When the meaning of words changes, will people of all ages follow the trend, or do older speakers remain linguistically stuck in the past while the younger generations are leading? Large-scale linguistic analysis published in Proceedings of the National Academy of Sciences This suggests that the semantic changes are more comprehensive than previously thought. Younger people tend to adopt new meanings a little faster, but older speakers usually continue within a few years and sometimes lead.
This finding is contrary to sociolinguistics' longstanding view that language evolves primarily through generational departure. Instead, the results demonstrate a more dynamic process in which speakers of all ages participate in real-time shifts in how words are used.
The researchers set out to test basic assumptions in the study of language change. It is whether older people maintain stable linguistic patterns throughout their lives, or whether they update their language use in response to changes in the broader speech community.
For decades, sociolinguists have relied on the “apparent time” method. This compares the language of older and younger people at one time point and infers change across generations. This approach rests on the idea that adult language use is relatively fixed. Instead, if older speakers regularly adjust to current trends, these assumptions may not retain these assumptions as to how the meaning of words evolves.
Although previous research has primarily supported generational change models, particularly in terms of pronunciation and grammar structure, the question of whether words' meanings follow the same pattern can remain relatively unexplored, especially on a large scale.
“What led us to explore the topic was the fact that simple questions were not actually answered yet. Does the words change meaning and people of all ages continue?” said Gaurav Kamath, a doctoral student in linguistics at McGill University and the lead author of the paper. “It's a more broad and important question for language change because (i) sociolinguists often assume that older speakers are windows into the past (this is only true if they don't adopt changes), and (ii) as adults, they also teach us something about our ability to change the way we speak.
To address this gap, the researchers analyzed speeches from more than 7.9 million US Congresses between 1873 and 2010. These speeches were given by thousands of speakers whose age was known at the time of each speech, and provided a rare opportunity to control the age of speakers while tracking verbal behavior for nearly 140 years.
Researchers focused on a set of around 100 words that are likely to mean change in the 20th century. Examples include words such as “monitor”, “article”, “satellite”, and “exceptional”. Each of these words was examined for multiple possible meanings called “sensory,” using advanced linguistic models that predict the context-based use of each word. These predicted meanings were grouped using clustering algorithms to identify a clear sense of each word.
For example, the word “article” can refer to a physical product, legal provision, or written story. By analyzing the context in which words appear and modeling the rise or fall of each sensation over time, researchers were able to demonstrate how meanings change over different periods.
To determine whether age influenced the adoption of new meanings, the team used a statistical model that predicted the likelihood that speakers would use certain word sensations based on both the year and the speaker's age. These models estimated whether older speakers used outdated sensations or whether they adopted new sensations at slower or faster speeds compared to younger colleagues.
The researchers also conducted Bayesian meta-analyses to calculate mean age-related delays across all word sensations. This allowed them to quantify how slow the old speakers are to adopt a new meaning.
Throughout the dataset, researchers found that changes in the meaning of words are overwhelmingly driven by collective changes over time, not just by generational exchanges. Younger speakers tended to adopt new meanings a little earlier, while older speakers were less late. On average, older speakers were about two to three years behind younger speakers when it came to adopting new word meanings.
In many cases, this delay was so minimal that older speakers could not be viewed linguistically in the “back”. For example, older lawmakers in the 1960s might use a new sense of new words like “articles” just a few years after their younger colleagues have already started doing so. For a small number, older speakers actually led the shift. This is similar to the geopolitical sense of the term “satellite” that became prominent during the Cold War era.
“The main result of the old speakers being able to adapt very well to the meaning of new words was a surprise in itself,” Kamas told Psypost. “But the bigger surprise was that for some of the words we saw, we even found evidence that the old speaker was that. reading change. ”
This result provides evidence that meaning change tends to be a “Zeitgeist” effect, a product of cultural and temporal moments, rather than a strict intergenerational handoff. Even at individual levels, the speakers adjusted their use over time. When examining a handful of prolific speakers who frequently use the same words for decades, researchers observed significant individual changes in how they used those words, closely following the broader shifts in usage patterns.
“In a nutshell, older people pick up new meanings of words,” Kamas explained. “Another way to put it down – this is proof that your parents/grandparents can actually use words like “sick” (i.e. “cool”) and “model” (i.e. “AI model”) in new sensations that are increasingly dominant. ”
These findings influence the way linguists model and interpret language changes. If older speakers frequently adopt modern usage, the differences observed in cross-section data may not fully capture the speed or nature of ongoing changes. In fact, obvious time comparisons may underestimate the extent of already ongoing changes, as the linguistic behavior of older speakers quickly converges with the younger ones.
The results also illustrate the power of computational approaches to studying large-scale semantic changes. By leveraging large-scale text corpus, speaker metadata, and advanced natural language processing models, researchers were able to draw conclusions that are difficult to reach using small-scale observational studies.
“This study demonstrates the possibility of using natural language processing (NLP) tools to study human language, and we believe it hopes to encourage further work on using NLP tools for language research,” Kamath said.
However, there are some limitations. This study focused only on adult speakers as US Congress membership requires that individuals be at least 25 or 30 years old. Since teenagers and young adults are often the earliest adopters of language innovation, this analysis may overlook the beginning of a particular change in meaning.
The dataset also reflects certain sociopolitical groups (particularly those that tend to share in previous decades) that tend to share certain demographic characteristics. Results may not be entirely generalised to a wider population or speakers outside the US.
“The main limitation to keep in mind is seeing the Congress speech,” Kamas said. “The reason we relied on data for this genre was because it was the only data that could track the ages of thousands of speakers over 140 years. However, the speakers we studied (members of Congress) were underestimated by arrangement and minority, and, importantly, our research typically did not include language.
Furthermore, the linguistic models used in this study were generally effective in identifying different meanings, but they are undoubtedly. Several errors in sensory classification may remain, especially when word usage is ambiguous or frequent.
“The next step is to broaden the scope of this research and find ways to address the limitations mentioned above,” Kamas said. “Can we include a more balanced demographic sample that extends beyond North American English? What about other languages and societies? And what about speeches from youth?”
“Adult semantic changes are not primarily generational phenomena,” the study was written by Gaurav Kamath, Michelle Yang, Siva Reddy, Morgan Sonderegger, and Dallas Card.
