In a new project, Andrew Higgins (Urbana-Champaign), Jacob Levernier (Oregon), and I are studying obituaries as a perspective on values. The basic idea is to sift through what people say about the dead to identify patterns in values. As a first pass, I read through a month’s worth of obituaries from my local paper, the Eugene Register-Guard, and noted all of the agent-level descriptions and evaluations of the deceased. In other words, I ignored what was said about hospice workers (not that their work is unimportant — quite the contrary) and single actions of the deceased, recording only the words used to describe him or her as a whole person. One thing I immediately noticed was that there were systematic differences in how the men and the women were described. Unsurprisingly, a huge proportion of the men had fought in World War II, and thus were described as soldiers. The women — not so much. This suggested that there might be other, more subtle, systematic differences. Andrew took this initial batch of data and created semantic maps of the things said about the men and the women. Here’s the one for men:
These maps can be a little hard to read, but I think this one is pretty straightforward. The size of the words indicates the number of other words that co-occurred with them in a single obituary. So, for instance, men who were described as Ducks fans were described as quite a few other things, but men who were described as chefs were described as only a couple of other things. The thickness and brightness of the edge connecting a pair of terms indicates how many times those terms co-occurred. The color of the terms is less important in this context.
From this map, we can see what people found most worth celebrating publicly about the men who died in Eugene last month and were mourned in the local paper. We can see, as it were, the Eugene-male constellation of virtues. By contrast, here’s the map for women:
This map is, as it were, a guide to the Eugene-female constellation of virtues. It’s very similar to the male one, as you can see, but there are some differences. The most noticeable is the one I already mentioned: no soldiers among these women. The women seem almost as fanatical about the Ducks. They’re more humorous, less athletic, and of course not patriarchs.
Finally, here’s a map everybody, regardless of gender:
The size of the terms and the thickness of the edges have the same meaning in this map. The colors this time have a more interesting meaning. Terms that share their color grouped together in much the same way that items in a factor analysis group together: they tend to co-occur with each other and not with other terms. The pink and light-blue groups are probably too small to interpret, but the others seem to meaningful. For instance, the green group is centered on humor and agreeableness. The red group seems to be mostly a matter of political liberalism. And the dark-blue group seems to be about commitment to the local community.
As we continue to work on this material, I’ll be posting more maps and discussing them. Andrew and Jacob may also do some guest-blogging, if I can talk them into it….