Inaugural Speed

« previous post | next post »

Yesterday, prompted by a note from Geoff Nunberg, I cited a passage from Heejin Lee and Jonathan Liebenau's essay "Time and the internet" (published in Hassan and Thomas, Eds., The New Media Theory Reader). Their idea seems to be that "speed is contagious", and so the increased speed of modern life — faster cars, planes, computers, and so on — makes people do everything else faster. Their evidence for this included allegedly faster modern tempos for a Beethoven symphony, and a claimed 50% increase between 1945 and 1995 in the speaking rate of Norwegian parliamentarians.

Some helpful commenters pointed to evidence that symphonic tempos have not, in fact, increased in any sort of reliable way over the past century, and that Beethoven's own metronome-markings are very fast by modern standards. Other helpful commenters located the source of the claims about Norwegian parliamentary speeches, which seems to have been based on an analysis of stenotypists' tapes. I thought I'd bring a little English-language data to the table by looking at the inaugural addresses of U.S. presidents from Truman in 1949 to Obama in 2009.

I used the simple technique of measuring the overall duration of each speech (from the mp3 file), asking the computer to count the overall number of words in it, and dividing the word count by the duration in minutes.

(This is basically a Breakfast Experiment™, but I had to do it over lunch-time because I had a physical therapy session this morning.)

President Date Words Seconds Words/Minute
Truman 1949 2255 1173 115.3
Eisenhower 1 1953 2445 1231 119.2
Kennedy 1961 1341 834 96.5
Johnson 1965 1479 669 132.6
Nixon 1 1969 2104 1042 121.2
Nixon 2 1973 1785 967 110.8
Carter 1977 1222 893 82.1
Reagan 1 1981 2421 1197 121.4
Reagan 2 1985 2561 1237 124.2
GHW Bush 1989 2308 1221 113.4
Clinton 1 1993 1598 840 114.1
Clinton 2 1997 2155 1626 97.3
GW Bush 1 2001 1598 865 110.8
GW Bush 2 2005 2083 1266 98.7
Obama 2009 2368 1104 128.7

Here's the result in graphical form. If politicians were really talking 50% faster in 1995 compared to 50 years before — with further increases expected through 2009 — we ought to see more of a trend, I think, even allowing for individual variation:

This tends to undermine Lee and Liebenau's argument, I think, but it's somewhat weak evidence, since the total amount of speech is fairly small (only about 4.5 hours), and the change across time is also associated with a change of speakers. But next week I should have some better evidence for you.

Jerry Goldman at has been working on a time-aligned digital archive of U.S. Supreme Court oral arguments, from the 1950s through the present. He's been using some tools developed by Jiahong Yuan with a little help from me (our collaboration was funded by NSF grant IIS-0325739 "A Resource for Collaborative Research in Speech Technology, Linguistics, Decision Processes and the Law"). Jerry's crew have now got things digitized and time-aligned back through 1966, which gives us 44 years, with a large amount of speech (about 100 hours/year) from a gradually-changing cast of characters.

I've set aside a breakfast period next week to assemble and analyze this material from the point of view of potential secular trends in speaking rate. This is closer to the Norwegian parliamentary situation, and has the additional advantage that the recordings and transcripts are available on line for people to check or revise our work.


  1. Terry Collmann said,

    September 14, 2010 @ 1:41 pm

    Interesting that Kennedy is the second-slowest speaker on this list, since the received opinion seems to be that he was a fast talker, and this site claims that "his Inauguration Speech was powerful in content but Kennedy also delivered it with a rapid rate of speech." Maybe he just sounded fast …

    [(myl) Well, it's also possible that I made a mistake — you can follow the link and check up on me.

    Another issue is the difference between overall speaking rate and what you might call "duty cycle", that is, the proportion of overall speaking time that is speech rather than silence, or equivalently the ratio of speech to pauses. Someone can speak rapidly and also have a lot of long inter-phrase pauses, resulting in an overall slower rate. An analysis here showed that in one of the Bush-Kerry debates, Bush spoke faster (when he was speaking) but used longer silences, resulting in a lower overall rate.

    And then again, the received opinion might just be full of it, as the R.O. often is.]

  2. aaron said,

    September 14, 2010 @ 2:02 pm

    It seems to me that the length and frequency of applause could be as important as rate of speech in causing the variation in "words/minute."

  3. chris said,

    September 14, 2010 @ 2:03 pm

    the change across time is also associated with a change of speakers

    Of the four presidents who appeared twice in the table (why is there no entry for Eisenhower's second term? Was he not recorded?), only Reagan speeded up from his first inaugural to his second. But that's a *really* small sample size and it's likely that presidents giving their first inaugural address would be more nervous than giving their second.

    [(myl) The sites that I checked didn't have an mp3 file for Eisenhower's second inaugural, or at least not one that I managed to find during my lunch break.]

    Also: is there hard evidence that people with Southern "drawls" are actually talking slower than people with other regional accents? Carter's status as slowest-talking President would seem to be a point in favor, but then maybe it's worthwhile to note that four of the last five addresses (and the four slowest out of the five, at that) have been given by former governors of Southern states, which isn't true for the 60s-70s presidents. On the other hand, fast talker Johnson was also from Texas.

    [(myl) On the contrary, there's hard evidence that Southerners don't talk any slower, overall, than Northerners do.]

    P.S. Since these speeches were given live, I assume they included pauses for applause — did you take those out in measuring the length? If not, audience response could have a substantial effect on "speaking rate" without having anything to do with actual speaking rate.

    [(myl) Like I said, I just used overall elapsed time and overall number of words. You could do it the other way — I might have, if I'd had more time.]

  4. Morten Jonsson said,

    September 14, 2010 @ 2:15 pm

    Here's a recent collection of essays, edited by a sociologist and a political scientist, dedicated to the premise that life in general is indeed speeding up.
    The problem with this book, it seems to me, is that the essays take for granted that that premise is true, and that it's a modern phenomenon—one of the defining aspects of modernity, in fact—and draw some rather dubious conclusions.

  5. Jean-Sébastien Girard said,

    September 14, 2010 @ 3:17 pm

    FWIW my French 1994 Guinness Record Book (ISBN 2-87761-053-5) attributes (p. 249) to Kennedy the fastest talking speed, saying that he managed over 300 words per minutes "in a December 1961 speech".

    [(myl) Interesting. I wonder whether this is complete fabrication, or whether there is some germ of truth behind it. Does it give any further details about the speech, or cite a source for the measurement?

    It's common for speakers to reach 300 wpm for ten or twenty seconds at a time — and a fluent reader reading as fast as possible would generally be faster than that, depending of course on the distribution of word lengths in the material. ]

  6. Eugene van der Pijll said,

    September 14, 2010 @ 8:03 pm

    Re Kennedy's speech:

    This article from December 1961 seems to suggest that the speech in question is this one, to the National Association of Manufacturers on December 6, 1961. It claims a rate of 327 words per minutes, but the 47-minute speech is only about 6,500 words (if I'm counting correctly), so that cannot be the average for the entire speech.

    An audio link to the speech is included, but I don't have QuickTime installed.

  7. Dan Lufkin said,

    September 14, 2010 @ 8:51 pm

    To illustrate an application of cliometrics, I calculated the regression line of presidential speech with time. Counter-intuitively the speech rate is actually slowing down. The increment is quite small (-0.126 words/year), but the good news is that the word count reaches zero in June 3030.

    The correlation coefficient is only -0.09192, so there are probably other factors in play.

  8. Rubrick said,

    September 14, 2010 @ 10:19 pm

    I recall that Kennedy "record" from various Guinness Books of my past, and it always struck me as peculiar, since they didn't give any criteria for establishing it as a record or points of comparison. Surely the professional "fast talkers" who give the boilerplate at the end of radio ads could hit a mark far surpassing what any sane person giving an actual speech would achieve.

    I'd guess that this "fact" originally had more context attached to it ("fastest public speech" or something), which eventually withered away over years of editing.

  9. dirk alan said,

    September 14, 2010 @ 10:47 pm

    a sample taken from one speech ? better to average 100 speeches. whats wrong with you people ? do some research.

  10. Bill Walderman said,

    September 14, 2010 @ 11:05 pm

    What are we to make of the differentials between inaugural tempos for the same speakers? Reagan stayed about the same, but Nixon, Clinton and Bush II slowed down, in each case by roughly 10%. Maybe all that it suggests is that nothing can be inferred from the data.

  11. Will Steed said,

    September 15, 2010 @ 12:28 am

    I can't help but think that the news-readers on CCTV-1 (Chinese Central News) read the news at a rate far faster than any other news broadcast service.

    Unfortunately I think that's actually just a reflection of my poor political-Chinese skills.

  12. Jean-Sébastien Girard said,

    September 15, 2010 @ 2:33 am

    "Does it give any further details about the speech, or cite a source for the measurement?"

    Here's the entire blurb (with additional related records):

    "Word flow

    Few people can exceed 300 words per minute. President John Fitrzgerald Kennedy (191701963) was the most voluble public personality. In a December 1961 speech, he exceeded that limit.

    René Tramoni, from Nice, Alpes-Maritime, managed to pronounce 668 word in one minute on February 24, 1989. [Remember, this is a French national edition. There are random French equivalents/records dropped everywhere]

    In Reverse
    On February 6, 1990, Steve Briers (UK) recited the entire lyrics to the songs of Queen's A Night at the Opera backwards. He pronounced the 2,343 words in 9:58 m."

  13. Jean-Sébastien Girard said,

    September 15, 2010 @ 2:39 am

    ETA: The article found by Eugene talks about a "burst" of 327 words per minute (which I assume is not necessarily impossible: I bet the Major General songs and other patter songs may reach similar numbers, for example, though extemporaneous speech is another matter entirely), which probably would not be noticeable by just doing an average.

  14. maidhc said,

    September 15, 2010 @ 5:03 am

    This is not scientific, but as I remember Kennedy, he had a speech style that had a lot of dramatic pauses in it. "Ask not what your country can do for you PAUSE etc." I don't remember thinking that he talked particularly fast, either. I suppose you could find a few samples where he did, but on ceremonial occasions, which is mostly what I remember, he went for more of a stately style.

    Modern politicians don't use the style that Kennedy did. I think it would be perceived as elitist these days ("he must have studied oratory!"), whereas modern politicians want to seem more like one of the regular people. Consequently they tend to rattle through an entire sentence, then pause for applause from their claque.

    I believe that recordings of presidential speeches go back to Taft. There may be some Teddy Roosevelt available from after his Presidency.

    In the days before microphones, speeches to large audiences had shouters spotted through the audience. The speaker would say a sentence, then the shouters would relay it through the audience, then the speaker would say another sentence, etc. This must have had an effect on the way you would structure a speech.

    The Kennedy-Nixon debates were supposedly the first time that television had a major influence on an election. Franklin Roosevelt had used radio very effectively, but television even more so emphasized the intimate personal approach.

    Contrast that with Lincoln, whose political career was driven by his skill as an orator: the Lincoln-Douglas debates, the Cooper Union speech, the inaugural addresses, etc. But very few people would have heard them first-hand. Most people would have read the text in a newspaper.

    I wonder if the question should be whether there used to be more modes of talking than there are today. I don't see any reason why ordinary conversation would be any faster now than in days gone by. But consider that there may have been a different way of talking when giving a sermon, when reading a book aloud (common in the 19th century), when giving an address, etc.

    My local PBS station has been filling up odd intervals in programming with short videos of modern poets reading their poetry. Modern poetry doesn't have rhyme, doesn't have meter or any other structure. So how do you know it's poetry?

    The answer seems to be that there is a particular mode of speech that indicates that you are a poet. It seems to be fairly standardized among all poets.

    I've been trying to learn it by turning the sound down and concentrating on the pitches and rhythms. So far I haven't quite got it.

    One characteristic is that you have a series of pitches repeated at the same rhythm, terminating in a final lower pitch, but only about half as low as would occur in ordinary speech. Another variation is that the terminal pitch slides up (never down).

    For longer phrases you get intermediate up and down pitches, like —-^—-v—V. (The preview shows that this is not going to show up very clearly as syllables, but I don't know how to do better.)

    I'd like to master this way of talking, because then I could take any kind of prosaic subject matter and transform it into poetry, but I'm not there yet.

    Does anyone know if there is somewhere you could learn to talk like a poet?

  15. George said,

    September 15, 2010 @ 8:52 am

    Frankly, I find poetspeak a little distracting and annoying. Although I don't expect them to speak as if extemporaneously, I would prefer a more natural style.

  16. groki said,

    September 15, 2010 @ 1:53 pm

    @maidhc, @George:

    it bugs me some too, but I think the formality and flatness of poetryspeak (like white walls in a museum) is aimed at letting all the meanings in the lines come through. more natural speech patterns might focus on one or another interpretation too much.

    @maidhc: The answer seems to be that there is a particular mode of speech that indicates that you are a poet.

    per a lit prof years ago: poetry is what's laid out on the page in that weird way.

    @maidhc: It seems to be fairly standardized among all poets.

    does NPR include any, say, hip-hop poetry–with its greater number of (near-)rhymes–and if so, do you detect the same intonation patterns?

  17. maidhc said,

    September 16, 2010 @ 2:22 am

    groki –No hip-hop on PBS! Just a lot of old people.

    When I said all poets, in my mental model hip-hop is a different genre.

    Hip-hop has a lot of rhythmic and rhyme structure which would be considered hopelessly out of date by the PBS poets. I find those its most interesting features.

    It certainly has its stereotyped intonation patterns, but they are not the same as "serious" poetry.

RSS feed for comments on this post