Transcription conventions

From Lane Greene on bluesky:

This is an unusual transcript (of Harris/Walz by CNN), filled with "gonna", "wearin'", "I'll tell ya", "pulling outta the race". These things are ubiquitous in speech, but most transcribers would change these to "going to", "wearing", "I'll tell you", "out of". edition.cnn.com/2024/08/29/p…

[image or embed]

— Lane Greene (@lanegreene.bsky.social) Aug 30, 2024 at 7:06 AM

[Here's an image if the embedding doesn't work for you…]

The cited CNN page is a transcript of the whole interview, not subtitles for the video version.

And I agree with Lane that the page's transcription conventions are unusual — when I compare other CNN transcripts, I find the conventional conventions, where e.g. "going to" is spelled "going to" even if the pronunciation is closer to "gonna".

In 4,718 words of the Bash/Harris/Walz transcript on CNN's website, there are 16 instances of "gonna" and 0 of "going to". That would be a plausible choice for representing how people talk these days, just as "can't" is plausible in place of "can not" — but formal transcription conventions these days prescribe "going to".

And in contrast, there are 7 instances of "want to", and none of "wanna", even though "want to" is usually (and invariably, here) pronounced as "wanna", in the same way that "going to" is pronounced "gonna". Here's the first of the 7 examples — "But I wonder what you say to voters who do want to go back":

Your browser does not support the audio element.

And zeroing in a bit further, on the words "do want to (=wanna) go back", with "wanna" outlined in red:

Your browser does not support the audio element.

Whatever the reason for the choices, they're applied in the same way to everyone in the interview. 13 of the 16 instances of "gonna" are in the speech of Harris or Walz, while 3 are due to Dana Bash. All 7 instances of "want to" are attributed to Dana Bash, as are both instances of "outta".

There's one instance of "wearin'" (attributed to Walz), and none of "wearing".

Only one word (sequence) is variably transcribed: there's 1 instance of "tell ya" (due to Harris), and 5 instances of "tell you" (3 attributed to Harris, 1 to Walz, and 1 to Bash).

If you want to look into things further, there doesn't seem to be full audio on the CNN site, but YouTube presents the interview in three parts: Part 1, Part 2, Part 3.

The YouTube transcript for Part 1 is very weird and not at all the same as the CNN page — it starts this way:

The YouTube transcript for Part 2 is less weird, but the lines are mostly truncated:

And the YouTube page for Part 3 has no transcript at all.

For another recent transcription weirdness, see "Putin: 'pollutant'? 'pooch and'?"

Some older relevant posts:

"The history of gonna", 5/9/2010 "True Grit isn't true", 12/29/2010 "Ask Language Log: Writing 'gonna' or 'going to'", 6/25/2011