obvious one, but there are kind of more, they're kind of less obvious ones. So a good example comes from work by well-known sociolinguist, William Labov, who's actually considered to be like a founder of sociolinguistics. And this is actually a case of forensic linguistics where a small detail of language really narrowed down where somebody was from, or in this case, not from.

So he worked on a case involving a bomb threat. This was back in the eighties, a number of decades ago. And at one point the threat, it was an anonymous threat called in. And this was in LA. It was in LAX, Los Angeles airport. The bomb threat caller called in and said something like, “There's a bomb on the plane,” blah, blah, blah, blah, “I know you're on the-at.” So he said "the-at" for "that." That's not an exact pronunciation, but it's approximate. 

People, the investigators and airport officials in LA, listened to the bomb threat tape. And they said, “Oh, that guy has a New York accent, New York City accent. So we have a cargo handler working here who has a New York accent. It's him. It's that guy.” 

So they actually arrested the guy. He was put in jail. His defense team contacted Dr. Labov and said, “Hey, you know, you're an expert in New York City English,” which Bill Labov is, he's done foundational work in sociolinguistics on features of the New York accent. And he said, “You know, the prosecutors think this guy is from New York, he assures us he did, or thinks that the bomb threat caller is from New York and thinks it's this guy, you know, we're convinced our guy's innocent.” So Labov listens to the audio, and he doesn't even actually need the audio. Any time to analyze it, he knows he basically knows right away from that “the-at” that person is not not from New York City.

And it's actually very subtle detail because in New York City English at the time, there is a pronunciation like that. Somebody might say be-yag for bag. But in New York, that “ee-ya” for eh is not pronounced that way before T sounds, like in that. So that's a really subtle pattern. And in that case, it was enough for Labov to go into court and to testify in front of the judge and say look, here are the patterns of New York City English. Here are patterns of other Englishes on the East Coast where “ee-ya” works differently. Guy who's from New York City is not your bomb threat caller. You need to look for somebody else. And so it actually worked to get this guy out of prison.

MIGNON: Wow. I just got chills.

NATALIE: Yeah, yes. It's a powerful example, and that happens.

I would, I mean, that sort of thing I've been involved in as well, where, you know, it's involved looking at sometimes obvious features of language. Oh, I know that person has to be from Philadelphia because they said "hoagie." Sometimes it's something less obvious. And these less obvious features can come into play when people are attempting to disguise their language.

So if you're disguising yourself, you might say you wanted to fake an accent because you're calling in a threat, and you want to make sure people don't know it's you. You might be able to mimic a few more obvious features, but you're probably, you're not likely to be able to fake, well, I'm going to pretend I'm somebody not from New York, so I'm going to say "the-at" in the wrong place.

That's people aren't going to have that level of awareness about their language to be able to do that.

MIGNON: Because there's so many little subtle things. Are there grammatical, like are there sentence structure tells too that you look for? Or that can be a sign of something?

NATALIE: Sure. So maybe there would be, maybe there would be say a note that somebody wrote in connection with a crime. I don't know, maybe a ransom note in a kidnapping or something like this. Maybe somebody writes "something, something needs washed" or an example like that, where that's a feature that is found in certain areas of the U.S. instead of "needs to be washed" or "needs washing." It's found in certain areas of the U.S. but not others. And so that could be a clue. So a subtle feature of dialect variation where people aren't even aware that they're doing it. 

So, you think, well, why would a bad guy who's writing a ransom note put something so obvious in it like that, that localizes him? And the reason could be, well, you don't even realize it's a local pronunciation. You grew up saying it, all your friends grew up saying it. And so you just think everybody does it. I remember being surprised as a linguist when I realized that saying something like, "Whoa, anymore, there's really a lot of traffic around here."

I remember learning in graduate school that that's a dialect feature. That's a regional dialect feature. I said, no way. That's just regular talk. How else would you make the sentence? "Anymore there's a lot of traffic," which means something like "Nowadays, there's a lot of traffic." And you would think somebody who's immersed in the study of linguistics wouldn't be surprised by something like that.

So you can imagine how somebody who's trying to cover their tracks and write a nefarious communication, something can actually leak through. So it could be a grammatical feature. Most definitely. And in terms of forensic linguistics, it doesn't only involve audio data. It also involves written data.

So questions of author profiling. Hey, we've got this anonymous note. Where might the person be from? And also author identification or what we often call authorship attribution. So we have a communication. We have three suspects. Can you help us assess which of the three suspects it might be? And of course you're going to look at grammar among other features because, you know, your pronunciation channel isn't going to be there in the written communication.

I guess some pronunciation features could sort of be evident in the spelling, but it's not going to be as likely.

MIGNON: Yeah. I'm curious how much of the work you do is audio versus written. You know, you've talked about both.

NATALIE: Yeah. I've done some of both. So I've been involved in speaker profiling, speaker identification, author profiling, and author identification. Feels like nowadays maybe there's more on the written side. I mean, there's an awful lot of, as you well know, there's an awful lot of internet communication, social media communication, text messages, all these things where it seems like just more and more people are communicating more in written form than spoken form.

And unfortunately, you know, criminal communications are moving that way too. You know, there are unfortunately lots of problematic communications that come to investigators and come to law enforcement in written form.

MIGNON: I'm curious how much of someone's writing you need to look at to be, you know, when you've got the three people, and you're trying to decide which one it is, how much of their other writing do you have to look at to make a determination?

NATALIE: That is a great question. One of the biggest challenges in doing forensic linguistics is how much language do you need? How much linguistic data do you need? And in forensic linguistics, almost always the data is very short. There's not much. It's rare that you would get, you know, you're not going to get except in one very bizarre case, you're not going to get a ransom note that's pages and pages long. 

So there's no, there's no set answer. The more data you have, the better. But very often, especially for the problematic communication, it’s probably going to be short. A threat will be short. Okay. You know, a defamation or something like that will be short, a ransom note would be short, but the comparison communications, maybe you can get some more, you know, maybe the person is a prolific blogger or something like this.

And you would be able to get more data. The more you have, usually the more convincing, the more firm your analysis and your conclusions can be, but you also have to be very careful that more data isn't always better data, because ideally, you want it to match as closely as possible in genre, type of communication. So if I write a threatening email to a coworker, and the comparison data that the investigator and the linguist has is my shopping list or, you know, I don't know, something crazy like that or my tweets to my mom or something like that. It might not be, we all change our language a lot based on the type of communicating we're doing.

What's the purpose of the communication? Who are we talking to? Is it humorous or serious tone, these types of things. And so if we get a question document and some documents of known authorship that are quite different in purpose and tone and audience, then we might accidentally conclude it's a different person rather than correctly concluding it's the same person. So we would strive to get communications that are as close as possible to each other.

MIGNON: Yeah. You mentioned tweets and digital communication. Have you seen any cases where things like emoji or memes came into play? Like this person always uses this meme or they tend to end things with a laughing emoji or anything like that?

NATALIE: I've been involved in a case involving authorship of some question text messages. And while the person or the … nobody's style is marked by, oh, they always use, you know, smiley face number 72 or something like that. But I could clearly see texting styles where one person's communications include these types of cutesy things.

And another person's really just kind of don't include them. So that's an example I can think of, but I wouldn't be surprised if there are very specific emojis that mark, um, particular people's styles.

MIGNON: Mm hmm. And what are some of the … just tell us some stories. What are some of the really interesting stories in forensic linguistics? Like, famous cases, maybe, are there cases that every student of forensic linguistics learns about?

NATALIE: Oh, that's a good question. Yes. The case I just talked to you about with William Labov is one that I have always taught my students, and it tends to appear in say textbooks and that kind of thing. So this is the case, we often just call it the "Pan Am bomb threat case" because the threat was made against the now defunct Pan Am airlines.

So that's one. It's a great case because it illustrates so well how knowledge of subtle patterns, subtle yet highly regular patterns of dialect variation can be used in a forensic context, can really pinpoint something about demographics, who a person is, and who a person is not, in terms of identifying an individual. Another case that people often point to is one that Roger Shuy was involved in where there was a ransom note in connection with a kidnapping.

A juvenile had been abducted from their home, and there was a ransom note instructing people where to put the money. And it said something like, take the money, put it in a diaper bag, and take it to the corner of, take it to the corner of 5th and Carlson, something like this, and leave the bag on the devil strip. And this is, I guess we might call it a linguistic smoking gun. So it's another example of a word like "hoagie," which is very regionally confined. So it's a lexical item that's tightly confined, and Professor Shuy didn't actually know when he got the case where that term is. 

But thankfully, we have this great resource in sociolinguistics and linguistics and dialect study. That's called the Dictionary of American Regional English. And you can go to that dictionary. It's multi-volume. It's huge. It's also very fun to read through because it catalogs all these regional terms, these regionalisms in American English. So Professor Shuy was able to go to that great resource and look up "devil's strip."

Oh, well, that's really confined to the area of Akron, Ohio, and that's about it in the U.S. And so in that case, we're able to say, well, you've got a suspect pool. Do you have anybody from the Akron area? And also in that case, it was a weird letter because, you know, something, something diaper bag, but there were also, these misspellings, like "cop" was K-O-P, "daughter" was something like D-O-T-T-E-R, and Professor Shuy said, "Well, no, wait a minute. If you can spell a word like 'diaper,' that's not the easiest word in the world, but yet you're writing like this cartoonish spelling, K-O-P for 'cop.'"

He's like,"No, you're, it's just somebody who's disguising themselves. They're trying to appear like a, you know, kind of low-literacy author. But I think they're actually somebody with a higher education level," and he was, it turns out that that to be the case as well that yes, it was somebody with more education who was attempting to dumb down the communication.

MIGNON: Interesting. You talked about generally not having a lot to work with, and it made me think of the Unabomber case, because there was a man—, like, a long manifesto. Can you talk about that case?

NATALIE: Yes, it is a long and involved case, and there are so many ins and outs and so many details. But yes, essentially, it involved a serial bomber who was active in the U.S. for many, many years who, along with the improvised explosive devices that he would send to people, he did kill several people with his bombs.

It was a, you know, very serious, horrible crime. So he would send these devices that he built himself to people. And along with them would be notes, would be communications to the people. He also sent some letters to other people warning or threatening with violence. And among his communications, in addition, he wrote what is nowadays called the Unabomber's Manifesto.

It was a long, 50-something page document, in which he outlined his grievances against the world. In particular, he was very anti-technology. He didn't like universities and professors. He spoke out against, he spoke out against the increasing prevalence of computers and the development of computational technology, very interestingly, and he wrote this long manifesto, and he mailed it back in the day when you had to physically go, you know, buy envelopes, put stamps on things, go to the post office. He mailed it to several media outlets and said, "Hey, if you publish this long document, I'll stop bombing in order to kill people."

He  also said, "I reserve the right to keep bombing to destroy property, but you know, I won't kill anybody." Yeah. Yeah. "I won't kill anybody. If you publish this." And this is something that the media outlets really wrestled with. Meanwhile, there's an ongoing FBI investigation. Who is this bomber who's mailing these packages to different states across the U.S.? That's why it's a, you know, it's a federal case because he's working all over the U.S. What clues do we have? They tried some before this long document was sent to media outlets. The FBI tried some linguistic profiling. Okay, we have these letters that came with bombs. What can we say about the person based on these communications?

Once the manifesto came to light, once the media outlets alerted the FBI, "Hey, we've got this document," then that gives us a lot more communication. That we can, you know, try to figure out, are there some clues as to where this person might be from? Is he from the U.S.? Where in the U.S.? What else about him can we tell? 

Then finally, at one point, Ted Kaczynski’s brother David, David's wife was reading ... I think she was actually on the early Internet, looking at an FBI site of some type. I don't exactly remember. But she was reading this manifesto and she said, "David," she said to her husband, who was Ted's brother, she said, "This is your brother." So she thought she recognized the language. David looked at it as well and agreed and said "Yeah, I’m afraid this really may be Ted." 

So David contacted the FBI, and the Kaczynski family then began turning over documents that Ted Kaczynski was known to have written. So letters that he had written to his family, he was a very prolific writer. He wrote a ton of letters, including many, many letters to his brother, David. And they gave these documents to the FBI. So the case initially involved author profiling. We have these anonymous documents. We have some letters that were attached to bombs. We have a long manifesto. What can we say about the person who wrote them from any linguistic indicators and interestingly, Roger Shuy was also involved in that case. And he gave an opinion. He looked at some spellings, for example. So "instalment," I-N-S-T-A-L-M-E-N-T with one L instead of two, a couple other spellings with one L instead of two. "License." I always get confused which way "license" is supposed to be spelled in American English. But the manifesto and the letters with the bombs used a particular spelling of "license" that was more of a British or Canadian spelling than an American spelling. And Professor Shuy actually gave a profile in the early days saying, well, the Unabomber, because of these spellings, he might be somebody possibly more northerly or Canada, or he could be somebody based on this "instalment" … and "wilfully" was another one with W-I L-F-U-L-L-Y, one L. He said, you know what, these spellings come from the Chicago Tribune style guide of a particular time period, I think like '50s and '60s.

And so he's like, you know, this might be somebody who's not that young, who is from the Chicago area. So he gives this kind of abbreviated profile, but hey, maybe this has something to do with who this guy is. But then when they get these known writings of Ted Kaczynski, then you can do what we call authorship attribution, where you can compare the anonymous criminal communications with the known communications. And then you can do what we call authorship attribution, where you can compare the anonymous criminal communications with the known communication, so you can take Ted's letters to his brother, his letters to his family, et cetera, et cetera.

There was a fairly long document known to be written by Ted in which he also is outlining some grievances, so you can compare something that's a very like genre with the anonymous manifesto. So that's what FBI investigators did. So there was a Unabomb task force and they had, within the task force, they had a team that was looking at doing this. They didn't call it authorship attribution at the time. They didn't actually bring in any other linguists per se at the time. So this was headed by a supervisory special agent, James R. Fitzgerald, who just with his team did this very detailed comparison where they considered, say, for example, lexical items.

MIGNON: What's that? Explain what that is…

NATALIE: Oh, right.

So terms, vocabulary words. So, for example, there was something like "edible roots and plants" in, you know, unusual things that most people wouldn't write about or talk about that was found across the known and unknown communications. So vocabulary, he also in both sets of documents, there's also reference to like "chicks" for women, so like an older derogatory term where you think, "Oh, that's not a young person who's going to call women 'chicks.'"

Then they also considered grammar. Are there any kind of grammatical tells? The Unabomber wrote in a very, what we would consider grammatically correct and standard way. So there wasn't necessarily any kind of tell that was an obvious, you know, mistake that we see across the known and unknown documents.

But all of these documents are very, you know, grammatically correct according to accepted standards. Though there are some interesting, say, lack of commas where one might put a comma, something like that. Nothing that necessarily really leaps out a lot, but there are consistencies. There are also what we, I actually noticed later when I was taking a look at the Unabomb documents, I found some other usages, whether we call them vocabulary or grammar, things like whether or not you write — and these are all perfectly proper English again — so people don't think about it or notice it, but do you use "anyone" or "anybody"?

Do you use "everyone" or "anybody"? Do you use "no one" or "nobody"? Believe it or not, these things tend to be, we tend to do them consistently, even though we don't think about them, and they do have a little bit of a regional aspect to them. So you can actually look at these features that really fly under the radar.

And so if somebody's trying to mask who they are, they wouldn't necessarily notice. So I found some interesting consistencies in that regard across the sets of documents. The main thing in the Unabomb case that was the linguistic smoking gun, there isn't always one, rarely is there one. But Agent Fitzgerald, noticed in the manifesto, he had highlighted this phrase called that's something to the effect of, "Well, you can't eat your cake and have it too."

And he's like, "Oh, my gosh, finally, the Unabomber made a mistake because the phrase is 'You can't have your cake and eat it too.' I don't believe it. I thought he was perfect." Then in reading Ted Kaczynski's work, unbelievably, there's that same phrase again, "You can't eat your cake and have it too," inverted the other way around.

It's like, "Wow. Oh my God." And you can even do quantitative research where you can go look, you can search the internet, you can search newspaper archives, and you can find in American English usage, it's very rare for people to put the phrase "eat and have" instead of "have your cake and eat it." It turns out it was never a mistake in the first place because the original way that the phrase was ever written was "you can't eat your cake and have it too," not "have your cake and eat it too."

So the Unabomber turned out in the long run to have actually been more correct. So yeah, so that case was, it's just an interesting and quite foundational case in forensic linguistics, even though at the time the FBI wasn't calling it forensic linguistics, but it was definitely a case of, uh, author comparison.

Technically in forensic linguistics, we would call that author verification where you have some anonymous documents and some documents and documents from a suspect, you're comparing two sets of documents when we compare anonymous communications to a number of authors. We call that authorship attribution.

It's a very similar task either way. 

MIGNON: That's amazing. That's more, that's more detailed than I've ever heard about the Unabomber case. And it's just fascinating. It makes me wonder if you can tell someone's state of mind. Have you ever called on to tell someone's state of mind by comparing like different documents from different time periods to try, you know, prove maybe they were angry or murderous or something when they wrote a certain letter?

NATALIE: Yeah, that's a great question.

MIGNON: Or whether someone has declining cognitive abilities too, you know, like how does someone's writing change as they age, and does that ever get in the way?

NATALIE: Yeah, yeah, yeah. There are, there are definitely studies looking at psychological state and particular language features. So, a researcher by the name of James Pennebaker, I believe he's a psychologist, not necessarily a linguist per se. He's done a lot of work, where it’s quantitative. 

We quantitatively study people's writings and what can you say about their psychological state. So, for example, he and his team have come up with findings such as people who are depressed will use more first-person pronouns “I” than people who are not depressed, who would use, you know, second-person, "you," third-person, "he" and "she."

So that would be an example of quantitative, at least correlation between particular and again, small little details of language use and something about the person. There's also work in sentiment analysis, which is more in the computational linguistic side of things where you consider are there words as you said, indicative of maybe anger?

There are researchers in forensic linguistics who have considered the linguistic characteristics of threats and do threats have more anger words. "Hate." I'm trying to think of some angry words at the moment. "I'm mad at you. I hate you. You're horrible." Do they have these types of strong words that we think, you know, carry strong emotional connotations?

Or are there other characteristics of threatening language? Interestingly it does seem like you cannot neatly categorize threats that way. And another type of word that forensic linguists have looked at in connection with say, for example, is this a threat that's likely to be carried out or not, which you can imagine is very important.

If you get a bomb threat called into a school, for example, it would be really helpful to know if this is something you need to close down the school for. You need to quickly evacuate everybody. Is it a kid, you know, playing a prank who wants to get out of their exam or something like that? But one type of word that linguists look at in this regard is what we would call models.

So "I might hurt you," versus "I will hurt you." "I could hurt you," versus "I'm going to hurt you." You know, these types of things. It's not a clear-cut picture, whether a person who really is going to carry out a threat uses sort of stronger rather than weaker language. "I will do this," versus "I'm considering this," or "I might do it," or "If you don't do this, I might do that."

So it's kind of a mixed bag, and it turns out that threateners actually use a mix of strong and weak language because I guess like all of us, you know, communicating involves kind of a, you know, balancing, so you want … many threateners want something, you know what I mean? So they're making a demand, and so you might include softening language in addition to your more harsh language if you want your demand met. So you might not just say, "I insist this, this and this," you might say, "Well, if you would do this, then I could do that." That sort of negotiation.

MIGNON: I mean, we think of the Unabomber with his type of bombing negotiations.

NATALIE: Yeah. Yeah.

MIGNON: Yeah. Yeah. Well, this is just fascinating. Thank you so much. We're going to wrap up the main section, but if you're a Grammarpalooza subscriber, we're going to continue the discussion and talk about  the interplay of forensic linguistics and AI, maybe some other topics — authorship attribution and the ethics of forensic linguistics. So if you're a subscriber, definitely look for the bonus in your feed. But Natalie Schilling, forensic linguist, what would you like to tell people about where to find you if they're interested in more of your work or hiring you or something like that?

NATALIE: Oh, thanks so much. Yeah. Please look for me on LinkedIn. You can find me there, and that's where my current up to date info mostly resides.

MIGNON: Great. Natalie Schilling, forensic linguist. Thank you.

NATALIE: Thank you.