Language Log

Audio photoshopping at NPR

Last Friday, NPR's On the Media re-broadcast a segment about how NPR news and public affairs programming is edited ("Pulling Back the Curtain", 5/25/2007).

Ever wonder what goes on behind the scenes here at OTM? (Hint: Not everybody speaks as cleanly as it might seem.) A few years ago, we invited reporter John Solomon backstage to see how the sausage is made.

(The transcript is not on their site yet, but a bit of internet search turned up the version of the piece that was broadcast on 11/14/2003.) From my perspective, the most interesting part was what Solomon left out.

Nearly all of the 13 minutes is spent worrying about the relatively inconsequential topic of what you might call "audio airbrushing" -- editing out verbal blemishes like filled pauses, false starts and so on. Solomon also mentions that Car Talk edits in fake laughter from the co-hosts. (He didn't mention that editors -- so I've been told -- don't just remove false starts and dead air, they also sometimes add add fake pauses and breath noises, as a form of verbal punctuation.)

All of this audio touch-up stuff is interesting, but hardly a matter for listener indignation (though I can imagine the flap if a politician's office were caught similarly touching up interviews or news-conference recordings!).

A few slightly more serious sorts of fakery are mentioned . One is studio dubbing of voice-overs on top of background sounds from the field:

Nothing is so identified with NPR News as its signature field pieces, and nothing is more of a cut and paste production collage. My first such story was a media workout day held for reporters by the National Football League's New York Jets at their practice facility. [JETS FOOTBALL PRACTICE AMBIENCE UP & UNDER] Before I left, the producer suggested I record ambient sound during the drills that could be mixed later into the piece underneath my narration. That confused me. As a listener, I had thought that the narration in field pieces was recorded by the reporter at the location. I guess I should have realized reporters wouldn't always be able to deliver their clear, concise, well-organized prose while also reporting the story.

Another is the creation of apparent real-time conversations that never took place. We only get a hint of this:

JOHN SOLOMON: It also may be worth examining whether there are normal production elements that may unnecessarily deceive listeners, for example this is how All Things Considered's Melissa Block introduced a report on President Bush's visit to Singapore last month.
MELISSA BLOCK: NPR's Vicki O'Hara is traveling with the president and joins us now from Singapore. Vicki, the center piece of this trip to Asia for the president was the annual...
JOHN SOLOMON: Kern and Dvorkin agreed that the use of "joins us" in an introduction strongly implies that the interview is live when in fact it may have been taped hours before.
JEFFREY DVORKIN: I think it's a morally slippery slope. I think it makes more sense to say "we spoke to Vicki O'Hara a few hours ago," - boom - and you take that. The idea of being live is a construct and almost a vanity of the electronic media now.

I can say from personal experience that such concocted conversations are not limited to pre-recorded field reports spliced in as if they were arriving in real time. At least sometimes -- and maybe often, I don't know -- pre-recorded "conversations" are cut-and-paste jobs, just as fake as a photoshopped picture of two people who never met.

I described (my memories of) one such experience in an earlier post ( "Imaginary debates and stereotypical roles", 5/3/2006):

I wound up participating, on the air, in a vivid debate with my friend and colleague Cecil Coker in which we appeared to disagree fairly sharply on a topic that in fact we mostly agreed in being uncertain about. And curiously, though the [Radio Personality] spent half a day interviewing a half a dozen of us at length, the interviews had all been individual.

Thinking back over the experience, we realized that the RP had approached us from opposite sides of the question, and then stitched together bits of our answers. The interview technique was roughly like this:\

RP: So, in short, we can say that it's now apparent that EITHER?
Me: Well, the answer isn't clear. To be fair, there's quite a bit of evidence pointing toward OR, such as X, Y and Z. At least some people think that way, though I don't find the arguments very convincing myself; and P and Q do seem to point towards EITHER.

[ . . . ]

RP: As I understand it, a lot of people have concluded that OR.
Cecil: Well, some people think so, but I'm not convinced. EITHER seems much more likely to me, because of P and Q.

and the broadcast "conversation" then consisted of a series of "exchanges" like this:

Me: There's quite a bit of evidence pointing towards OR, such as X, Y and Z.
Cecil: Some people think so, but I'm not convinced. EITHER seems much more likely to me, because of P and Q.

In that case, no real harm was done -- it's true that my views were distorted, but I doubt that anyone else cared or even noticed. Still, as NPR's Ombudsman Jeffrey Dvorkin said to John Solomon in 2003, in reference to the practice of splicing in pre-recorded segments from reporters in the field, this is "a morally slippery slope".

For centuries, print journalists have been asking leading questions and then presenting the answers in a completely different context; omitting crucial frames and qualifiers; juxtaposing quotes from different segments of an interview as if they had a rhetorical connection, thereby creating meanings that the speaker never intended; highlighting aypical asides from a long interview as if they were the subject's main reaction; and so on. And I'm not just talking about Fox News, but about the Washington Post and the New York Times.

It shouldn't come as a surprise that these techniques are also commonplace in the broadcast media. And the audio or video version can be even more misleading, because it seems to be a real person who is participating in a real conversation -- which never took place, or at least never took place in the form in which it's presented to listeners.

In news photography, analogous staging, cutting and pasting would be (I believe) a firing offense. But radio seems to be governed by the culture of the writer rather than the culture of the photographer -- for the obvious reason that considerable staging, cutting and pasting is usually required by the nature of the medium, in order to create a coherent program out of bits and pieces from various places and times.

In the case of print journalism, the better publications have well-defined standards, even if they're routinely violated in practice. There are similar words in (for example) the NPR News Code of Ethics:

8. NPR journalists make sure actualities, quotes or paraphrases of those we interview are accurate and are used in the proper context. An actuality from an interviewee or speaker should reflect accurately what that person was asked or was responding to. If we use tape or material from an earlier story, we clearly identify it as such. We tell listeners about the circumstances of an interview if that information is pertinent (such as the time the interview took place, the fact that an interviewee was speaking to us while on the fly, etc.). Whenever it's not clear how an interview was obtained, we should make it clear. The audience deserves more information, not less. The burden is on the NPR journalist to ensure that our use of such material is true to the meaning the interviewee or speaker intended.

(This applies specifically to NPR News -- I don't think it covers programs like On the Media, Marketplace, Day to Day, or All Things Considered. I haven't checked to see if these individual shows have similar statements of principle.)

I can't cite many examples of radio photoshopping that violate these standards, because I rarely have access to the raw materials from which the finished products were created. But standard audio production techniques -- the same ones involved in editing out stammers and flubs, or splicing pre-recorded reports into an apparently live feed -- make it as easy to do this sort of thing in radio as it is in print. And if you think that that radio journalists haven't given in to the temptation from time to time, then you have a higher opinion of human nature than I do.

[Update -- David Warner sends in a link to this fictional example from Frontline, where the resulting radio version would have been perfect, but a small problem of wardrobe consistency spoils the video.]

Posted by Mark Liberman at May 31, 2007 07:53 AM