The study, published in the peer-reviewed journal Science Advances, introduces an innovative analytical approach called "ousiometrics," utilizing advanced mathematical methods to measure the essential meaning of language across vast datasets. Among its most striking conclusions is that the words humans use in everyday communication display a consistent, widespread, and largely unrecognized preference for safety over danger.
This hidden safety bias has the potential to transform how researchers study linguistics, analyze human behavior, and develop artificial intelligence systems designed to interpret human speech and writing. For decades, the academic community has relied heavily on emotional frameworks to evaluate text. However, this new research argues that those emotional metrics may actually be masking a more primitive, survival-oriented linguistic structure.
Scientists Discover a Hidden Safety Bias in Human Language
Since the 1950s, much of the scientific understanding regarding word meaning has relied on a foundational framework developed through the pioneering research of psychologist Charles Osgood and his contemporaries. Known as the VAD model, it describes emotional meaning through three distinct dimensions: valence, which measures positive versus negative feelings; arousal, which measures excitement versus calmness; and dominance, which measures control versus submissiveness.
This influential approach has shaped decades of work in psychology, linguistics, and computer science, particularly serving as the backbone for artificial intelligence systems designed to perform sentiment analysis and identify the emotional tone of written text.
However, the team at the University of Vermont found compelling evidence that this long-established model overlooks crucial aspects of how human meaning is structured and organized. Their rigorous analysis examined billions of individual uses spanning more than 20,000 distinct words across a wide and diverse range of real-world texts. Utilizing advanced computational techniques, the research team identified a completely different underlying architectural structure to language.
They discovered that the three dimensions within the traditional VAD model are not independent of one another, as previous generations of scientists had assumed. Instead, their comprehensive results indicate that word meaning can be much more effectively and accurately described through three entirely separate dimensions: power, measuring weak versus powerful; danger, measuring safe versus dangerous; and structure, measuring ordered versus chaotic.
The research, which received financial support from the US National Science Foundation, Google, MassMutual, and other funding bodies, revealed that this new framework accounted for more than 90 percent of the variation in measured word meaning. By comparison, the traditional VAD approach explained approximately 72 percent of that variation.
This significant improvement in explanatory power could prove especially vital as large language models, automated content moderation tools, and other complex AI technologies increasingly depend on an accurate, nuanced understanding of human communication.
Why Human Language Favors Safety Over Danger
One of the study’s most unexpected and revealing discoveries emerged when the researchers investigated how words actually appear in everyday communication, rather than merely cataloging words in isolation. Across an expansive library of books, daily news reporting, social media feeds, and spoken language transcripts, words associated with safety appeared with a much higher frequency than those associated with danger.
This striking finding offers an entirely new explanation for a familiar linguistic and psychological phenomenon known as the Pollyanna principle, which describes the well-documented tendency for human language to lean toward positive rather than negative words. For decades, researchers generally interpreted this recurring pattern as direct evidence that people naturally favor positive emotional expressions and optimism.
The new analysis, however, suggests that something much more fundamental may be responsible for the skew. According to the researchers, the apparent preference for positivity in daily communication largely reflects an underlying, systemic preference for safety.
"The Pollyanna principle’s positivity bias," the study concludes, "is, in fact, a one-dimensional projection of an underlying safety bias."
Peter Dodds, director of the University of Vermont’s Complex Systems Institute and senior author of the study, emphasized the broad implications of the discovery. "This is a big observation that comes out of this work," Dodds said. "Expressions of safety are crucial to all language."
Could Human Language Be Shaped by Survival?
The discovery raises an intriguing and profound evolutionary possibility: the very structure of human language may reflect ancient evolutionary pressures associated with staying alive. From this perspective, words do far more than merely communicate transient feelings or emotional states. They also help people recognize external threats, evaluate uncertainty, establish vital trust, and coordinate complex actions with others within a community.
Being able to reliably distinguish between safety and danger is essential to survival. The researchers suggest that this basic human priority may also heavily influence how meaning is organized and prioritized in language. Throughout daily life, people constantly exchange information regarding whether individuals, places, physical situations, and specific behaviors are threatening or safe.
The new findings suggest that these vital distinctions may be far more central to human communication than previously recognized. They also offer a compelling alternative interpretation of why positive language is so common in daily discourse. Rather than simply expressing happiness or fleeting approval, many positive words may actually convey a deeper sense of security, stability, or predictability.
Julia Zimmerman, a postdoctoral researcher in the University of Vermont’s Computational Story Lab and a co-author of the study, noted that the three newly identified dimensions reflect experiences that are universally fundamental to human life. "Power, danger, and structure," she said, "are relevant to every person that’s ever lived."
Researchers have also long recognized that people tend to favor language associated with goodness and low aggression. The research team argues that these familiar linguistic tendencies may arise from the exact same deeper pattern. "We now understand," the team writes, that these are "shadows of an underlying linguistic safety bias."
Although the results are entirely consistent with a possible connection between language and biological survival, the authors caution that the findings do not definitively establish that evolutionary pressures directly caused the safety bias. Further interdisciplinary research will be required to explore the exact mechanisms driving the pattern.
A Discovery That Could Change How AI Understands Language
The implications of these findings extend well beyond theoretical linguistics, offering practical applications for technology and the social sciences. Many contemporary artificial intelligence systems rely heavily on sentiment analysis to determine whether written or spoken language expresses positive, negative, or neutral feelings. Some advanced approaches also draw directly on the traditional VAD framework to measure emotional meaning.
If that long-standing framework does not fully capture the true underlying structure of human language, systems built entirely around it could easily overlook critical distinctions in what people are actually communicating. The researchers suggest that incorporating power, danger, and structure into algorithmic models could eventually help AI systems interpret language more accurately and explain their underlying assessments more clearly. This capability could prove particularly useful when analyzing sensitive matters involving trust, financial risk, and human decision-making.
In the field of linguistics, the findings offer an alternative way to understand the very foundations of meaning. Instead of treating emotional tone as the primary organizing principle, the new framework emphasizes core distinctions related to strength, safety, and predictability.
Psychology could likewise be impacted by the shift. Decades of behavioral studies have utilized the VAD model to investigate human emotions, environmental perception, and social behavior. A different set of fundamental dimensions could encourage researchers to reconsider how some of those historical findings are interpreted.
The discovery may also capture the attention of neuroscientists. The human brain is exceptionally sensitive to potential threats and signals of safety, and the newly identified linguistic bias suggests that these biological priorities are deeply mirrored in our symbolic communication. This potential connection could help researchers explore complex relationships between neural activity and language, though additional targeted research would be required to establish direct links.
A New Tool Measures the Hidden Meaning of Words
To investigate these intricate patterns, the researchers developed sophisticated analytical methods designed to measure meaning across enormous, unprecedented collections of written and spoken language. One of their central technical innovations is a new instrument called the "ousiometer," which rapidly evaluates the essential meaning of large bodies of text and produces a comprehensive average meaning score. The term "ouisa" is derived from Ancient Greek, serving as a root for the English word "essence."
The ousiometer builds directly upon an earlier tool developed by the research team known as the "hedonometer," commonly referred to as a happiness meter. While that earlier instrument was designed to measure the overall emotional tone of language, the new ousiometric approach examines broader, more foundational patterns of meaning.
The researchers successfully demonstrated that the ousiometer can analyze material drawn from remarkably diverse sources, ranging from Jane Austen’s classic novels and Arthur Conan Doyle’s Sherlock Holmes stories to the New York Times, Wikipedia, public Twitter feeds, and transcripts of talk radio programs. The tool is also capable of tracking precisely how meaning shifts and evolves as a narrative unfolds over time.
For example, the researchers analyzed an English translation of Victor Hugo’s sprawling novel Les Misérables and mapped its changing meaning throughout the narrative arc. They termed this changing pattern an "ousiometric trajectory." The resulting data visualization resembles the twisting, complex shape of a multicolored protein, with the narrative moving across a multidimensional grid based on four contrasting pairs: dangerous versus safe, weak versus powerful, gentle versus aggressive, and bad versus good. By carefully following this computational path, researchers can observe how the underlying meaning of a text shifts from one section of a story to another.
Why Counting Word Frequency Changes the Results
Another crucial innovation involved distinguishing between words themselves as dictionary categories, known in linguistics as "types," and how often people actually use those words in practice, known as "tokens." For instance, as a general category, the word "apple" is a type, while every single time the word "apple" is spoken or written in a sentence counts as a token.
This distinction is critically important because a word that appears millions of times in daily discourse contributes vastly more to everyday human communication than a word that is rarely used. Earlier research often gave every dictionary word equal analytical importance, regardless of its actual frequency in real-world usage.
The Vermont team instead incorporated real-world word frequencies into their models, allowing subtle, widespread patterns to emerge that might otherwise have remained completely hidden. This methodological approach was particularly essential for identifying the widespread and pervasive preference for safety-related language across society.
The ten-member research team was led by Peter Dodds and Chris Danforth, both professors in the University of Vermont’s College of Engineering and Mathematical Sciences. Their broad international collaboration included researchers from the Santa Fe Institute, the Complexity Science Hub in Austria, the Howard Hughes Medical Center, the University of California, Berkeley, the University of Adelaide, and MassMutual Data Science.
Rethinking What Human Language Reveals About Us
Understanding this hidden safety bias could ultimately have profound consequences for how researchers study communication across human society. If words consistently reflect an underlying preference for safety, that psychological tendency may influence how stories are structured, how news and information spread across populations, and how people interpret everyday events and personal experiences.
The findings could eventually inform future research into political communication, public health and mental health messaging, and the design of conversational AI systems intended to safely interact with human users. More broadly, the study challenges the long-held assumption that human language can be understood primarily through surface-level emotions such as happiness, fear, and excitement.
Instead, the researchers propose that human meaning possesses a much deeper architectural structure connected fundamentally to power, danger, and order. Their findings suggest that language directly reflects the persistent challenges humans face when navigating uncertain environments, maintaining fragile social relationships, and identifying potential threats in the world around them.
By revealing this previously overlooked structure, the research team offers an entirely new way to understand what words communicate and why certain kinds of meaning dominate everyday conversation. Human language, their empirical research suggests, may be far more closely connected to our fundamental need for safety than scientists have traditionally recognized.