Lessons - The Transformative Power of Language | Valerie Fridland, Professor, Sociolinguist, and Author

Welcome to Lessons episodes of Success Story, part of the HubSpot podcast network. These lessons episodes will be shorter conversations with past guests, valued members of the success story community, and myself. They'll be focused on teaching you actionable, insightful takeaways that you can use to upscale your personal and professional life. I think anybody listening understands how important language is and everybody who's listening has had these moments in their life when they've been judged or they've even judged someone else. But it doesn't seem like quite literally before I met you, I've never heard of somebody studying this as a profession. I mean, the only application that I can see this in day to day would be maybe actors and actresses and people that use this as part of their profession. But this is so, language is so powerful and it seems like the lack of understanding of language and how it impacts is something that could change people's careers, change people's lives, change their ability to negotiate the next job or present a certain way. And there's all these other skills that are about, are sort of adjacent to communication that people focus on, but they never focus on language. I've never heard somebody say, like, don't say this word or say this word as part of, you know, they say, don't say, um, or what not and try and cut out the filler words when you're doing a presentation and maybe to be a bit more confident. But I've never seen somebody, and actually, apparently according to your research, that's not even the best way to present. But the point is, I've never seen somebody focus on this before. So what does this impact, what kind of impact does this study have? What does this actually teach people? What kind of influence could this have on somebody's life? Like what are some examples of how drastic altering language could be on an individual? Well, you know, it's interesting that you say that because it's true, people generally have never heard of the social linguist or even a linguist that's a theoretical linguist like I am. So we talk about linguists that study a lot of languages, but the type of linguist that I am is someone who studies the underlying structure of language, how we produce sounds, why certain things happen a certain way? How languages change over time? What are the social and linguistic forces that create the language we speak? Why did old English sound so different than modern English? All those kinds of things. And we also try to trace back the evolution of where all languages come from. Were they from one language, when proto-language, or was it sort of multiple origin stories for language? Were all those big questions that we don't think about as speakers of a language? We think about what's a noun, what's a verb, and you know, where's my dangly modifier? And those are really interesting because those are actually social preferences that we have learned when we study English language arts. What we're studying is one person's version of what they like about language and what they have claim language should be. Those are not the actual cognitive rules that create language and drive it forward. So it's really kind of striking to me that we don't know this side of language because it's so pivotal in everything we do. Do you have a voice assistant, like a virtual assistant that you talk to Alexa or Siri? Linguist, linguist, linguist, right? If it wasn't for people like me, you wouldn't be able to talk to them. So someone has to figure out how to program those computers to understand human speech. What is the structure of a senate so that we make sure that when Siri spits something out, she follows that structure, right? And analysis, and she's using large data pool. So she's doing analysis that is a syntactic analysis of those speech features. So this predictive language model that is driving like chat GPT today, what it does is it looks at huge quantities of language data and a lot of them have been pruned and approved by linguists like this is something the types of language data we want. A lot of the language data comes from linguists. In fact, I have a friend that runs a large corpus that is a linguist and a lot of his lately, a lot of his work is trying to field request from these large language data runners to get his access to his corporate to help them build language models for these big chat GPT type things. So that chat GPT is a predictive model, meaning that based on its analysis, it's sort of linguistic analysis of all these sets of data, it predicts what comes next. So if you have a sentence, it's looking at massive quantities of data and saying, what is the statistical probability that this type of word comes next? And this is exactly what you do as a human language model, right? You are predicting, when you hear someone talk, you are predicting, oh, they said the nouns follow the, that's probably going to be a noun. And that's how I understand if you use a word that's not typically noun and put it after a the. All of a sudden it becomes a noun because my brain understands that the makes things a noun, that that slot in syntactic structure makes a noun. So another great example is adulting, right, that we've made into a verb, right, people of course hate that, but what they don't know is parent and parenting is actually the same relationship parent, the noun came centuries before we started talking about parenting. So you could be a parent in the 14th century, but you didn't do parenting. Really, the first reference to parenting we find is in the 16th century and that it was hardly ever used until the 20th century, but parent has been used for a long time. Well adult adulting has that exact same distribution or adult came first and it meant to be a grown up, right, a grown up person, but adulting carries with it not just the idea of I'm being I'm grown up. It carries all the things associated socio culturally with being an adult. And it's by adding that ing that it signals to me, okay, I've shifted that noun to a verb. So these are actually linguistic ways of looking at language that are super helpful in both, you know, data for language learning models in making sure Siri recognizes a southern accent versus a non-southern accent by giving it the understanding of the vowels. For example, that a southern uses versus a non-southern or it's also really important in educational context because a lot of times we have non-native or non-standard speakers. Both of those go to schools along with native and standard speakers and we can have some problems in terms of how educational attainment is met for different pools of language learners or different pools of dialect speakers and it doesn't mean they're sending you wrong with those dialects, but we can use the models of standard English and the models of dialects or non-standard to try to compare them to help children achieve learning attainment by giving them really solid understandings of where their dialect differs from the one that is the socially preferred dialect. Now notice I said social preferred because there's nothing wrong with a standard dialect. It's simply a different set of rules, but how do you learn rules? You learn it by having them articulated in English language arts. We don't articulate. Like here's the rule where English uses a noun after a the, here we don't articulate those kinds of rules. We give people ideas like nouns or person places or things which actually doesn't explain most nouns, especially abstract ones. That's not really linguistics, that's actually just sort of ideas about language that get floated around. So there are lots of ways that linguists are relevant to our daily lives. Many professional circles, linguists do a lot of things like studies on that as well as communication researcher studies on how different types of language use is perceived in different contexts. So one place for example that I give talks on a lot is accent discrimination and how what kind of research linguists do can show us where we're making some errors in the way that we approach accents in workplaces. And sometimes it's really surprising. So for example, if you see, if we do experiments where we see an Asian face, a South Asian face and we play a standard English native speaker voice and then we compare that to a different control group that saw a white face with the same exact voice, what we find is ratings of intelligibility and ratings of non-native itness go significantly up when you show them a picture of a non-white person, which means you're not actually listening to the actual signal because it's exactly the same voice, but just your stereotypes about what that person should sound like, influences strongly what you think they sound like and can affect intelligibility. But what we find is people that are massive bias when it comes to hiring and what not. And then what's the solution, that's the other thing. So what we find in also doing the same studies is if we expose speakers to more talkers from diverse backgrounds, they get better at not processing it in a negative way. So it can decrease accent bias. So there are a lot of different places where linguists are important and people just don't