Stanford researchers analyzed the language in 125 performance reviews from a tech company and found something disturbing

It's no secret that women's prospects for professional advancement are often dimmer than men's.

Across a range of industries, they're less likely to get promoted, and as a result, they're underrepresented in senior leadership.

New research, spotlighted in The Wall Street Journal, yields some insight into the factors that contribute to women's stalled progress. A team of researchers at Stanford University's Clayman Institute for Gender Research is currently reviewing hundreds of performance reviews from four unnamed technology and professional-services companies.

Though the research is still underway, initial findings from an analysis of 125 performance reviews at a tech company suggest that managers use substantially different language to describe their male and female employees.

Here are some of the key findings, provided by the Clayman Institute:

• Women's evaluations contain nearly twice as much language about their communal or nurturing style (e.g. "helpful" or "dedicated.")
• Managers are nearly seven times more likely to tell their male employees that their communication style is too soft. Women, on the other hand, receive 2.5 times as much feedback related to their aggressive communication style.
• Men are more than three times more likely to hear feedback related to a general business outcome.
• Women's evaluations contain 2.39 times the amount of references to team accomplishments, as opposed to individual ones.
• Men hear nearly twice as many references to their technical expertise and their vision.

Interestingly, it didn't matter whether the manager administering the evaluation was male or female - biases still existed.

Caroline Simard, director of research at the Clayman Institute, told Business Insider that these differences in performance reviews could be highly problematic because "language is powerful at shaping perceptions."

For example, a man might get promoted over his female coworker because he was described in his evaluation as a "strong visionary," as opposed to a "dedicated team player."

Fortunately, Simard said, these findings are highly "actionable" in the sense that we can do something about them. She pointed out that the companies who participated in the analysis approached her research team, and not the other way around, because they wanted to find out how they might be unintentionally hurting women's chances of success.

In other words, at least some organizations are open to the idea that they might be biased and that they have some work to do.

One solution is to be clear and consistent about the criteria on which employees are evaluated. For example, does everyone get judged on how aggressively they communicate - or just women?

Another possibility is to have coworkers read each others' evaluations of subordinates to hold people accountable for what they write. "Knowing that other people will review [the evaluations] and ask questions can help bring everybody's level of accountability up," Simard said.

Ultimately, these often unconscious biases are hard to control or eliminate. But we can certainly figure out ways to prevent them from hurting other people.

"We can't stop it," Simard said of gender biases. "But if we diagnose how it plays out, we can come up with solutions."