256 — The readability of scientific texts is decreasing over time

Plavén-Sigray, Matheson, Schiffler & Thompson (10.7554/eLife.27725)

Read on 03 May 2018
#language  #readability  #paper  #reading  #linguistics  #Flesch-Kincaid  #fluency  #jargon 

I remember sitting with friends in fifth (sixth?) grade trying our hardest to write a paper that would max out the Microsoft Word Flesch-Kincaid Grade Level metric (which, if I remember correctly, only went up to 12 at the time). The writing was totally unintelligible and we were immensely proud. (Alli, if you’re reading this, I still have that file somewhere and it’s ugly)

The Flesch-Kincaid Grade Level (and, separately, Readability Index) is a metric that captures the complexity of writing, comparing the total number of words in a passage with the total number of sentences, and the total number of syllables with the total number of words. For example, this blog has a FKGL of 9.8 and a Reading Ease of 54.6.

Novels trend toward higher grade levels. Presidential speeches trend toward lower grade-levels, which makes sense, since the goal is for most Americans to understand what is being said. (The current stable-genius-president achieves a score of 4.6, versus Barack Obama’s 9.7 or Eisenhower’s 9.4 [source]. And that’s including the speeches written for him by Literates.)

Academic papers trend the highest of all: In fact, as this research found after polling hundreds of thousands of abstracts from scientific journal articles since 1881, readability scores (FK and otherwise) have been dropping dramatically for years. By the early 1900s, the Flesch Reading-Ease (FRE) scores rested around the mid 30s (about as readable as the harder-to-read Wikipedia articles). Today, though variance is much higher, they average closer to 10.

This is pretty striking because it means that jargon use has gone dramatically up and the accessibility of papers to the layperson has nearly vanished. (In my planned epilogue to this #365papers endeavor, I’ll speak a bit more to this point.)

I think it’s difficult to regress out which parts of this decrease in readability are due to more complex topics or simply longer words (“electrophysiology” has eight syllables and completely throws off a readability score; but ephys didn’t exist prior to midcentury), and which are due to an actual reduction in the accessibility of material. But either way, if we continue at this linear rate, soon papers will be so hard to read that not even the authors can understand them.

That is a joke.

By the way: The second-lowest recorded presidential FKGL score was Truman’s, at 5.9. Truman held office from ‘45–’53, and so even accounting for academic “readability” deflation, Trump is an idiot.

That is not a joke.