President Donald Trump’s behavior on Twitter routinely drives entire news cycles. This weekend, he showed that a single word within a single presidential tweet can be explosive.
Trump raised alarm bells in his published response to the news that his former national security adviser, Michael Flynn, pleaded guilty to lying to the FBI.
The tweet published to Trump’s account clearly implied that he already knew that Flynn had deceived the Feds when he fired him back in February: “I had to fire General Flynn because he lied to the Vice President and the FBI. He has pled guilty to those lies. It is a shame because his actions during the transition were lawful. There was nothing to hide!”
That unleashed a frenzy of speculation about whether Trump had just admitted to obstructing justice, since it seems he must have known that Flynn had committed a felony when he was pressuring then-FBI director James Comey to ease up on the Flynn case.
But then came word that maybe Trump didn’t write the tweet after all. The Washington Post reported that “Trump’s lawyer John Dowd drafted the president’s tweet, according to two people familiar with the Twitter message.” The Associated Press also identified Dowd as the one who “crafted” the tweet, citing “one person familiar with the situation,” though Dowd himself declined to make a comment to the AP.
Attributing the tweet to Dowd set off a new round of incredulous chatter. Would the president’s lawyer really compose a tweet like that on his client’s behalf, especially one that seemed so incriminating? One widely shared response, from a person who tweets from the account @nycsouthpaw, focused on a single word in the tweet as grounds for skepticism: “We’re supposed to believe John Dowd wrote pled instead of pleaded?”
Others argued that Dowd could very well have used pled as the past tense of plead. Harvard Law School professor Jonathan Zittrain noted, “I’ve seen lawyers write each. It’s not like, you know, hung and hanged.” Indeed, both pleaded and pled are considered acceptable by American usage guides—though, in many newsrooms, pled is considered a rookie mistake, which helps explain why some journalists seized on it.
Pled actually dates back to the 16th century, and though it never gained much traction in British English, it has been gaining in popularity in American English over the past few decades. Some prefer pled because they think pleaded sounds wrong, based on analogous past-tense forms like bleed/bled and feed/fed. Plenty of legal types don’t seem to mind pled, at least not in the United States. In fact, when the blog Above the Law polled its readers in 2011, 57 percent of the 1,311 respondents preferred pled to pleaded.
But what of Dowd himself?
I searched through the LexisNexis news database to try to find his preference for forming the past tense of plead, and I discovered an example from January 2010, when Dowd was representing the billionaire hedge-fund manager Raj Rajaratnam, who was standing trial for insider trading. As quoted in The Wall Street Journal, Dowd said of Rajaratnam, “He’s pled not guilty and we intend to try his case and demonstrate that he’s innocent.” (Rajaratnam was later found guilty and is currently serving an 11-year prison sentence.)
So Dowd, too, is on record as a pled user. That single word does not betray some nonlawyerly voice—Trump’s or anyone else’s—so we can’t point to it as evidence for who really wrote that tweet. It would be a tidy solution to isolate the use of pled as a kind of “tell” disproving the attribution of the tweet to Dowd, but it is in fact exceedingly difficult to be able to identify such a linguistic smoking gun.
One skeptic on Twitter wrote, “A forensic linguist could rule out Dowd in 5 minutes. Once that happens, Trump has no backpedal.” Actual forensic linguists would be hard-pressed to rule Dowd in or out on the basis of a single tweet, however. The field of authorship analysis requires significant amounts of textual data in order to be reliable. First, one would need to compile past texts firmly attributed to the potential authors—in this case, Trump and Dowd. That could at least establish idiosyncratic patterns of style and usage, but for a low-frequency word like pled, even that approach may prove fruitless. (For what it’s worth, Trump had never previously used pled in a tweet, according to the Trump Twitter Archive. Trump’s only use of pleaded is from a news article he quoted.)
Authorship analysis has had some notable success stories, but not involving something as slender as a tweet. In 2013, I wrote in The Wall Street Journal about how forensic analysts helped determine that Harry Potter author J.K. Rowling had written a crime novel, The Cuckoo’s Calling, under the pen name Robert Galbraith. I asked one of the experts, Patrick Juola of Duquesne University, to detail his approach in a guest post for Language Log, a blog about linguistics that I contribute to. When a commenter remarked that it would be interesting to analyze the anonymous Twitter tip that had set off the investigation, Juola replied, “It would indeed be interesting, but authorial analysis of tweets is HARD. Not enough data, you see.”
That sort of challenge has been taken up by some forensic linguists, such as Tim Grant of Aston University, who has been working on techniques to analyze tweets and other short-form messages. But such an analysis would be even trickier in this case, since Dowd—if he is indeed the true author of the controversial tweet—may have been attempting to mimic the Twitter voice of Trump. That could explain the exclamation point at the end, for instance—a classic Trumpian touch. Or Dowd could have “drafted” the tweet with Trump subsequently making revisions or at least adding some finishing touches. Then we’d be dealing with an even murkier co-authorship situation. It looks like we’ll simply have to wait for further illumination of the story behind the tweet’s composition: No single word or punctuation mark is going to give the game away.
We want to hear what you think about this article. Submit a letter to the editor or write to firstname.lastname@example.org.