AstroTRENDS: No so weaselly after all


Substitute damn every time you’re inclined to write very; your editor will delete it and the writing will be just as it should be. — Mark Twain

Damn right!

In my last post, I showed a plot of the number of abstract that contained weasel words, as tracked by AstroTRENDS:

chart-2I interpreted this trend as a steady change in the style and “audacity” of astronomy papers, and I believed that a possible cause was hedging. (See Writing without conviction? Hedging in science research articles.) Note that I was not making a statement about the quality of the research, but merely about an interesting trend I had not seen mentioned elsewhere.

Perhaps I should have used more weasel words in my post! I acted upon Ben Weiner’s suggestion: use a set of non-weasel words as a control to verify whether the trend was due to an increase in verbosity instead. I track this set of keywords (a mix of adverbs and adjectives that I deemed to be neutral):

Fast OR Slow OR Large OR Small OR Before OR After OR High OR Low OR Many OR Few OR More OR Less OR Inside OR Outside OR Recently OR Just

This is available as “Non-weasel keywords” in the AstroTRENDS drop downs.

And here’s the plot!

The two appear to be tracking each other pretty well. It seems to me to be a strong indication of the correctness of Ben’s guess that verbosity is the main driver here. However, simple keyword search may still not be telling the whole story (e.g. because certain keywords “saturate” as the abstracts get longer, appearing more than once), so a better approach could be to study a small sample of abstracts through the years.


Weasels (green) vs. non-weasels (yellow)
Weasels (green) vs. non-weasels (yellow)

Turns out that there’s a comprehensive ADS API, described on GitHub here, so with a bit of rejiggering I will be able to let AstroTRENDS do free-form queries (via Michael Kurtz.), and do a bit of abstract munging myself.