The coded language that holds women back at work

Share on Linkedin
Woman listening while male colleague talks
Workplace language encodes gendered expectations that feed into who we hire, how we assess people and who we promote, holding some people back.

Erin Oldford, an assistant professor of finance at Canada’s Memorial University of Newfoundland, was having a hard time getting women to join the student-managed investment fund she advised. Female students made up only 10% of applicants for what should have been an attractive opportunity to develop tools important for a career in finance. 

On a trip to Toronto in February 2020, Oldford spoke to another woman in the finance industry, the managing director of an investment company. She’d said she’d been having a similar struggle with recruitment. That conversation, coupled with Oldford’s research, made her realise that the language of the job postings might be deterring qualified women from applying. 

She decided to overhaul the language of the fund’s website, job application portal and marketing materials, and launch a social media campaign to showcase female role models. The fund is about to start its first round of recruitment since these changes were made, so it’s too early to determine their impact. But Oldford, who hopes to eventually reach a rate of 50% of applications coming from female students, says it’s clear to her that language we use “is an important impediment to progress”. 

It may be hard to believe that simple changes – to verbs, pronouns, adjectives or even phraseology – could lead to substantial changes in workplace gender balance. But plenty of social science research suggests that the language we use at work does indeed encode gendered expectations in ways that may not be obvious to employers or employees. These gendered expectations feed into who we hire, how we assess people and who we promote – and may play a role in holding some people back.  

The agentic-communal spectrum 

Gender researchers often discuss the differences between agentic and communal language and behaviours. To be agentic is to be confident and decisive; to be communal is to be warm and helpful. In many countries, including seemingly progressive ones, men are still perceived as more agentic, while women continue to be viewed as more communal. 

This feeds into persistent ideas about women, particularly women of colour, being less suited for leadership. Much of the language around leadership (and certain types of jobs) inclines towards agency – sending, intended or not, a gender-coded message about who the ideal candidate is.

Gendered language feeds into all aspects of the workplace, from hiring to performance evaluations and promotions (Credit: Getty)

Gendered language feeds into all aspects of the workplace, from hiring to performance evaluations and promotions (Credit: Getty)

For example, Oldford and her colleague John Fiset, of Canada’s Saint Mary’s University, have found that male-coded language (think ‘dominant’ and ‘competitive’) is rife in finance internship postings, which are an important part of building finance careers. These send subtle signals of welcome to stereotypically masculine applicants. In contrast, women are more comfortable applying for jobs advertised using terms like ‘interpersonal’ and ‘understand’, even for the same positions and sectors. 

Oldford provides some examples from actual job posts. This is more communal:

“We’ll support you with the tools and resources you need to reach new milestones, as you help our customers reach theirs.”

This is more agentic:

“Tell us your story. Don’t go unnoticed. Explain why you’re a winning candidate.”

In their study, Oldford and Fiset found that women felt that they were better fits, and were more interested in applying for, the jobs whose posts were high in communal language and low in agentic language. The results suggest that tweaking the language of job postings may help to improve the gender diversity of job applicants, particularly in male-dominated fields like finance. 

Similar studies have shown that the same phenomenon exists in other fields as well. One multi-field study inspired behavioural designer Kat Matfield to create an online tool called Gender Decoder. Based on a collection of male-coded and female-coded words, this tool allows users to paste in the text of a job advert, to quickly determine the possible presence of subtle gender bias. 

Even if there’s a gender-diverse pool of applicants, recruiters may still use gender-coded words and views in hiring decisions. For someone of any gender to be hired, they generally need to demonstrate competence (e.g. ‘efficient’). Yet when the candidate is a woman, hiring decisions are more likely to be factor in their sociability (‘friendly’) and morality (‘trustworthy’) as well, compared to male candidates. 

Essentially, it’s evident in hiring committees’ language that women not only have to prove their ability; they have to clear the peculiar extra hurdle of demonstrating that they’re good people, too. 

Gender bias in performance reviews 

The hurdles don’t stop once a woman is hired. Research on performance evaluations suggests that regardless of the gender of the evaluator, different words are used to describe how men and women perform. 

A study of military performance evaluations showed that more negative characteristics were recorded for women overall, possibly because stereotypes of women didn’t match the conventional mode of military leadership. ‘Compassionate’ was a common descriptor of women, whereas ‘competent’ was frequently used for men. In terms of negative feedback, ‘irresponsible’ was more often applied to men, whereas judgemental and gender-coded terms like ‘frivolous’ and ‘temperamental’ were more likely to be mentioned for women.

Women were praised more for being helpful, but being helpful wasn’t highly valued

Similar disparities exist in different industries. One recent study by research associate Alison T Wynn and colleagues at the Stanford VMware Women’s Leadership Innovation Lab analysed the full text of performance evaluations for an internet services company in Silicon Valley. The researcher determined that 61% of the time, when communal terms were used, they were being applied to women

These kinds of communal descriptors, though apparently positive, did not translate into higher ratings. For instance, women were praised more for being helpful, but being helpful wasn’t highly valued. Instead, ‘taking charge’ was viewed favourably. Thus, women’s chances of higher pay and positions were being affected by gendered attitudes, expressed in specific words. 

Even where certain words were being applied to women and men at the same rate, they were being valued differently, according to gender. Wynn explains that women “were not any more or less likely to be described as taking charge. But when men were doing it, they were getting rewarded, whereas women… it wasn’t getting them into that top category.” Women were receiving comments suggesting that they were too assertive and taking charge too much. In contrast, men were being criticised for being ‘soft spoken’. 

As well, men were more likely to be described in exceptional terms; ‘visionary’, for instance, was applied to men 61% of the times it was used. ‘Genius’ was also mostly reserved for men. This matches research suggesting that men are associated with being ‘brilliant’, across cultures and ages. 

When it comes to race and other aspects of identity, including sexuality and non-binary gender identity, there’s insufficient research when it comes to linguistic bias at work. On the short-term job site Fiverr, black women and men are more likely than white workers to be described using negative adjectives; this may be a reflection of social bias on the part of the reviewers. 

Alex Kapitan, an editor and consultant who blogs as Radical Copyeditor, says that the “way we think about people dictates the way we treat people”. Kapitan refers to the “mutually reinforcing cycle” of gender-coded language. “When we use particular language for a particular group of people, and that language is different than the way that we talk about other groups of people, that reinforces the sorts of biases that are baked into that language.” 

Fairer and more constructive language 

While it’s very challenging to overturn ingrained stereotypes related to gender, there are plenty of ways to be more thoughtful about the language that transmits those stereotypes. 

When it comes to performance evaluations, Wynn recommends, “the more specific you can get, the better”. For one thing, performance feedback for women tends to be more vague (and thus less actionable) than for men. Precision also helps to avoid the trap of relying on stereotypes as a substitute for grounded, specific comments. 

Related to this, Wynn advises companies to avoid descriptors of personality and communication style, as these are areas where gender and racial bias creeps in. If it’s essential to include some metric related to communication style, it should be results-focused.

More clarity, consistency and transparency about how criteria translate into reward would be fairer for everyone

This involves reducing manager discretion – “trying to put procedures in place that make it harder for the manager to act on whatever biases they may have”, says Wynn. “It doesn’t mean these managers are bad people,” she stresses, but more clarity, consistency and transparency about how criteria translate into reward would be fairer for everyone. This would involve “trying to have formal procedures in place, nice clear criteria, and accountability structures that make it harder for them to let bias impact their decision making”. 

One concern is some companies attempting to overcompensate for the presence of bias by getting rid of formal performance reviews altogether. But this is likely to worsen the problem. Wynn worries that “then you don’t really have clear criteria or processes for evaluating. And so, it can be possibly more prone to bias, depending on how it’s set up”. 

Similarly, if managers try to overcorrect by avoiding any criticism of female employees, they’ll be doing a disservice to those employees, who need the feedback to progress. To avoid this problem, Wynn suggests, “tie everything to business impact”. 

“I think the onus is not on the women to change, it’s really on the companies,” notes Wynn. Ultimately, it’s up to organisations to ensure with their criteria that they’re “actually weighting these things appropriately and valuing them equally, regardless of whether the behaviour is shown by a man or a woman”.