My post on the gendered use of the word bossy has gotten a lot of attention in the past week. In it, I presented a modest bit of data to support Sheryl Sandberg’s campaign attempting to raise awareness of the obstacles that hinder young girls from developing leadership skills. Writing for The New Republic, Alice Robb used my own and other linguists’ data to argue that usage of the word bossy is gendered.
However, Sandberg’s campaign has had many detractors, and thus so did Robb’s article. As a result, the legitimacy of my own and others’ arguments about language has been questioned by Sandberg’s critics. For example, writing for Reason, Cathy Young questioned the entire campaign around bossy suggesting that it was designed to address “a fictional problem”.
Young dismisses the language data that I and others have presented supporting the gendered nature of the word itself, stating that “text searches can yield complex and contradictory results.” I would partially agree with that statement. Yes, text searches can lead to contradictory findings. In particular, they reveal the diversity of actual language usage. However, as I’ll demonstrate below, in this case, the contradiction is largely a product of Young’s use of unprincipled methods for studying language that make it appear that there is more contradiction in the data than there is. It’s also true that interpreting the results of textual analyses can be a complex endeavor, yet, in this case, the complexity of language use does not prevent us from making cogent claims about differential treatment of men and women.
First, allow me to explain why the results are not contradictory even though some writers like Young have tried to make it appear that my and others’ findings are more “selective” than they really are. Young contradicts our findings by picking several different phrases to search through Google’s search engine and stating that these show that bossy was used more for men and boys than for women and girls, which would contradict my and other linguists’ findings. I replicated Young’s results and present them in the table below.
|Table 1: Cathy Young’s analysis of bossy using Google|
|Male phrases||# of results||Female phrases||# of results|
|he is bossy||227000||she is bossy||151000|
|he is too bossy||127000||she is too bossy||117000|
|bossy little boy||187000||bossy little girl||93900|
|my brother is bossy||11700||my sister is bossy||20000|
|my son is bossy||808||my daughter is bossy||37400|
|Young’s total||553508||Young’s total||419300|
Young’s analysis would suggest that actually bossy is used slightly more frequently for men and boys than for women and girls, which is the opposite of what researchers, including myself, have reported. However, any good researcher would ask Young the following question: “Why have you restricted the analysis to these particular phrases?” In effect, Young’s analysis appears to be nothing more than an exercise in hunting down the data that would fit her preconceived conclusion. If we expand her data set to include similar but absent phrases we find a quite strikingly different result as in my table below that shows my expansion on Young’s Google search technique.
|Table 2: My further analysis of bossy using Google|
|Male phrases||# of results||Female phrases||# of results|
|bossy boy||7000||bossy girl||419000|
|bossy boys||6890||bossy girls||203000|
|bossy man||11000||bossy woman||53400|
|bossy men||8160||bossy women||42800|
|he’s bossy||36800||she’s bossy||31400|
|he was bossy||198000||she was bossy||454000|
|he’s being bossy||60700||she’s being bossy||93000|
|he is very bossy||414000||she is very bossy||102000|
|he is extremely bossy||5770||she is extremely bossy||31300|
|he is super bossy||1660||she is super bossy||6970|
|boys are bossy||3760||girls are bossy||53300|
|men are bossy||7680||women are bossy||107000|
|My total||761420||My total||1597170|
|Combined totals||1314928||Combined totals||2016470|
Expanding on Young’s analysis, I arrive at the finding that bossy is used for women and girls about one and a half times as frequently as it is for men and boys. At this point, it would be understandable if readers thought to themselves “well you could always add more and different phrases and perhaps arrive at a new conclusion, so this whole process of text analysis is fraught with problems”. That conclusion would be correct if all attempts at text analysis were as methodologically flawed as Young’s.
However, the techniques used by linguists such as myself to study this topic are quite different from what Young did. They are designed to address the problems related to researchers selecting the data that confirms their pre-conceived conclusion (or “cherry-picking” as we call it) by being as inclusive about what is included in the data as possible and not arbitrarily restricting it as Young has done.
The first technique is to rely on more flexible approaches to searching. For example, Alon Lischinsky (Lecturer in Media and Culture Studies at Oxford Brookes University) used Google books Ngram viewer to show that the ten nouns that occur most frequently after the word bossy include four nouns referring to women (woman, women, wife, mother) and no nouns referring solely to men, as you can see in the graph below (or by clicking here). This approach compares a seemingly infinite number of different phrases including bossy men, bossy women, bossy girls, and bossy refrigerators to arrive at the most frequent of them all and thus avoids the selectiveness of only arbitrarily selecting a few phrases taken from a seemingly infinite number of possibilities.
Another possible approach is to use a concept known as collocation. This approach asks the somewhat complex question of which words tend to occur around bossy (x number of spaces to the left or to the right) at a rate that is higher than these same words occur elsewhere? I employed this approach using the US sub-section of the Corpus of Global Web-based English. The words that collocate with bossy in this corpus are presented in the table below ranked according to a statistic that calculates the word’s tendency to occur in the vicinity of bossy (the mutual information score). They include the female-gendered words sister, she, and her but no male-gendered words. This approach captures and aggregates across an even greater number of possible phrases than the previous approach because it allows for such diversity in phrases as these two, which occurred in the data:
- … older sister can be quite bossy. She’s the middle child. I’m…
- …the younger sister with a bossy older sister who I adore…
The diversity of phrases captured through collocation then is much greater than in the approach used by Young.
|Table 3: Collocates of bossy|
Finally, the most flexible approach is one that is much more labor intensive. It involves gathering a random sample of instances of bossy and then simply reading through all of them with our own eyes to determine who is being labelled bossy. This is the approach I took in my recent blog post. Because of the amount of time involved, I looked at far fewer examples than any of the approaches I’ve discussed, but I also was able to classify instances that the above approaches would have missed. The graph below illustrates what I found, namely that bossy was applied to women and girls three times more frequently than it was to men and boys.
You might think to yourself, “But there’s only 101 examples! That’s so few!” It’s true that this is far fewer than normally done in automated textual analyses. However, consider this. If we assume that other linguists and I are mistaken and that the word is not used in a gendered fashion, then what is the likelihood that I would have found the large difference that I did in a random sample of bossy occurrences? This is easily calculable, and it happens to be less than .01% (for the statisticians in the crowd, chi-square = 24,02, p<.0001). That means that if we assume, for the purpose of argument, that Cathy Young is correct and the word is not used in a gendered fashion, then the data I found in the Corpus of Contemporary American English has to be for whatever reason extremely uncharacteristic of the way the word is generally used. Given that I am working with a random sample from a well-constructed corpus of language use, I’m personally going to place my confidence in the idea that Cathy Young is wrong.
All three of these approaches provide the flexibility necessary to search across different contexts of usage to arrive at a principled analysis, and they all point to the same conclusion: bossy is used more frequently to describe women and girls than men and boys. As a result, the contradictions in the data that Young alludes to, while curious in that they suggest phrases that buck the trend (and if anyone wanted to explore this further, I’d be happy to hear about it in the comments), are nonetheless a relic of her own flawed attempts at textual research and do not generalize to the broader picture of the way bossy is used. When we avoid cherry-picking the data, the findings are clear, and they support Sandberg’s and others’ contentions that bossy is gendered.
Another issue Young and others have raised is that how the word is used is more complex than can be revealed through textual analysis. That textual analysis is complex is made evident by the numerous questions in the comments of my previous post on this topic about whether I looked at other aspects of usage for the term (I regret to say the answer to all of your questions is “No I didn’t look at that, interesting as it is, but I’d be happy to look at your findings on the subject”).
Young’s claim about the complexity of this enterprise, however, exaggerates the degree to which this complexity prevents us from drawing well-supported claims about the gendered nature of bossy. In particular, she claims that two of the examples that I provide in my blog post are “positive” or “neutral” and thus cannot possibly point to a problem for women and girls. For example, I found this use of bossy in my data: “It seems like we [men] hate it when you’re [women are] bossy, but really, we only hate it when you boss us around.” The writer goes on to describe women’s bossiness as “totally hot” when directed at others. There is complexity here, namely what to do about an occurrence of bossy that is framed by the original writer in a partially complimentary fashion. In other words, women’s bossiness when directed at others is apparently “totally hot”, which could be a desirable thing. However, this confuses the issue. The claim that linguists such as myself are making is that the word is gendered, in other words that it is applied disproportionately to one gender (in this case, women and girls), not that it is universally used in a fashion that is intended by the speaker or writer (at least consciously) to be disparaging.
However, this raises the question of whether this comment ostensibly intended as a compliment is still itself part of a larger problem. To take another common example, we might find that women’s physical appearance is more likely to be commented on than men’s. Although these comments may very frequently be intended as positive or complimentary, they still point to different societal treatment for men and women. Specifically, these compliments, as good as they make some people feel, may reflect different expectations about how women should dress or look or problematic prioritization of women’s physical appearance over their other traits. Likewise, if people comment on women’s “bossiness” more than men’s, even if it is sometimes done in an apparently positive fashion (for example, when Tina Fey defiantly uses the term Bossypants as the title of her book), it still suggests the presence of different treatment for different groups of people.
Since feminism aims to bring about equality in the political, economic, and societal treatment of men and women, from the point of view of feminists such as myself, the gendered nature of the term’s use in society is an indication of a problem. Of course, the frequency counts and statistics I’ve presented here say nothing about the moral goodness or political expediency of a campaign against bossy or any other gendered language; they only show the observed tendency of the term to be used more frequently for women and girls than for men and boys.
Whether or not we choose to think that the gendered usage of bossy is a problem and the amount of attention we feel it deserves are two questions that cannot be directly addressed by this data or really any empirical observation. For this, we have to turn to our political or moral beliefs. In this case, I interpret this gendered use of language as a problem drawing on the perspectives of feminism and anti-sexism. Specifically, because I believe in equal treatment for men and women (as advocated for by feminism), I interpret the different treatment revealed in the gendered usage of bossy as a problem. You could, of course, approach the data making sexist assumptions about the world, such as the idea that women are inherently more bossy than men or the belief that women should have less of a right to be bossy than men. There are countless issues with such assumptions, not the least of which is that they are sexist. There are also empirical problems as well. For example, how is it possible that women could be both more likely to “be bossy” and yet more likely to have less economic and political power than men? In addition, what evidence is there to support this idea other than people’s anecdotes? Furthermore, if for the sake of argument we were going to take this suggestion seriously, how would we define bossy in an objective manner that gets away from its highly subjective but ubiquitous usage which boils down to something like this: “a characteristic of someone exerting authority that we resent”?
Defining the nature, scope, and urgency of the problem of gendered language as well as what should be done about it are of course other issues. It has been repeatedly pointed out that “banning” the word (in the sense of a prescribed social avoidance of the word as opposed to governmental censorship, since no one is advocating that) alone is unlikely to be a good solution. Robin Lakoff (Professor of Linguistics at UC Berkeley), for example, notes that even if somehow we were successful at ridding ourselves of the word bossy (which is quite unlikely to begin with), there are numerous words like aggressive just waiting to fill the functional void left by bossy. Social policing of language alone is insufficient to bring about the types of changes that Sandberg, myself, and numerous others desire. Nonetheless, it is valuable to think about the language we use as it reflects the societal and political norms that govern our behavior and structure our society. The gendered usage of bossy thus is not itself the problem but rather a symptom of society’s differential treatment of women, a problem worth solving. Drawing attention to the word is merely one way of pointing to a problem that has proven to be difficult to wrap our heads around.
Update (9 April 2014)
In response to Lesley Jeffries’ comment below (pointing out that men tend to be overrepresented in corpora), I have gathered some data that demonstrates this point in a new post. This means that some approaches to looking at the gendered nature of bossy underestimate the tendency to which it is applied mainly to women and girls. Specifically, because men and boys are referenced more in corpora, we would expect a greater number of instances where bossy applied to men and boys. Since we find the reverse, this is further evidence that bossy is gendered.