What Is Goldberg's Lexical Hypothesis?

The lexical hypothesis holds that the personality differences that matter most in social life have been encoded in the natural language of a community. If a trait is stable, shows in how people deal with one another and has consequences, the language very probably has a word, or several, to name it.

The idea predates Lewis Goldberg, but he was the one who stated it explicitly and turned it into the method that led to the Big Five model. That’s why it’s often known by his name.

From Galton to Goldberg: a century of counting words

The intuition that the dictionary holds a map of character has a long history:

Francis Galton (1884) was among the first to try it: leafing through a dictionary, he estimated that English had around a thousand words to describe character.

Gordon Allport and Henry Odbert (1936) did the systematic count. In Webster’s dictionary they found 17,953 terms related to individual differences, of which about 4,500 described stable traits, such as “sociable” or “meticulous,” rather than passing states or value judgments.

Raymond Cattell, in the 1940s, cut that list down by grouping synonyms into a few dozen variables, which he analyzed with factor analysis. That produced his sixteen-factor model.

Ernest Tupes and Raymond Christal (1961), reanalyzing Cattell’s variables with data from the U.S. Air Force, found five recurrent factors. Warren Norman (1963) replicated them. The five-factor model was already there, although it would take twenty years to catch on.

Lewis Goldberg, from 1981 on, gave the hypothesis its explicit formulation and put it to the test with large lists of adjectives. In 1990 he published the studies showing that, with different lists and methods, English personality adjectives always organized themselves into the same five dimensions, which he called the Big Five. The full history of the model is in who invented the Big Five.

What the hypothesis says exactly

Goldberg (1981) set it out on two levels. In paraphrase:

• Fundamental hypothesis: the individual differences that are most significant in people’s everyday interactions with one another eventually become encoded in their language.
• Corollary: the more important an individual difference is, the more people will notice it and want to talk about it, and the more likely it is that a word for it exists, or even several, and in more languages.

The corollary has a direct methodological consequence: traits with many synonyms (in English, “kind,” “considerate,” “cordial,” “amiable”) should correspond to important dimensions, and should appear across different languages.

How it is tested

The lexical method follows a fairly stable sequence:

1. Collect all the personality adjectives in a reference dictionary.
2. Clean the list, removing evaluative terms (“good,” “awful”), transient states (“angry”) or rare words.
3. Gather ratings: hundreds of people indicate how well each adjective describes them or someone they know well.
4. Analyze which adjectives tend to go together using factor analysis. If “talkative,” “sociable” and “lively” are rated similarly for the same people, they share a dimension.

The advantage is that the researcher doesn’t start from a theory about what the traits “should be,” but from what the language already distinguishes. In English, the result was the five dimensions that a Big Five test measures today: extraversion, agreeableness, conscientiousness, emotional stability and openness (or intellect).

What happened when it was applied to other languages

The hypothesis predicts that the important dimensions should show up in many languages. When this was checked, the results were more nuanced than expected:

• Six factors instead of five. Ashton, Lee and colleagues (2004) reanalyzed lexical studies in seven languages (Dutch, French, German, Hungarian, Italian, Korean and Polish) and repeatedly found a sixth factor, honesty-humility, which gave rise to the HEXACO model.
• The “Big Seven” in Spanish. Benet-Martínez and Waller (1997), working with Spanish adjectives, found seven factors. The five familiar ones were joined by two purely evaluative dimensions, positive valence (“exceptional,” “admirable”) and negative valence (“cruel,” “wicked”), which appear when evaluative terms aren’t excluded from the list.
• Only three fully universal factors. De Raad and colleagues (2010) compared fourteen taxonomies from twelve languages and concluded that only three dimensions replicated in all of them: extraversion, agreeableness and conscientiousness. Emotional stability and, above all, openness changed in content depending on the language.

These findings don’t overturn the lexical hypothesis. Rather, they confirm it in its weaker form: language does store the important differences, but each language groups them in its own way, and some dimensions are more universal than others.

Criticisms and limitations

The lexical hypothesis also faces more fundamental criticisms:

• Not everything important gets a name. Some relevant differences aren’t well captured by everyday language, such as need for cognition or certain motivational styles.
• The dictionary reflects a culture. Vocabulary shows what a society has considered worth naming, not necessarily what matters most from a biological perspective.
• It captures the visible better. Adjectives describe how someone looks from the outside. Internal, physiological or unconscious processes are underrepresented.
• It describes; it doesn’t explain. Finding five factors in language doesn’t tell us why they exist. That takes the explanatory theories that were developed later.

Even so, it remains one of the most solid methodological foundations of personality psychology, precisely because it is empirical and can be repeated in any language.

If you’d like to see in practice the five factors that emerged from this hypothesis, you can take our Big Five test.

Sources

• Allport, G. W., & Odbert, H. S. (1936). Trait-names: A psycho-lexical study. Psychological Monographs, 47(1), i-171.
• Ashton, M. C., Lee, K., Perugini, M., Szarota, P., de Vries, R. E., Di Blas, L., Boies, K., & De Raad, B. (2004). A six-factor structure of personality-descriptive adjectives: Solutions from psycholexical studies in seven languages. Journal of Personality and Social Psychology, 86(2), 356-366. https://doi.org/10.1037/0022-3514.86.2.356
• Benet-Martínez, V., & Waller, N. G. (1997). Further evidence for the cross-cultural generality of the Big Seven factor model: Indigenous and imported Spanish personality constructs. Journal of Personality, 65(3), 567-598. https://doi.org/10.1111/j.1467-6494.1997.tb00327.x
• De Raad, B., Barelds, D. P. H., Levert, E., Ostendorf, F., Mlačić, B., Di Blas, L., Hřebíčková, M., Szirmák, Z., Szarota, P., Perugini, M., Church, A. T., & Katigbak, M. S. (2010). Only three factors of personality description are fully replicable across languages: A comparison of 14 trait taxonomies. Journal of Personality and Social Psychology, 98(1), 160-173. https://doi.org/10.1037/a0017184
• Galton, F. (1884). Measurement of character. Fortnightly Review, 36, 179-185.
• Goldberg, L. R. (1981). Language and individual differences: The search for universals in personality lexicons. In L. Wheeler (Ed.), Review of Personality and Social Psychology (Vol. 2). Sage.
• Goldberg, L. R. (1990). An alternative “description of personality”: The Big-Five factor structure. Journal of Personality and Social Psychology, 59(6), 1216-1229.
• Norman, W. T. (1963). Toward an adequate taxonomy of personality attributes: Replicated factor structure in peer nomination personality ratings. Journal of Abnormal and Social Psychology, 66(6), 574-583. https://doi.org/10.1037/h0040291
• Tupes, E. C., & Christal, R. E. (1961). Recurrent personality factors based on trait ratings (Technical Report ASD-TR-61-97). U.S. Air Force.