THE SCIENCE
Psychological measurement derived from language, not inferred by a model.
Scores computed from what someone actually wrote, reproducible across runs, models, and time.
In the early 1990s, Dr. James W. Pennebaker began asking whether the words people use could predict their mental and physical health, their personalities, their social behavior, and their cognitive processes. The answer was yes. That work became the foundation of Receptiviti's measurement.
WHAT PSYCHOLINGUISTIC MEASUREMENT IS
A transparent measurement method - not a black box.
A deterministic, theory-grounded feature-extraction layer. Closer to a scientific instrument than to a learned model.
Each measure is an explicit, documented mapping from language to a psychological category. The frameworks are published and the computation is fixed, so the same input yields the same output across runs, versions, and environments.
These dimensions have detected patterns of distress escalation in LLM-generated responses that standard safety evaluations missed, and have distinguished the language dynamics preceding suicidality and self-injury in a study of 992 individuals with borderline personality disorder.
COGNITIVE
Analytical thinking, cognitive load, certainty, insight, reasoning complexity
AFFECTIVE
14 discrete emotions, positive and negative affect, anxiety, sadness, resolved at the level of specific states rather than valence
SOCIAL
Rapport, affiliation, dominance, authenticity, clout, formality
WELLBEING
Distress signals, inward focus, wellbeing trajectory
INTERPERSONAL
Status dynamics, over-reliance, dependency signals, empathy, agency versus communion
STYLISTIC
Language style matching, deception markers, temporal orientation, fast versus slow thinking
200+ dimensions across cognitive, affective, social, interpersonal, and stylistic categories, traceable to published science.
Example response, before normalization against reference distributions.
LIWC
The most established computational text-analysis tool in psychology and the social sciences.
LIWC, Linguistic Inquiry and Word Count, was created by Receptiviti co-founder Dr. Pennebaker and refined over five major versions spanning three decades. The current version, LIWC-22, was developed and validated with per-dimension reliability and validity reporting.
Receptiviti holds the exclusive commercial rights to LIWC, whose dimensions are a foundational part of the Receptiviti API.
Receptiviti's science team has extended LIWC with hundreds of additional categories using the same validated methodology. Some are available as proportion-based measures; others inform the construction of normed algorithmic scores.
They are built from theory-driven and data-driven methods, with LIWC as a foundational ingredient, and each dimension carries its own reliability and validity reporting rather than inheriting a single figure for the framework as a whole.
Dr. James W. Pennebaker is Regents Centennial Professor at the University of Texas at Austin and was elected to the U.S. National Academy of Sciences in 2025. The foundational methods paper, Tausczik & Pennebaker (2010), The Psychological Meaning of Words, Journal of Language and Social Psychology, is among the most cited articles in its field.
HOW RECEPTIVITI DIFFERS
Psycholinguistic measurement makes psychological characteristics observable.
Self-report surveys tell you how someone perceives themselves. Model inference tells you what a model predicts, without producing anything that can be audited or cited. Psycholinguistic measurement produces explicit, observable scores derived from how someone actually communicates.
SELF-REPORT SURVEYS
X
X
X
X
Social desirability bias. People answer to look good rather than accurately.
Reference-group effects. Self-perception is skewed by who someone compares themselves to.
Requires participation. The person must actively cooperate.
Point-in-time. A snapshot, not a signal that can be tracked continuously.
LLM-BASED INFERENCE
X
X
X
X
Implicit and unauditable. Psychological state influences model behavior without ever becoming an obser vable variable.
Prompt-sensitive. The same conversation yields different inferences across model versions and prompt phrasings.
Circular. Asking the model being evaluated to assess its own impact on the user is not independent measurement.
Not inherently traceable. A model-generated inference cannot necessarily be traced to a documented, validated measurement method.
RECEPTIVITI MEASUREMENT
Derived from natural language. Measures how someone communicates, not how they describe themselves.
Version-stable and deterministic. The same input produces the same score across runs and environments.
Unobtrusive and continuous. Computed from language people produce naturally, per turn, without an assessment.
Traceable to peer-reviewed science. Citable in research, in product decisions, and in accountability conversations.
Self-report requires stopping to ask, which rules it out as a per-turn signal. Psycholinguistic measurement runs on language the person has already written.
RESEARCH
Published research, internal studies, and open questions.
Receptiviti's team and academic advisors publish peer-reviewed research on the psychological dimensions of language and human behavior. Receptiviti also conducts its own research and experiments - contributing to the questions AI safety, alignment, and evaluation teams are actively working on.
PUBLISHED: PNAS Nexus, 2024
Large language models display human-like social desirability biases in personality surveys
Salecha, Ireland et al.,2024. Large language models display human-like response biases when they infer they are being evaluated, with effects up to 1.20 human SD across GPT-4, Claude 3, Llama 3, and PaLM-2. Co-authored by Molly Ireland (Receptiviti).
PUBLISHED: npj Artificial Intelligence · 2026
PsychAdapter: adapting LLMs to reflect traits, personality, and mental health
Vu, Boyd, Eichstaedt et al., 2026. A lightweight LLM architectural modification that generates text reliably reflecting Big Five personality traits (87.3% accuracy) and mental health variables (96.7% accuracy). Co-authored by Ryan Boyd (UT Dallas), academic partner to Receptiviti.
PUBLISHED: Clinical Psychological Science, 2026
Replicability and validity of a new AI assessment of PTSD from patient language
Kjell, Ganesan, Boyd et al., 2026. AI-based psychological assessment from language, when grounded in validated measurement, produces replicable, clinically meaningful results across sequential evaluation with preregistered models.
Co-authored by Ryan Boyd (UT Dallas), academic partner to Receptiviti.
PREPRINT: arXiv, May 2026, under review
When support escalates distress: regulation and escalation in LLM responses to venting and advice-seeking
Chi, Ganesan, Boyd, Ungar & Guntuku, 2026. Across 178,800 Reddit posts, LLM responses to venting simultaneously regulate and escalate distress — and the escalation is invisible to standard safety evaluations. Therapist personas reduce escalation; friend personas increase both. Measured using LIWC-22. Co-authored by Ryan Boyd (UT Dallas), academic partner to Receptiviti.
PUBLISHED: npj Mental Health Research, 2025
Psychosocial dynamics of suicidality and nonsuicidal self-injury: a digital linguistic perspective
Entwistle, Hoemann, Nightingale & Boyd, 2025. Large-scale naturalistic study of the language dynamics surrounding suicidality and self-injury in 992 individuals with borderline personality disorder (66,786 posts). Co-authored by Ryan Boyd (UT Dallas), academic partner to Receptiviti.
PUBLISHED: Perspectives on Psychological Science, 2026
Artificial intelligence and the psychology of human connection
Boyd & Markowitz, 2026. Introduces the MIRA model - a theoretical framework for when and how AI functions as a relational entity in human ecosystems. Language is the primary modality through which that relationship operates. Co-authored by Ryan Boyd (UT Dallas), academic partner to Receptiviti.
INTERNAL STUDY
Providing psychological user state as context to educational AI
GPT Study Mode: Providing psycholinguistic signals produced a +4% improvement in overall educational effectiveness across 25 blinded evaluations. Largest gains in reasoning & scaffolding (+6.3%) and cognitive-load management (+5.9%).
SCOPED
Psychological state as a missing dimension in AI evaluation
The case for human-state signals as a first-class eval criterion alongside accuracy, helpfulness, and harmlessness.
34,000
Peer-reviewed citations to the psycholinguistic science behind these dimensions
30+
Years of foundational research validated across health, personality, social behavior, deception, and clinical contexts
34,000 citations.
Not a marketing number - a research record.
That record spans health psychology, clinical research, personality science, organizational behavior, forensic contexts, and deception research. The dimensions were not built for AI. They were validated across decades of work in other fields first, which is why they hold still when models change.
Every dimension is traceable to published science. That traceability is what makes the scores auditable, and what separates them from inference that cannot be cited or challenged.