A couple of researchers at Stanford University took a look at scientific papers which had to be retracted and compared them with ones that were found to be sound or which were retracted for reasons other than academic fraud.
Academic fraud can cover everything from fudging observational data to just flat-out making stuff up. Scientists are human beings, and sometimes when they confront data that doesn't match their theories, some of them will do like many of us will do and adjust the facts to fit the theory instead of the other way around.
The pair examined some 20 million words in papers published between 1973 and 2013, and they found that both third-person pronoun use and linguistic obfuscation were higher in retracted papers. Both are often seen as techniques used by people when they are lying. Evidently we feel we can create some distance between us and the untruth if we don't use first-person pronouns, and piling on the verbiage can leave so many interpretations available that the fibber can say, "Whoops! That's not what I meant." Or it may be that when linguistic obfuscation -- which is probably more commonly referred to by initials signifying bovine alimentary end-product -- is laid on thick enough, smart scientists are afraid to question it lest they be thought of as dumb for not understanding what everyone else does.
One of the researchers said it might be possible to develop software that would scan papers for that kind of language and prompt journal editors to take a second look at ones that exceed even the usual dry academese word salad that sounds a whole lot like linguistic obfuscation to folks outside of the relevant discipline. Programmers would need to be careful, of course, to create algorithms that wouldn't show too many false positives. Those could harm reputations and bring suspicion to people who write with perfect honesty but dull style.
They would also need to keep the software away from all political speeches and statements. It's bad enough to know that almost every one of those rat bastards is lying; seeing it proven by science would probably lead to voter turnouts in the single digits.
No comments:
Post a Comment