Woman, interrupted

In 2015 Jessica Bennett wrote an article for Time magazine about the problem of men interrupting women. ‘My friends’, she said, ‘have come up with terminology for it: Manterrupting. Manstanding. (Or talk-blocking if you want the gender-neutral version)’. ‘Manterrupting’, defined by Bennett as ‘the unnecessary interruption of a woman by a man’, joined ‘mansplaining’ and ‘manspreading’ in the lexicon of everyday sexism. And in case anyone doubted that we needed such a term, along came Donald Trump, who interrupted Hillary Clinton 35 times in one 90-minute presidential debate.

But while Trump’s boorishness is not in doubt, on its own it doesn’t prove there’s a larger pattern. Bennett’s article, whose title was ‘How not to be manterrupted in meetings’, belongs to a genre which I have criticised many times on this blog because of its tendency to invent problems so it can sell women solutions (like the app that removes ‘just’ and ‘sorry’ from their emails, and the courses that teach them to stop tilting their heads). Whenever you encounter a generalisation of this form (‘women over-use the word “just”‘; ‘men interrupt women constantly’) it’s always worth asking if it’s supported by reputable evidence. So, what does research say about men interrupting women? Like so many things about language, it’s complicated.

The complications begin with the basic definition of ‘interruption’. If person B begins speaking before person A has stopped, does that mean B is interrupting A? Some researchers would say yes; others would say ‘not necessarily’. What we usually mean when we say that ‘B interrupted A’ is that B infringed A’s speaking rights by taking the floor before A was ready to cede it. By that definition, most cases of simultaneous speech are not interruptions at all.

Simultaneous speech is a common by-product of the way turn-taking works. We don’t usually agree in advance that A will speak first, then B, then C. Rather, who speaks when is something we negotiate as we go. We monitor the unfolding interaction and figure out from various clues when a potential ‘turn transition point’ is approaching. At that point, if no one has been selected to speak next, anyone can bid for a turn. And people often make their move slightly before the current turn has finished, resulting in a brief period when two speakers overlap. As long as the second speaker has correctly predicted that the first is about to finish, this won’t be perceived as violating their rights.

To illustrate the difference, here are two examples (they’re from a transcript of a British TV election debate broadcast in 2015). In the first example, the moderator invites SNP leader Nicola Sturgeon to respond to a point made by Labour leader Ed Miliband. The brackets show where there’s a stretch of simultaneous speech:

MOD:              Nicola Sturgeon do you agree with what Ed Miliband [is saying]
STURGEON:                                                                                                       [ well  (.)  ] I

This is overlap, not interruption. Though Sturgeon starts to speak before the moderator has finished his question, it’s already clear he’s giving her the floor: she knows her turn is coming and just slightly misjudges the timing.

In the second example, Ed Miliband is speaking when Sturgeon comes in uninvited:

MILIBAND: and that’s not [ going   to  ]
STURGEON:                          [we need to] replace the Tories

This is an interruption: Miliband is in the middle of a sentence, and Sturgeon cuts him off before he’s had a chance to make his point.

As it happens, Nicola Sturgeon produced more interruptions than anyone else in this debate–and it was virtually always a man she interrupted. But her behaviour no more disproves the ‘manterruption’ thesis than Trump’s behaviour proves it. To assess the validity of the claim about gender difference, we need to look at studies which investigated it directly.

I’ll start with one of the earliest (first published in 1975 and still frequently cited), which was carried out on a California college campus by Don Zimmerman and Candace West. They collected 31 recordings of students talking informally: ten were conversations between two men, ten were between two women and eleven were between a man and a woman. Their analysis of the interruptions (which they distinguished from overlaps along the lines I’ve just explained) showed a very striking pattern. In same-sex conversations the interruptions were fairly evenly distributed between the two speakers, but in cross-sex conversations the male speaker was responsible for 96% of the interruptions. Zimmerman and West concluded that men ‘deny equal status to women as conversational partners’.

I often see this study cited in popular sources (like Bennett’s Time article) as definitive proof that men interrupt women more than vice versa. But clearly it isn’t definitive: if we’re going to make general claims we need more to back them up than a single study, done nearly 50 years ago, which looked at a specific population (US college students) engaged in a particular kind of talk (informal, peer-to-peer and one-to-one). The good news is that since 1975 a lot more studies have been done. The bad news, however, is that their findings have been far from uniform.

In the early 1990s Deborah James and Sandra Clarke reviewed the accumulated evidence, and concluded that there was no clear pattern. Some studies had found that men interrupted more, a smaller number had found that women interrupted more, and the majority had found no difference. These reviewers also pointed out, however, that comparing the various findings wasn’t easy: different researchers had defined interruption in different ways, and consequently they had counted different things.

One issue that may arise in this kind of research is whether to count cases which are formally interruptions (i.e., not just overlaps), but which don’t match the prototypical definition of interrupting as taking the floor from someone who isn’t ready to give it up. It may sound like an oxymoron, but there is such a thing as a supportive interruption–when one speaker breaks into another’s turn, not to make their own point but to display their engagement or agreement with the current speaker’s point. Here’s an example from a conversation among women friends:

A: she didn’t like Katy she didn’t ge[t on with Katy at all                   ]
B:                                                               [no she didn’t get on with Katy]

B’s interjection meets the formal criteria for interrupting (it starts too early to be an accidental overlap, and it’s too long to be classified as a minimal response like ‘yeah’ or ‘right’), but B isn’t trying to take the floor from A; rather she’s reinforcing A’s point, in this case by echoing A’s actual words. Then she stops speaking, and A goes on with her story. The whole conversation is like this: there’s so much talking at the same time, you wonder if it even makes sense to call what the speakers are doing ‘interrupting’.

In a 1982 article called ‘Who’s got the floor?’ Carole Edelsky asked the same question about some data she’d recorded at academic committee meetings. In theory a committee meeting is much more formal than a conversation among friends, but Edelsky noticed that the participants hadn’t observed the formalities consistently. Mostly they had followed the expected one-speaker-at-a-time pattern of turn-taking (Edelsky calls this a ‘singly developed floor’, or ‘F1’); but there were moments when that arrangement yielded to what she calls a ‘collaborative floor’, or ‘F2’. In F2 episodes it was difficult to say who ‘had the floor’: it seemed more like a free-for-all, with people chipping in frequently but briefly, and often speaking simultaneously. Whereas F1 talk was male-dominated, with men holding forth at length while women took fewer and shorter turns, the talk that occurred during F2 episodes was more equally distributed. Edelsky offers the following explanation:

F1s, characterized by monologues, single-party control and hierarchical interaction where turn takers stand out from non-turn takers and floors are won or lost, share features with other contexts in which women have learned they had best not assert themselves. F2s, however, are inherently more informal, cooperative ventures that provide both a cover of “anonymity” for assertive language use and a comfortable backdrop against which women can display a fuller range of language ability.

Later researchers (including, perhaps most famously, Deborah Tannen) would echo the suggestion that women feel more comfortable speaking when interaction is organised in a collaborative way. But where Edelsky links this preference to women’s subordinate social status (when there’s a contest for the floor they have ‘learned they had best not assert themselves’), Tannen sees it as a quasi-cultural difference: men relish competition, women prefer collaboration. Though politically they’re very different, these two accounts make similar predictions about gender and interruption: crudely, that men in ‘F1’ situations will produce more interruptions of the competitive, floor-grabbing kind than women, but in ‘F2’-type situations women will equal or outstrip men in the production of supportive interruptions.

What all this means, though, is that we can’t answer the question ‘is there a general problem of “manterruption?”–which is essentially about the first type of interruption, not the second–by simply counting all the interruptions. To ensure we’re comparing like with like, we also need some way of deciding what kind of interruption we’re dealing with.

But how do we decide, given that we have no access to the thoughts of the people involved? One answer is to use what we do have access to–the reaction of one speaker to another’s intervention. Some conversation analysts argue that you can only count something as an interruption if there’s evidence it was taken as an interruption by the person on the receiving end. And what they mean by ‘evidence’ is the kind of reaction which is known in the jargon as ‘doing being interrupted’–acting in a way which signals to others that you feel your speaking rights have been infringed. You can convey that message verbally (e.g., by saying ‘stop interrupting me!’ or ‘please let me finish’), paralinguistically (e.g. by sighing deeply, or raising your voice while continuing to speak), nonverbally (using gestures or facial expressions), or a combination of these possibilities.

The conversation analyst Marta Baffy looked at ‘doing being interrupted’ in her analysis of the Congressional hearings which investigated Russia’s role in the 2016 presidential election. She focused on the testimony of Attorney-General Jeff Sessions, which was of interest because one of the people who questioned him, Sen. Kamala Harris, was reprimanded by the Chair for interrupting him. This reprimand, along with the subsequent criticism of Harris’s ‘aggressive’ behaviour in the media, prompted accusations of sexism from her supporters, who pointed out that women, and especially women of color, are often described as ‘aggressive’ when the same behaviour from a man would pass without comment.

Was a sexist double standard in play here? Baffy investigated by comparing the exchanges between Harris and Sessions to Sessions’s exchanges with a male questioner, Sen. Angus King. King, it transpired, had interrupted Sessions around the same number of times as Harris. In both cases Baffy counted eleven instances of simultaneous speech, most of which (six in King’s case and seven in Harris’s) could be classified as interruptions. There was, in other words, little difference between the two senators’ actual behaviour; but there was a big difference in the way Sessions reacted. With Harris he ‘did being interrupted’ nine times; with King he did it only three times.

As Baffy points out, there’s no way we can be certain that this difference was the result of sexist bias. There are other possible explanations: for instance, King questioned Sessions earlier in the day than Harris, so perhaps he just got grumpier as the hours ticked by. But the sexism interpretation fits with other evidence: some studies have found that women who interrupt are judged more negatively than men.

In one study Katherine Hilton asked 5000 American English-speakers to listen to scripted audio clips containing simultaneous speech, and then say if they thought one of the speakers had interrupted. To test whether gender had an effect, she recorded the same scripts in two versions, with the role of the putative interrupter played by a man in one and a woman in the other. She found that male judges rated female interrupters as ruder, less friendly and less intelligent than men performing the same script.

If we put these two studies together, we might well conclude that men have a problem with women who interrupt. And though neither study investigated the manterruption pattern directly, their findings may be a clue to what’s behind it.

But wait, I hear you say, have we established that there is a manterruption pattern? You’re right: so far I’ve been emphasising that the evidence is mixed, and sometimes difficult to interpret. I think that’s a reasonable summary of the overall picture. But I also think there’s something to be learned from a kind of research I haven’t talked about yet: research dealing not with casual conversation (or laboratory simulations of it) but with institutional talk–for instance, business meetings, job interviews, academic seminars, political debates, legal proceedings and medical consultations. In these contexts the pattern is more consistent; it’s also very revealing.

In institutions there’s generally a hierarchy of status, and we’d expect that to be the strongest predictor of who will interrupt whom. Yet many studies of institutional talk have found that higher-ranking women are routinely interrupted by lower-ranking men. Women doctors get interrupted by male patients, women bosses by male subordinates, women teachers by male students and women judges in Australia’s High Court by the male advocates who make arguments before them.

What strikes me about this pattern, and about the attitudes uncovered by Katherine Hilton, is how well they fit with the patriarchal principle laid out by the philosopher Kate Manne–that men are entitled to take from women, whereas women are obligated to give to men. If we think of (non-supportive) interruptions as a form of ‘taking from’ (that is, taking the floor from someone else) Manne’s principle might explain why men apparently feel entitled to interrupt any woman, even one who by other measures outranks them, while judging women’s own interruptions illegitimate or hostile.

From this perspective, the reprimanding of Kamala Harris was an example not of sexism but of misogyny–the punishment of women who give too little and/or take too much. But Harris has lived to fight another day: this week it was announced that she will be Joe Biden’s running-mate–and if he wins, therefore, his vice-president. This wasn’t a foregone conclusion; though Biden was committed to picking a woman, many people expected him to choose someone more emollient. There had been rumours that his team regarded Harris as too ‘ambitious’ and ‘abrasive’. But in the event she was picked despite, or perhaps even because of, her reputation for being, as Donald Trump immediately put it, ‘nasty’ to men.

Of course, when the campaign gets going Harris may come under pressure to be ‘nicer’. If so, I hope she’ll resist it. ‘Be nice, be polite, be conciliatory, be gentle’–these injunctions to women have a long and depressing history. But history, like men, can be interrupted.