A 2024 review of the love-languages theory found that most people rate all five categories as important rather than one dominant preference, and that couples matched on their stated language were no more satisfied than couples who were not

Gary Chapman published The 5 Love Languages in 1992, proposing that every person has one primary way of feeling loved, drawn from five categories: words of affirmation, quality time, acts of service, physical touch, and receiving gifts. Chapman, a Baptist pastor, built the framework from his own counselling sessions with couples, not from a study, a survey, or a dataset. The idea has since sold more than 20 million copies and turned into an industry of quizzes, corporate workshops, and dating-app profile prompts. What it has not done, until fairly recently, is get tested.

What happens when someone actually tests it

The first real attempt came in 2006, when the communication researchers Nichole Egbert and Denise Polk asked 83 couples to report both their own preferred love language and how they actually behaved toward their partner, then measured the couples’ relationship quality. If Chapman’s theory held, couples whose partners’ behaviour matched their stated preference should have reported a better relationship than couples whose partners guessed wrong. Egbert and Polk found no such difference. Matched couples and mismatched couples reported statistically similar relationship quality. The study did find that people who engaged in general relationship-maintenance behaviour, the ordinary work of keeping a partnership functioning, also tended to express love in ways that mapped loosely onto Chapman’s five categories, but that is a different claim from the one the book makes.

This is one small study, not settled consensus, and it did not end the conversation. But it was the first sign that the theory’s central mechanism, matching a partner’s specific preferred language, might not be doing the work the book claims it does.

The 2024 review that actually took the theory apart

A more recent and considerably more thorough test came from Emily Impett, Haeyoung Gideon Park, and Amy Muise, in a 2024 review published in Current Directions in Psychological Science. The three relationship scientists set out to evaluate Chapman’s theory against its own three claims: that each person has one primary love language, that there are exactly five of them, and that couples are more satisfied when a partner speaks their preferred one.

On the first claim, the review found that when people are asked to rate the importance of all five categories rather than forced to rank them against each other, most people rate all five as meaningful, not one standout above the rest. Forced-choice quizzes, the format almost everyone encounters online, manufacture the appearance of a single dominant preference that a more open-ended measure does not reproduce. Fewer than half of the people studied across the reviewed research had a clearly identifiable single primary language once the forced-choice format was removed.

On the third and most practically important claim, whether satisfaction improves when a partner’s behaviour matches a person’s stated love language, the review found no consistent supporting evidence. Across the studies it examined, matching and mismatching partners did not reliably differ in reported relationship satisfaction. That is the same basic result Egbert and Polk found in 2006, now confirmed across a larger and more recent body of work rather than resting on one small sample.

Why the idea still feels true anyway

None of this means the five categories are meaningless as descriptions. Words of affirmation, quality time, acts of service, physical touch, and gifts are all real things people do for each other, and naming them gives couples a shared vocabulary for talking about affection that many did not have before. The Impett, Park, and Muise review does not argue that these behaviours don’t matter. It argues that the specific claims Chapman built around them, one dominant language per person, exactly five categories, and a satisfaction boost from matching, are not what the evidence shows.

The more useful comparison the review offers is a shift in metaphor: rather than picture love as a single language a couple needs to learn to speak correctly, picture it as something closer to a balanced diet, where a relationship draws on a full range of these behaviours, plus other ingredients like companionship, mutual respect, and emotional support, rather than optimising for one preferred nutrient. A couple that only ever expresses affection through gifts and never through time or affirmation is missing something, in this account, not because gifts are the wrong language for one partner, but because a relationship fed from only one source tends to be thinner than one fed from several.

What the corporate and dating-app version gets wrong

The love-languages framework’s real commercial life, in HR team-building exercises, dating-app prompts, and relationship-advice content, treats identifying a single love language as the diagnostic step: find your language, find your partner’s, then speak it correctly and the relationship improves. The research complicates that pipeline at its first step. If most people value all five categories rather than one, the entire premise of matching languages as a satisfaction lever loses its footing before the second step even begins.

That does not make the exercise of discussing preferences with a partner pointless. Asking a partner what makes them feel appreciated, and actually listening to the answer, is a reasonable thing for a couple to do regardless of whether the underlying five-category theory holds up. What the research suggests is that the value in that conversation comes from the act of asking and paying attention, not from correctly identifying which of five preset boxes a person belongs in.

What this does not settle

The Impett, Park, and Muise review is itself an evaluation of existing studies rather than a single new experiment, and the underlying research base on love languages remains smaller than the size of Chapman’s readership would suggest. It is possible that future, more precisely designed studies could still find a narrower version of the matching effect that current methods have missed. What the available evidence supports today is more modest than the popular version of the idea: the five categories describe real behaviour, but the theory that one of them dominates for each person, and that satisfaction depends on getting the match right, has not held up when researchers have actually gone looking for it.