Listen to Lexicon Valley Episode #18: The Rate of Exchange.
We’ve all known people who are deliberate, even plodding, talkers, taking their time with seemingly every word. And then there are those who spit out their sentences with barely a breath in between. Such variation among individuals is understandable (and at times even cultural), but what about among languages themselves? In other words, is Spanish in general spoken faster than English? Is English faster than Chinese? And how do we measure the speed of speech anyway? Listen as Bob Garfield and I talk about the common perception that foreign languages are spoken more rapidly than one's own.
You can also read the transcript of this episode below.
You'll find every Lexicon Valley episode at slate.com/lexiconvalley, or in the player below:
Send your thoughts about the show to firstname.lastname@example.org.
BOB: From Washington D.C. this is Lexicon Valley, a podcast about language. I'm Bob Garfield with Mike Vuolo and today: Episode No. 18, titled The Rate of Exchange, wherein we discuss the difficulty of measuring the speed at which we speak. Hey Mike.
MIKE: Hey Bob. How you doin'?
BOB: Splendid, thank you. Yourself?
MIKE: I'm good. I'm good. It's a beautiful day and I'm in a windowless studio, so what could be better?
BOB: Can't think of a thing.
MIKE: I wanna read first a recent review on iTunes from DrewInTN. He wrote, "I've enjoyed this podcast since it started, but only now am I leaving an iTunes review. Does that make me an asshole? Perhaps, but prior to 1970 or so I would have been a phony." So I want to urge listeners of this podcast to heed Drew's advice. Don't be an asshole. Subscribe to our feed in iTunes and while there leave a rating and a review.
BOB: I think Mike you really need to add, so this doesn't seem like a complete non sequitur and an insulting one, that that's a reference to our last show about Geoff Nunberg's book Ascent of the A-word, which is about the word asshole.
MIKE: Right, and I wanna say one more thing about that show. If you remember, Nunberg pointed out that "asshole" is used far more commonly in judging the behavior of a man than that of a woman. And he suggested that there's an inherent sexism responsible for the discrepancy, that when a woman is acting entitled and obtuse we attribute her behavior to some particularly feminine quality. Right, does that sound familiar?
BOB: Yeah, yeah, it sounds familiar to me. I listen to the show. I have to.
MIKE: Well, a listener named Avi Lichtenstein pointed out that more than 20 years ago the late, great reporter Molly Ivins wrote a column about Camille Paglia, of whom she was not very fond. And her column ended with the following quote: "There is one area in which I think Paglia and I would agree that politically correct feminism has produced a noticeable inequity. Nowadays, when a woman behaves in a hysterical and disagreeable fashion, we say, ’Poor dear, it's probably PMS.’ Whereas, if a man behaves in a hysterical and disagreeable fashion, we say, 'What an asshole.’ Let me leap to correct this unfairness by saying of Paglia, Sheesh, what an asshole."
BOB: [laughing] Wow. Molly.
MIKE: Yeah, more than 20 years ago Molly Ivins had already pointed out the inherent sexism of the word asshole. And for anyone who wants to read the entire column, which is really classic Molly Ivins, it's fantastic. It's called "I Am the Cosmos" and it's in the September/October 1991 issue of Mother Jones magazine, which you can find at Google Books. All right. Today's show. Bob, does the name Pimsleur mean anything to you?
BOB: Did you say Pimsleur?
MIKE: I said Pimsleur, yes.
BOB: Yes. Yeah, it's a language method. It says right here on my Google search. Sorry.
MIKE: [laughing] Did you just Google it?
BOB: Yeah I Googled it.
MIKE: Paul Pimsleur was a professor of French at various colleges and in the early 1960s he became very interested in how it is that students could best learn a foreign language. He ended up developing a series of principles that became known as the Pimsleur method, principles he then used to create instruction courses in French and Spanish and a number of other languages. Now, as an instructor, as a teacher, Pimsleur noticed that a common complaint among students taking a foreign language was that the teacher spoke too fast. I remember, many years ago, in between my junior and senior years of high school, I took an eight-week French immersion course in which the instructors, who were native French speakers, from the very first day spoke no English at all. I had only taken Latin up to that point, so I had no experience with spoken languages and I thought, there's no way I'm gonna be able to tease out one word from another. It just sounds like an avalanche of speech coming at me. And I spent the first few days of that course actually saying, "répétez s'il vous plaît, répétez s'il vous plaît," asking the instructors to please repeat what they were saying.
BOB: Did you use that kind of Israeli accent when you asked them to repeat?
MIKE: [laughing] Does my French accent sound Israeli?
BOB: It's not quintessential, let me just say that.
MIKE: I'm channeling my grandmother. So, here's how Paul Pimsleur described this problem back in the 1970s. He said: "The foreign words reach the listener's ear so rapidly that they soon pile up. The short-term memory overloads and the listener simply 'tunes out.' It is important to be able to control this factor in order to teach listening effectively."
The question that Pimsleur then set out to answer was, how fast do we generally talk? What's an average speaking rate in English? In order to measure this he wanted to choose a group of people whose speech was calibrated in a sense, consciously or unconsciously, for optimal listening comprehension. A group of people that he said spoke "slowly enough to be readily understood, yet rapidly enough to avoid boring the listener." So what group of people do you think he chose Bob?
BOB: News announcers.
MIKE: Yes! Exactly. Radio announcers in particular.
MIKE: Yeah. People like you!
BOB: You know, I'm not even cheating.
MIKE: Pimsleur and a couple of his colleagues determined that an average speech rate is between 160 and 190 words per minute. And it makes sense that human speech, all human speech, would exist within a particular range. Because at one end of the spectrum it's socially untenable to speak slower than a certain rate. If I were telling you a story, Bob, and I spoke one word every five seconds, let's say, very quickly you would get frustrated and be like, you know, dude I got stuff to do, I gotta go. You need to speed this up.
BOB: Yeah, it's like being in the fast lane of the interstate and some moron is ahead of you observing the law and going at 55.
MIKE: That doesn't make any sense to me.
BOB: It makes perfect sense. I think we should keep this in the show and ask the listeners to write to us to weigh in on whether they think the metaphor is accurate or not.
MIKE: Fair enough. So, at the other end of the spectrum it would be impossible, computationally impossible, to process language coming at you faster than a certain rate. In fact, it's been shown that people can maintain comprehension at speech rates approaching 500 words per minute. Once you get much beyond that, we start to lose the plot so to speak.
BOB: All right. So let's say Paul Pimsleur had it about right and that the average speaker goes from 160 to 190 words a minute, let's say 175 is the golden mean and it's typical and comfortable in English. Is it true in French and Spanish, German?
MIKE: That's the question, right? We've all had the experience of perceiving that some languages are spoken faster than others but is that in fact true? If you were comparing the speech rate of one language to another in terms of words per minute, what do imagine the problem with that might be?
BOB: Well, if it were German, when every word has 117 characters, there's no way that they could possibly speak at the standard 175 English rate. It's gotta be a lower number for German, for example.
MIKE: Yeah, exactly. Some languages have a disproportionate number of very long words, like German. Others are made up primarily of very short words, like Chinese. So a linguist from Finland named Jaakko Lehtonen conducted an experiment where he had native Finns both talk extemporaneously and also read a prepared text in Finnish. He then had native English speakers do the same in English. And when he compared the speech rate of the two languages counting words, English was faster, but when he counted syllables, Finnish was faster.
BOB: OK, my Finnish is a little thin, but I'm guessing that, like German, its words are longer on average than English words.
MIKE: Yes, that's true for a number of reasons. One of which is if I were to say for example in English "I'm going to the studio" as opposed to coming "from the studio," I would need a preposition to do that, "to" or "from." But in Finnish, which has cases, you add different suffixes onto the word for "studio" to differentiate between "to the studio" and "from the studio" and "in the studio," et cetera.
BOB: Uh-huh, uh-huh. So in a minute's speaking time in Finnish you use fewer words but more syllables than an English speaker in the same period.
MIKE: Yeah, think of it this way: If you took a page at random from an Ernest Hemingway novel, it's likely to have many more words than a random page of the same size from a Henry James novel, because Henry James, as Pimsleur points out, used words that were on average about 45 percent bigger than those of Hemingway. Hemingway, in our example here, is English. And Henry James is Finnish.
BOB: Henry James is Finnish?!
MIKE: [laughing] You didn't know that? Maybe that's why you had such trouble getting through his novels.
BOB: [laughing] Yeah, I thought they were a little ponderous, but I guess it's just a long word thing. So that suggests a problem measuring rates of speech, right? What do you use as a standard for counting? Is it syllables?
MIKE: Well, syllables have their own problems because they're more variable from language to language than you'd think. Some languages have what can be called a relatively simple syllable structure. For example, the vast majority of syllables in the Hawaiian language are one of two varieties, either a single vowel sound or a consonant sound followed by a vowel sound. So if you take the word "aloha" - three syllables, the first is a single vowel sound, the other two have what's called CV structure, a consonant sound followed by a vowel sound. A-lo-ha. English, on the other hand, allows for a much greater variety in which you can pack many more sounds, or phonemes as they're called in linguistics, into just one syllable. I spoke to a phonetician at the University of Michigan, her name is Patrice Beddor, and I asked her for a good example of a complicated syllable in English, one that would never be possible in Hawaiian or many other languages. And she gave me the example of the word "strength."
BEDDOR: It starts with an "s" phoneme, followed by a "t," followed by an "r," then the vowel "eh," and then another consonant that's called a nasal consonant, the "ng," and then the final sound which is the "th" but as a single sound is a fricative. So we have one, two, three, four, five, six sounds.
MIKE to BEDDOR: And if you made it plural then it would be seven.
MIKE to BEDDOR: So that would be s-t-r-e-ng-th-s.
BEDDOR: [laughing] Pretty good, yeah.
MIKE to BEDDOR: Strengths.
MIKE to BEDDOR: Seven sounds in one syllable.
MIKE to BEDDOR: Which is a pretty complicated syllable.
BEDDOR: It is.
MIKE to BEDDOR: That's about as complicated as syllables get.
BEDDOR: Ummm, they don't get much more complicated than that. It's way up there, let's put it that way.
MIKE: So "strengths" has three consonant sounds, or phonemes, followed by a vowel sound, followed by three more consonant sounds. Its structure would be CCCVCCC.
BOB: So you're saying not only isn't it accurate to measure speech velocity by word, even the syllable measurement is variable.
MIKE: Hold that thought for a minute. I wanna take a brief break to mention our sponsor this week Audible.com
MIKE: Years ago, a couple of linguists, Harry Osser and Frederick Peng, wanted to compare the speech rates of English and Japanese. Now, Japanese has a relatively simple syllable structure like Hawaiian. You can't pack very many sounds into a Japanese syllable. As Osser and Peng put it, "there can be as many as seven phonemes in an English syllable" - we saw that with the word "strengths" - "but at most only three phonemes in a Japanese syllable." And so they concluded that "a comparison of syllable counts would be biased."
BOB: So that leaves us with sounds, individual sounds.
MIKE: Exactly. They had a group of native English speakers and a group of native Japanese speakers talk about "student life" for five minutes each. These were all college kids at Cornell. And when they counted the number of sounds in a given one-minute period, there was no statistical difference between the English speakers and the Japanese speakers. And yet, all of the Japanese speakers thought that English sounded faster and almost all of the native English speakers they asked thought that other languages, including Japanese, sounded faster than English. So why do we feel this way? Why do many of us have this impression that other languages are being spoken faster than our own if it appears that at least when you count the sounds that we're making, the phonemes, we seem to be about the same?
BOB: Well, there's two things that occur to me. The first is that, duh, of course languages that you don't understand as well seem to be moving much faster than you can keep up with. It's a cognition issue, right? But the second thing is, when I tried to learn Spanish, more or less unsuccessfully, take the sentence "¿Cómo está usted?" - "How are you?" Native speakers of Spanish don't say it that way. They say "¿Cómo estáusted?" and they make essentially one word out of "está usted," essentially you lose an entire syllable in the process. Which makes everything go faster.
MIKE: Yeah, I think those ideas are pretty good and they comport more or less with some of the theories that are out there. One theory divides languages into different rhythmic categories. As one linguist put it a long time ago, there are languages spoken mainly with a kind of machine-gun rhythm and other languages that are spoken with more of a Morse code-like rhythm, if that makes sense.
MIKE: And the theory goes that speakers of the Morse-code rhythmed languages, like English, tend to think that the machine-gun rhythmed languages, like Spanish, sound faster. This theory is somewhat controversial because there's no really good evidence backing this up.
I really what Harry Osser and Frederick Peng said at the end of their study of Japanese and English. They formulated a hypothesis that went something like this: "When we, as native English speakers, listen to English we attend to the way in which the speaker distributes his speech and his pauses over time, i.e. we hear the speech and the gaps between the speech. However, when we listen to a foreign language being spoken we do not hear the pauses (other than the very long ones), rather we hear 'a continuous flow of speech.' ... As our acquaintance with a foreign language develops, we learn more and more about the units in the flow of speech, so that we are more likely to be able to judge the actual rate of speech correctly."
That makes a lot of sense to me. I think, you know, earlier I was saying how when I took that French immersion course I found it hard to imagine how I would be able to tease these words apart that were coming at me, but of course once you gain a familiarity with the language and the words then you do. Your ear develops in that way. And I'll just read to you one more thing that Osser and Peng wrote about this hypothesis. They talked more specifically about Japanese and English and they said, "When the Japanese speaker hears the bundle of dense consonant clusters of English he hears them in terms of the syllabic structure of Japanese, which of course does not have so many consonant clusters, and he therefore judges the speech to be faster than it really is." And then they said, "Similarly when an English speaker hears the successive vowel" sounds in Japanese, which we don't have as much of in English, that we judge Japanese to be faster.
BOB: You know, I think you kind of hit it on the head. It's really sort of an illusion, mainly one coming of unfamiliarity.
MIKE: Yeah, and let's go back to syllables for a moment. It's curious that languages would evolve in such a way that some, like Hawaiian, are made up mainly of these very small syllables that each on their own pack relatively little cargo, let's say.
MIKE: Exactly, right. That's the official state fish of Hawaii. Humuhumunukunukuapua'a—we call it the triggerfish in English—that's a very long word but it's made up entirely of, like aloha, either syllables that are one vowel sound or syllables that are a consonant sound followed by a vowel sound. Most of Hawaiian is like that. English, on the other hand as we discussed, makes use of a much greater variety of syllables, including lots of larger ones that pack more cargo.
BOB: You know what I always thought? I always thought that there should be an exchange program between Hawaii and the Balkans. Serbo-Croatian simply doesn't have enough vowels. Sometimes you see a street sign and there's like eight consonants in a row. What? You know, like a shoes and glove exchange for amputees, I think everyone benefits.
MIKE: [laughing] So, what would Humuhumunukunukuapua'a sound like in Serbo-Croatian?
BOB: Uh, Hrrmphyutchk.
BOB: I don't know.
MIKE: I didn't hear any vowels in that.
BOB: [laughing] That's what I was going for.
MIKE: Well, you know, that gets at this different way in which languages arrange their syllables. So very recently a group of linguists in France discovered an interesting insight into why languages might differ in this way. They thought that maybe we're asking the wrong question when we ask: Are some languages spoken faster than others? The point of speaking, after all, is not simply to produce sounds and syllables. It's to convey information, right? You've probably had the experience of seeing two sentences or paragraphs, let's say, side by side. They say the same thing in two different languages but one seems much shorter than the other.
So, here's what they did. They took a block of text, about five sentences or so, and translated it into seven different languages: English, French, German, Italian, Japanese, Mandarin Chinese and Spanish. Seven paragraphs in seven different languages all meaning the same thing. They then counted how many syllables it took each language to produce this written paragraph. And on one end of the spectrum were English and Mandarin Chinese, which required a relatively low number of syllables. And on the other end was Spanish, which required a much greater number of syllables.
BOB: So English and Chinese are more information dense. More is communicated in fewer syllables.
BOB: So, if I got this right, to express a certain idea it may take only 50 syllables in English and Chinese and 100 in Spanish.
MIKE: Yeah, and so English and Chinese had high information density and Spanish had a low information density. The other languages all fell somewhere in between and in fact they did this with 20 different blocks of text, so it wasn't just one anomalous paragraph. They then had a handful of native speakers for each language read the paragraphs and calculated how fast they spoke, by again counting syllables.
What they found was that the languages with the longer blocks of text, like Spanish, were spoken faster and the languages with the shorter blocks of text, like English and Chinese, were spoken slower - so that the rate of communicating information was approximately the same.
Now I know this may seem a little bit complicated, so I'm gonna try to couch this into a metaphor. I'm gonna pick up on the metaphor that you used earlier with lanes on a highway. Imagine that in one lane there are 50 Chinese cars all fully packed with cargo and in the other lane are Spanish cars that are only half-packed, so you need twice as many cars, a hundred, to get the same amount of cargo. Now, when these cars start moving, it turns out that the Spanish cars are moving twice as fast, so they end up delivering the same amount of cargo at the same time. In other words, the last car of each lane will cross the finish line at the same time. Now, I should say that this ended up being true for all the languages that they studied with the exception of Japanese, which was a bit of an outlier and it's not really clear why.
BOB: Let me see if I have this right. If you wanna measure the talking rate, you can do it by counting sounds and by that measure it seems that most languages come out at about the same place. But if you want to discuss the rate of information conveyed in a given amount of speech, you can do that best by measuring syllables where, again, most languages come out about the same. And I guess the third point would that Japanese is just weird.
MIKE: [laughing] Wait, Japanese is weird or the Japanese are weird?
BOB: No, not the Japanese, although, no I'm talking strictly about the language.
MIKE: If you would like to defend the linguistic honor of the Japanese feel free to do so by writing us at slatelexiconvalley@gmail dot com, that's slatelexiconvalley@gmail dot com. You can find past episodes of our show at slate dot com slash lexiconvalley. Remember, don't be an asshole, subscribe to our feed in iTunes where you can leave a rating and a review. I wanna thank Patrice Beddor at the University of Michigan and of course Andy Bowers, the executive producer of Slate's podcasts.
BOB: Once again Mike, I'm nervous here. That was another gratuitous swipe at non-iTunes commenters. It's a reference to our last show. He doesn't mean to insult you. There's an actual book about it. I swear. Geoffrey Nunberg wrote it, you could look it up.
MIKE: [laughing] Thanks for saving me.
BOB: Yeah, we done here Mike?
MIKE: We're done.
BOB: All right, man. Later gator.