Political sound and silence

« previous post | next post »

As part of an exercise/demonstration for a course, last night I ran Neville Ryant's second-best speech activity detector (SAD) on Barack Obama's Weekly Radio Addresses for 2010 (50 of them), and George W. Bush's Weekly Radio Addresses for 2008 (48 of them). The distributions of speech and silence durations, via R's kernel density estimation function, look like this:

Then I wondered what the 2D distributions would look like. So courtesy of R's two-dimensional kernel density estimation, here they are:

The durations of speech segments are on the horizontal axis, and the durations of the immediately following silence segments are on the vertical axis.

Or maybe they're interstellar gas clouds?

Anyhow there's clearly some structure there. And it's neat that it just took a few minutes of computer time and a few lines of R to create the pictures.

Update — contour plots make it a little clearer what's going on:

See also "Poetic sound and silence", 2/12/2016


  1. D.O. said,

    February 8, 2016 @ 10:31 pm

    Yeah, it's much more pleasant to recognize an interstellar cloud on a Rorschach test then your aunt Lidiya collapsed on a sidewalk after having too much to drink.

    But really, are Obama's relatively shorter speaking fragments due to his higher rate of speech (which would also be supported by shorter silences) or because he speaks in shorter bursts of syllables?
    Another question. Is the distinction between a speaking fragment and a silence a clear one or is it blurred? Because it seems possible that Obama's manner of speech might be with more clear ends of fragments followed by a very short pause where Bush might have continued to produce some sort of sound blending into another fragment making the break not recognizable to the software.

    It is also expected that the distribution of silences is bimodal — long pauses between sentences and short pauses within. Obama's distribution of silences is not technically bimodal, but clearly there are 2 typical lengths for both presidents. Now, as Mr. Obama's interstellar nebula shows, he rarely follows a longish speaking fragment with a longish pause, which to me suggests that he rarely ends a sentence on a long stretch of speech. That is, he prefers to end with a short fragment. I've never heard Obama's (or anybody else's) presidential radio address, but I didn't pick up this manner from his other speeches. Did anyone?

  2. David L said,

    February 9, 2016 @ 11:07 am

    If I interpret the plots as interstellar gas clouds, or perhaps blobs of hot material thrown off by a star, then I would say that Bush's gaseous emission could be the same as Obama's after it has moved away from the source, and cooled and dissipated somewhat.

    I leave it to others to explain how this relates to the speakers' diction patterns.

  3. Charles Antaki said,

    February 9, 2016 @ 4:48 pm

    On the issue of kinds of silences – I guess the difference that comes to mind is the internal-to-the-unit (phrase, clause or sentence) silence and the one between sentences (and larger units). An accomplished speaker can, presumably, deploy both kinds more effectively than can a more awkward one. But it's not obvious whether better deployment is associated with frequency, or with the speech/silence ratio.

RSS feed for comments on this post