You have recently joined a book club.

Before each meeting, one member of the literary collective sends an email to the club secretary offering a few thoughts on the assigned text. This month, it’s your turn to compose the brief review.

A new study suggests that the words you use may depend on whether the club secretary’s name is Emily (“a stereotypically White name,” as the study says) or Lakisha (“a stereotypically Black name”). If you’re a white liberal writing to Emily, you might use words like “melancholy” or “euphoric” to describe the mood of the book, whereas you might trade these terms out for the simpler “sad” or “happy” if you’re corresponding with Lakisha.

But if you’re a white conservative, your diction won’t depend on the presumed race of your interlocutor.

This racial and political disparity is among the discoveries made by a pair of social psychologists in a paper forthcoming in the Journal of Personality and Social Psychology, a peer-reviewed scientific journal published by the American Psychological Association. Cydney Dupree, an assistant professor of organizational behavior at the Yale School of Management, and Susan Fiske, a professor of psychology and public affairs at Princeton, documented what they call a “competence downshift” exhibited by white liberals in interactions with racial minorities and with black people in particular.

The findings, based on what the authors stress is “preliminary evidence,” raise difficult questions about aspirations for a so-called post-racial society. The results reveal how subtle forms of discrimination may coincide with progress toward equal treatment, or what the paper identifies as “a significant reduction in the expression of explicit prejudice and endorsement of negative stereotypes.”

The psychologists further discovered that white liberals rarely admit to the goal of appearing less competent, a fact that highlights the role of implicit bias and “the covert nature of the competence downshift strategy.”

“White liberals may unwittingly draw on negative stereotypes, dumbing themselves down in a likely well-meaning, ‘folksy,’ but ultimately patronizing, attempt to connect with the outgroup,” argues the paper, titled “Self-Presentation in Interracial Settings: The Competence Downshift by White Liberals.”

The findings could provide a new arrow in the quiver of those who decry identity politics practiced by liberals, and yet the paper hardly applauds conservatives for their approach, reasoning that they are simply “less motivated to affiliate with racial minorities.” In other words, the paper states, white conservatives “would not bother.”

“It’s somewhat counterintuitive,” said Dupree, who is the lead author and whose research was supported by the National Science Foundation as well as by Princeton’s Joint Degree Program in Social Policy. “The idea that people who are most well intentioned toward racial minorities, the people actually showing up and wanting to forge these connections, they’re the ones who seem to be drawing on stereotypes to do so.”

At the same time, she said, the findings are in line with what research has already concluded about the persistence of stereotypes even as more overt bias diminishes. What’s new is the paper’s focus on a population that has received less attention: people most likely to see themselves as allies of racial minorities.

White liberals, she said, may not endorse stereotypes painting black people “as lower status and less competent,” as the paper notes. But they’re nevertheless aware of these ideas, she explained, “and they may be using them to try to get along in a setting that we know is tricky — navigating an interaction with someone who’s different from you.”

The motive may be ingratiation, the paper suggests, since studies show that white liberals are “concerned about appearing racist,” as Dupree said. In their role as “impression managers,” white liberals may even take on the negative stereotypes they harbor toward people of other races, in an effort, as the paper puts it, to “get on their level.”

Their conservative counterparts, meanwhile, appear not to employ these stereotypes in the same way, as Dupree said, because, “we know empirically that white conservatives are less likely to be interested in getting along with racial minorities.” This became starkly evident to the behavioral psychologist when she turned to political campaign speeches for the first of several studies conducted to test whether political ideology shaped how white people presented themselves, on scales of competence and warmth, depending on the race of their audience.

In tracking the word choices made by white Republican and Democratic presidential candidates before white and black voters, her sample size was limited primarily by “the number of speeches in which Republican presidential candidates showed up for black audiences,” she said. The race of the audience was approximated by setting, at a black church for example, and by occasion, say the 40th anniversary of the Rev. Martin Luther King Jr.’s death.

What she found, by performing online text analysis of 74 campaign speeches over the past 25 years, was that white candidates who were Democrats used significantly fewer words about “agency or power” and more about “affiliation and communality” when addressing minority voters. There was no significant difference exhibited by Republican candidates.

The irony, as the paper notes, is that “Whites who may be more affiliative toward Blacks alter their verbal responses toward them in a way that matches negative stereotypes. Despite the patronizing behavior that they enact, these liberal candidates may hold more goodwill toward minorities.”

Additional experiments fleshed out the effect. In these studies, participants were either undergraduate students recruited through Princeton’s Survey Research Center or users identified through Amazon’s Mechanical Turk service, or MTurk, a relatively low-cost source of subjects for experiments and other tasks. Political ideology was mostly measured by authoritarian values and belief in hierarchy, which researchers commonly use to assess political conservatism.

In the hypothetical situation of the book club, participants viewed a list of 24 pretested words and were asked to choose 12 to use in their email to the group’s secretary. Liberal participants included words “that would make them appear significantly less competent with a Black interaction partner than with a White one,” the study found, while conservative participants presented themselves as equally competent with a black and white partner. A related scenario, in which participants chose personality traits for themselves in an introductory email, provided the weakest evidence of a competence downgrade, which the authors reasoned could have been a product of a “less competence-focused task.”

Finally, in an interaction that participants presumed to be real, they viewed the first name of an online partner, and an avatar that the partner had apparently chosen to present, and then filled out a profile for themselves based on a set of available traits, such as honest, capable, ambitious and helpful. Once their profile was complete, they indicated how they hoped they would appear by ranking several descriptions, including kind, intelligent, fair and friendly.

Unlike in previous experiments, liberal whites indicated that their goal was to appear less competent with a black partner than with a white one. Conservatives betrayed no such goal. The final experiment also stood apart because it was presumed by participants to be real, whereas the others were clearly hypothetical. It also presented them with visual evidence of their partner’s race, while the others relied only on names.

Across the studies, the paper concludes, “liberals’ competence downshift is a subtle, but consistent effect.” At the same time, it acknowledges the difficulties of tracking subtle shifts in self-presentation, and calls for additional research that goes beyond online interactions. It says that future studies should strive for larger sample sizes.

Further examination is also required, Dupree said, to determine whether people make themselves appear less competent with any group whose approval they’re trying to earn. It might be worthwhile, for instance, to look at Asian Americans, she said, because they are less likely to be stereotyped as incompetent.

The studies controlled for the gender of speakers but not of their audience, Dupree said, meaning that there could be an additional difference, for example, if someone were speaking to a black man or a black woman.

The data doesn’t point to conclusions about whether the competence downshift is effective in smoothing fraught interactions. As the paper observes, the behavioral difference is subtle, and Dupree said it’s “possible that racial minorities don’t necessarily pick up on the shift.”

At the same time, she said, research shows that racial minorities are more concerned about being respected than about being liked. “They may be attuned to the possibility of being patronized,” she speculated.

Dupree said she was driven to conduct the research by a gap that she had identified while in graduate school in work on prejudice, which barely addressed people less likely to be biased against minorities. So, too, was she driven to explore topics that were “personally meaningful” to her.

“I will say that this was a topic I was and still am very much invested in,” she said. “While the result may seem counterintuitive to some, it may not be to others. I fully understand both reactions.”