Abracadabra! I Win!

(image from Slate.com)

I tend to avoid Slate.com these days because, frankly, I typically find myself scoffing at some idiot article they've published that promotes such a ridiculous mis-reading of academic research that it's hardly worth finishing... like this one from today: A Better Way to Fight With Your Husband which linked to this article: The Healthiest Way To Fight With Your Husband. It's a classic piece of idiot journalism worthy of a Full Liberman* if only it weren't so trivial and obvious as to be beneath the man, so I'll take a crack at it.

The big point is that fabulous new research from real life scholars (psychologists nonetheless, and they're almost like scientists) proves that women should use particular words when yelling at their husbands (the experiment used heterosexual married couples). Pretty awesome, ain't it! Just use the right words, and like a magic key you can unlock the mysteries of the brain and make it do what you please (okay, I'm starting to exaggerate, but less than you might think).

First let's look at the way the academic article is summarized in the puff piece that Slate linked to:

A new study of married couples, however, has found physiological evidence for one technique to diffuse tension: choosing the right fighting words. Couples who used analytical language, such as “think,” “understand,” “because,” or “reason,” during heated arguments were able to keep important stress-related chemicals in check, according to research published in the latest issue of the journal Health Psychology. Cytokines are inflammatory chemicals that spike during periods of prolonged tension and can lower your immunity and lead to early frailty, Type 2 diabetes, arthritis, and some cancers. The authors noted a curious gender twist in their results. Husbands benefitted from their wives’ measured language, but a man’s carefully chosen words had little effect on a woman’s cytokine balance.

To be fair, here is a passage from the authors' abstract of the original article:

Effects of word use were not mediated by ruminative thoughts after conflict. Although both men and women benefited from their own cognitive engagement, only husbands' IL-6 patterns were affected by spouses' engagement. Conclusion: In accord with research demonstrating the value of cognitive processing in emotional disclosure, this research suggests that productive communication patterns may help mitigate the adverse effects of relationship conflict on inflammatory dysregulation.

And here is a passage from this interview with the first author, Jennifer Graham, Penn State assistant professor of biobehavioral health:

"We specifically looked at words that are linked with cognitive processing in other research and which have been predictive of health in studies where people express emotion about stressful events," explained Graham. "These are words like 'think,' 'because,' 'reason' (and) 'why' that suggest people are either making sense of the conflict or at least thinking about it in a deep way."

For the study, the 42 couples made two separate overnight visits over two weeks.

"We found that, controlling for depressed mood, individuals who showed more evidence of cognitive discussion during their fights showed smaller increases in both Il-6 and TNF-alpha cytokines over a 24-hour period," said Graham, whose findings appear in the current issue of Health Psychology.

During their first visit, couples had a neutral, fairly supportive discussion with their spouse. But during the second visit, couples focused on the topic of greatest contention between them.

"An interviewer figured out ahead of time what made the man and woman most upset in terms of their relationship, and we gave each person a turn to talk about that issue," said Graham.

Researchers measured the levels of cytokines before and after the two visits and used linguistic software to determine the percentage of certain types of words from a transcript of the conversation. (my italics)

The researchers' results suggest that people who used more cognitive words during the fight showed a smaller increase in the Il-6 and TNF-alpha. Cognitive words used during the neutral discussion had no effect on the cytokines.

When they averaged the couples' cognitive words during the fight, they found a low average translated into a steeper increase in the husbands' Il-6 over time. There were no effects on the TNF-alpha. However, neither couple's nor spouse's cognitive word use predicted changes in wives' Il-6, or TNF-alpha levels for either wives or husbands.

Graham speculates that women may be more adept at communication and perhaps their cognitive word use had a bigger impact on their husbands. Wives also were more likely than husbands to use cognitive words.

Well, thank gawd they used fancy computers to count cognitive words! After reading these three descriptions, it was clear to me that the original work is likely flawed. I don't have access to the original study, unfortunately, but taken together, the abstract and first author's interview suggests to me that it makes the same mistake most non-linguists make: they assume the linguistics part is easy and don't put enough effort into it. Dr. Graham's initial claim in the interview jumps out at me: "We specifically looked at words that are linked with cognitive processing in other research..."

Hmm? Words that are "linked with cognitive processing?" What does this mean? I would love to see the references page to follow-up on this "other research." Graham later refers to these as "cognitive words." They are alternately referred to as analytical language, measured language, conflict-resolution words, and cerebral words. From the puff piece and the interview we have five examples:

1. because
2. reason
3. why
4. think
5. understand

Huh? One conjunction, one interrogative, and three verbs of cognition. Hmmm. Is there any intuitive reason to believe that "because" is "linked with cognitive processing" in some special way that other words are not? Is it the fact that it grammatically links clauses? Many words do this. Are the verbs on the list simply because they are verbs of cognition? Are run and jump less "linked with cognition" because they are verbs of motion? I would have to speculate on what this "other research" discovered about the magical properties of the special words that make them the key to brain chemicals. Abracadabra! Poof! Also, it's not at all clear to me why they averaged the couples' frequency count. What is this average supposed to tell us?

However, the puff piece makes the leap into idiotsville all by itself:

"The study is significant because it’s one of the first to link language with biological markers and show what kinds of words help sparring couples rather than just recommending they “communicate more,” explains James Pennebaker, chair of the department of psychology at the University of Texas-Austin, who has studied the role of language on relationships." (my italics).

Nope. No link. Just a transcript. Given the study's methodology of counting words in a transcript, at no point could they possibly have been able to show any causal relationship between a particular word's utterance and the levels of a particular chemical in a person's brain.

The puff piece authors pull the classic journalist's trick of "being fair" by adding actual linguist Deborah Tannen's skepticism of the "link" between particular words and particular chemicals, but they abandon all skepticism just a few sentences later and end with a bang! "”Even when it seems like he is ignoring you, your words may be having an effect—at least on a chemical level,” says Graham"

Sigh.

*I'm going to start using the term "The Full Liberman" to refer to Mark Liberman's excellent manner of debunking bad journalism (see here and here for examples).

UPDATE (11/28/09): A nice summary of Full Liberman's at LL here.