What language is music?
March 15, 2009 6:52 PM   Subscribe

Western musical intervals are derived from speech tendencies, according to Duke scientists. Specifically, "most of the 12 chromatic scale intervals correspond to peaks of relative power in the normalized spectrum of human vocalizations." A somewhat more layperson-friendly summary of the study is here.

Some think that language and musicality evolved in tandem (see the singing Neanderthals), and Mendelssohn thought that the communicative ability of music is even more precise than that of language (as related by Oliver Sacks, at 9:05 - .mp3 link).
posted by univac (41 comments total) 37 users marked this as a favorite
i love this stuff. also see diana deutsch's work, (i found out about it via an awesome radiolab episode-

sometimes behave so strangely..
posted by localhuman at 6:56 PM on March 15, 2009 [2 favorites]

posted by delmoi at 6:58 PM on March 15, 2009

The weirdest thing: after having listened that radiolab poscast months ago, maybe even over a year, reading "sometimes behave so strangely" came out in my head as a melody. (listen to the podcast to understand)

But what does this say about other music traditions that use different scales?
posted by pantsonsteven at 7:04 PM on March 15, 2009

Steve Reich, among other composers, has composed works where fragments of speech turn into melodic fragments. His early pieces, It's gonna rain and Come Out already have this, as do Tehilim and Different Trains. Sorry for the lack of links; I just came back from a night out and I'm too tir1aswqw;.;ej zpieraw *THUD*.
posted by LMGM at 7:12 PM on March 15, 2009

Studies like this annoy me because this guy is acting like the 12 tones of the equal tempered scale somehow came out of nature and are imbued with all of these fundamental properties. When in fact, the equal temperament tuning system, as used on the piano, in western music and nearly all of its instruments, is an "imperfect" system, deliberately out of tune in order to allow for chromatic transposition to any key, which is a cornerstone of western music. If the intervals on the piano were tuned to be scientifically perfect and accurate, then western harmony begins to break down and an entirely new form of harmony must be created! Hmm, I don't even know if there is a point to my comment except to say, I wish he would've studied 'just-tempered' intervals instead of the equal-tempered system.
posted by ChickenringNYC at 7:16 PM on March 15, 2009 [6 favorites]

See here.

And check here for a really clear explanation of the fundamentals behind just intonation.
posted by ChickenringNYC at 7:19 PM on March 15, 2009 [1 favorite]

ChickenringNYC just said everything I was going to say.
posted by dfan at 7:38 PM on March 15, 2009

And check here for a really clear explanation of the fundamentals behind just intonation.

And here's a musical illustration:

J.S. Bach's The Well-Tempered Clavier (equal temperament)
La Monte Young's The Well-Tuned Piano (just temperament)
posted by jonp72 at 7:39 PM on March 15, 2009 [3 favorites]

There's a book that draws a similar conclusion about certain kinds of poetic meter. I realize it's a controversial thesis, but nevertheless such theories remain quite interesting (.pdf). Similar questions about the evolutionary role of motherese, which would appear to be universal, are also suggestive.
posted by ornate insect at 7:42 PM on March 15, 2009 [1 favorite]

What a very, very interesting post, thank you. This is the conclusion of the explanation:

These results imply that human preference for the intervals of the chromatic scale arises from experience with the way speech formants modulate laryngeal harmonics to create different phonemes.

The researchers seemed to have made a fundamental assumption that, it seems to me, is unwarranted: that if there is correspondence between musical intervals and speech, that the musical preference is derived from speech, when it could likely be the other way around. All they've shown (at least as detailed in their summary) is a connection, not a causal relationship. Mithen (author of The Singing Neanderthals, mentioned above) would likely reverse the conclusion, and say that this evidence could show our speech is derived from music.

pantsonsteven: But what does this say about other music traditions that use different scales?

Same thing, probably--most scales all over the world are also drawn from a 12-tone chromatic interval structure within the octave, it just depends which of the 12 are used to make a scale (the western major scale, for instance, uses only 7 of the 12). The few cultures that use micro-tonality (like quarter tones in some Chinese music) also make some use of the 12-tone division of the octave. (At least, as far as I know. Which isn't far, admittedly, when it comes to non-Western musics.)

ChickenringNYC, I hear you w/r/t pure tuning, but I don't think that's really a salient concern for this study. Pure tuning still uses a 12-tone octave, it's just that the divisions are a little different. Other studies (too lazy to look up, sorry) have shown that the human ear allows a fair bit of variance within recognition of interval qualities like consonance--something like 15 cents at least, I think, even though a reasonably sensitive ear will perceive a 2-3 cent variance as "out-of-tune"--we will still recognize a very out of tune perfect fifth as a perfect fifth, in other words.

(Also, FWIW, I find that musical ensembles always move toward pure intonation intuitively--no matter how hard they try to impose their scientific equal-temperament on us, musicians still want to lower that major 3rd by 14 cents, or that seventh in a V7 31 cents anyway.)
posted by LooseFilter at 7:42 PM on March 15, 2009 [2 favorites]

(Also, I'd love to see the post title reversed: What music is language?)
posted by LooseFilter at 7:47 PM on March 15, 2009

So are they saying that the formants of speech sounds have the same mathematically pure ratios as found in the intervals of the 12-tone scale (sort of, see above comment by ChickenringNYC)... or just about everywhere else in nature?
Snark aside, it would be fascinating if this could show how different languages make for preferences or tastes for specific intervals. E.g. Eastern European languages vs. the preference for the minor second in a lot of that music (grating to Western ears). Flamenco, and last but not least the sound of crickets, musical to Japanese ears (or so I heard).
Also, does anybody know if this means anything useful for the 12 tone scale, or music, or both?
Thank you for posting.
posted by yoHighness at 7:54 PM on March 15, 2009

Local (to me) musician Charles Spearin, whom you may know from Do Make Say Think or Broken Social Scene, has been exploring this recently in great detail.

His latest work -- The Happiness Project -- explores the melodies living inside everyday speech. I caught a bit of it on CBC radio the other night, and completely forgot about it until I read this thread.

Definitely worth a peek. And thanks for reminding me!
posted by tapesonthefloor at 8:04 PM on March 15, 2009 [2 favorites]

I couldn't help but think of this as i read the article.
posted by 5imian at 8:36 PM on March 15, 2009

Those formant bands are fuzzy swatches with widths of 500Hz or so, I don't think that the relatively microscopic adjustments of just intonation would be relevant at this resolution. After all, the whole point of just intonation is that, justified or not, a third is still perceived as a third. That's the third the authors are talking about.
posted by StickyCarpet at 8:55 PM on March 15, 2009 [1 favorite]

They have this all wrong.

Creation of those consonant intervals using the same vibrating meat-apparatus requires less energy, because they are (nearly or exactly) higher harmonics of one another. Evolution, taking the easiest path, would favor those in language, for lack of other options (ask any vocalist - it's hard to sing more dissonant intervals, which is why people naturally sing true perfect fifths instead of well-tempered ones). Consonance is the path of least resistance, which speech and music follow for the same reasons.

We hear through tuning-fork like hairs (cilia) in the organ of corti, each of which will be activated much like any other resonant system - strongest at a particular frequency and it's (sub)harmonics. Consonant intervals "activate" fewer such hairs, and are heard more... simple. The way things naturally vibrate produces consonant intervals (basically, because near-linear restorative forces are ubiquitous in nature) so it makes sense our ear would find them "normal" - I'd expect hairs with related resonant frequencies would be associated with one another neurally.

Also, the division of the octave in to twelve, regardless of how you solve wolf-tone related intonation issues, has mathematical rationale. It's simply harder to get the more consonant (speaking in the objective energy-efficiency sense here) intervals to drop out out of smaller bases (see here).
posted by phrontist at 8:56 PM on March 15, 2009 [8 favorites]

People seem to love just-so stories about art. Were any predictions made, or did they make up a plausible story explaining what we know we like?

The harmonic combination of intervals outside the standard 12 tone scale (modulo that slop factor that LooseFilter alluded to), sometimes make your head feel like it is leaning on a washing machine in spin cycle. There are certain kinds of sound I have played with using in my music that actually make parts of the inside of my head tickle if I try to do them with my mouth or trumpet, things that play around in the places where timbre and harmony and tuning system blend into one another a bit. It kind of makes sense that these things would be avoided in everyday speech or the kind of music you don't want to think about or be distracted by. I am not going to pretend that this pondering has any scientific validity though.

I get this feeling sometimes that someone is saying "Hey, you know that thing you really like already? I have mathematical and scientific proof it is the best thing ever, the best thing possible in all possible worlds! You are just so awesome like that!". I mean why don't we have cooking experts who are all like "we are genetically formed to acknowledge that sugar and butter are the best things ever, and that is why we are better than lower animals, and why it is good that we eat as much candy as we can as often as we can". I guess pop music doesn't make you fat or anything, but you know there are wonderful things to be found outside the most accessible areas of comfort.
posted by idiopath at 9:02 PM on March 15, 2009 [2 favorites]

On non-preview, phrontist may have made a better case for what I was trying to say in my second paragraph.
posted by idiopath at 9:04 PM on March 15, 2009

Let me take another stab at this...

They seem to propose: The Natural Order gives rise to Speech which gives rise to our system of musical intervals.

I'm saying: The Natural Order gives rise to speech and our musical intervals, which mimic one another due to the same underlying physics of resonant objects. In other words, the consonance of a given interval is solely a quirk of human evolution.

Don't believe me? The concept of an interval applies just as well to rhythms as notes...

A two against three polyrhythm is very easy to follow with the ear... just listen (this would be the rhythmic equivalent of a perfect fifth). A three against four is a little harder (this would be a perfect fourth). The difficulty of playing a given polyrhythm is directly related to the lowest-common-multiple of beats you have to count (consciously or otherwise), and most people need to learn to hear and feel comfortable with high order polyrhythm. Does this sound normal to you? Figure out the equivalent tonal interval name with this handy chart.

I think any conscious being capable of hearing would "hear" (or otherwise perceive - a blind person watching lights flash would rank them in the same order of complexity, I'd imagine) and judge rhythms in the same order of consonance...


So too do I think our appraisal of such intervals as tones (which are really just very, very, fast rhythms) is intrinsic to their nature (as oppose to cultural). What's cultural is our appreciation for kinds of dissonance, and how we expect to hear them (which can be learned).

It's great to see the beauty of which you're a part. It's precisely because you're a part of it that you see it as beautiful.
posted by phrontist at 9:39 PM on March 15, 2009 [3 favorites]

Also, I stumbled across this neurological study done on monkeys a while back, and while I don't feel qualified to summarize the results, it's interesting.
posted by phrontist at 9:43 PM on March 15, 2009

Pure tuning does NOT use a 12 tone octave. The concept of having 12 notes in an octave is COMPLETELY arbitrary and evolved over a very long period of time before it was more or less set in stone by a few "intellectuals" of the period who wrote textbooks and theses and treatises and composed mountains of religious music in that style. Harry Partch used as many as 41 tones per octave. Dean Drummond invented an instrument which plays 31 tones to the octave. Just like 12, those numbers are also arbitrary. Using just intonation frees the composer to use as many tones as he or she wants, all codified by the number/number ratio which is the simple beauty of the practice. i.e. "G" is 1/1, "D" is 3/2 and so on..
posted by ChickenringNYC at 10:25 PM on March 15, 2009 [1 favorite]

ChickenringNYC: See this link. You can, of course, use different numbers for equal-temperament, or eschew equal-temperament entirely, but that doesn't mean 12 is arbitrary. 12-TET specifically gets you intervals closest to perfect intonation without jumping up to a very high number of divisions. TET in general is popular because it allows an instrument to play in all keys and play music that modulates between keys without loss of intonation purity.
posted by phrontist at 10:51 PM on March 15, 2009

The concept of having 12 notes in an octave is COMPLETELY arbitrary

Well, except for the overtone series.
posted by LooseFilter at 10:59 PM on March 15, 2009

(Thanks for the great link, phrontist.)
posted by LooseFilter at 11:01 PM on March 15, 2009

Music from human speech - Steve Vai - listen starting at 1:07.
posted by plinth at 6:38 AM on March 16, 2009

Over tone series, smho-vertone series. Both the Partch and Drummond systems are also dervied from the overtone series. You can extrapolate as many pitches as you want from the overtone series, and some of them are quite wacky, harmonically. The notion that 12 is the "simplest to hear" or "provides the most "comfort" to the ear is still hogwash to me. Open your ears!
posted by ChickenringNYC at 8:07 AM on March 16, 2009

Open your ears!

You should know your audience before making assumptions.

The notion that 12 is the "simplest to hear" or "provides the most "comfort" to the ear is still hogwash to me.

I respect your perspective on this. Please stop being so dogmatic and dismissive of other, perfectly reasonable points of view.

Considering that, within a single fundamental, the most common scale tones occur first and repeatedly and are actually audible, and that this can be demonstrated objectively, I'm not sure this is an area open to opinion statements, however. If one wishes to use multiple fundamentals in pure intonation to derive 31 divisions within an octave, great! But don't claim that what you're doing is connected to western tonal practice; it's orthogonal to it.
posted by LooseFilter at 9:55 AM on March 16, 2009

I dunno, Indonesian gamelan music seemed to be okay sticking with 5- and 7-tone systems for hundreds of years. Other Eastern musics also were based on 5 tones, if I'm not mistaken. Those tones didn't "fit" into 12-tone equal temperament -- you can shoehorn them in and make the case that they are close, but that's a pretty strong rash of cultural bias you got going on there, you might want to get it checked out.

12 tones isn't arbitrary. Nor is it inevitable. It's just practical.
posted by speicus at 10:53 AM on March 16, 2009

"...I'm not mistaken. Those tones didn't "fit" into 12-tone equal temperament .."

the "yellow bell" pentatonic scale in eastern music is comprised of the first 5 overtones. so yeah, they fit quite nicely indeed.
posted by stubby phillips at 11:00 AM on March 16, 2009

Stubby, the overtones are part of the just intonation scale, not the 12-tone equal-tempered chromatic scale.

And actually, if you look at Figure 6 from the link, NONE of the tones of the equal tempered scale are actually present in the formants of the human voice. They're all just intonation intervals (as one would expect from a natural instrument like the human voice). Also, there are only 9 tones present -- the major 7th, major 2nd and minor 2nd are all missing.

If anything, this makes the case that the 12-tone chromatic scale is NOT natural. So yeah, this is a pretty horrendous misinterpretation of their own data.

They do say that this accounts for the relative perceived consonance and dissonance of the various intervals, which is a more reasonable case to make. But this can be explained by acoustic phenomena that occurs everywhere in nature, not just the human voice. And again, it doesn't relate to the equal-tempered chromatic scale. It relates to just intonation.
posted by speicus at 11:18 AM on March 16, 2009

Also, slendro tuning in gamelan music can have tones that are almost 50 cents off from equal-tempered tones (i.e. almost halfway between two of our tones).
posted by speicus at 11:24 AM on March 16, 2009

stubby phillips, as speicus alluded to, the overtone series and equal temperament are an uneven match, and thanks to equal temperament and instruments like the piano that cannot slide in pitch, we have intervals in western music that are alien to pretty much any other music ever made. For a long time ethnomusicologists would try to transcribe traditional music as if they were trying to play equal temperament dodecaphonic scales, but were just out of tune, but thankfully most of us should realize the folly of that now.
posted by idiopath at 11:55 AM on March 16, 2009

I've written and trashed two comments for this thread already, so let me make this short:

Chromaticism is justifiable; logarithmic scales are a means to an end.
12√128 ≈ 3/2. 3-limit harmonies are well represented in 12 TET.
posted by Monstrous Moonshine at 4:34 PM on March 16, 2009

Of course 12-tone chromaticism is justifiable. Did anyone say it wasn't? Maybe I wasn't being clear. All I mean to say is that 12-tone equal temperament is not the inevitability that this study, and some in the thread, are making it out to be. It's just one of many imperfect options.
posted by speicus at 4:43 PM on March 16, 2009

Does the article refer to the tempered scale? I missed that part.
posted by stubby phillips at 5:18 PM on March 16, 2009

stubby phillips: They start the article talking about a chromatic scale. A chromatic scale is 12 tone equal tempered, and they are either misinformed or lying when they say that a chromatic scale is universal or even substantially present outside the western tradition and its offshoots.
posted by idiopath at 6:05 PM on March 16, 2009

A chromatic scale is not necessarily tempered.
posted by stubby phillips at 6:06 PM on March 16, 2009

Mea Culpa, I checked their charts and it appears they are actually using the Pythagorean chromatic scale. This invalidates some of what I said earlier.

I have kind of a chip on my shoulder about smug confidence many seem to have that the way we tune pianos today is empirically the best set of frequency ratios for all music, ever. This study was comparing vocal overtones with a Pythagorean chromatic scale with just intonation, so my venting here was out of place, hardly anybody tunes their pianos by just intonation anyway.
posted by idiopath at 6:37 PM on March 16, 2009

That's true, chromaticism doesn't necessarily imply equal temperament. I am so used to seeing them used interchangeably... I guess I had my blinders on.

Their data still only accounts for nine tones, though, and doesn't really answer the question "Why are 12-tone systems so prevalent?" Instead it raises a new one: "Why aren't 9-tone systems more prevalent?"

The overtone series doesn't really account for the missing tones either since the missing tones are all very high, faint harmonics (9, 15 and 17).

I still think the most reasonable explanation is that the 12-tone system came about as a practical necessity, a kind of emergent property of the kind of music people were making around the Renaissance and Baroque periods, and it stuck around out of inertia. This won't satisfy the prescriptivists, though.
posted by speicus at 12:18 AM on March 17, 2009

The western 7-note scale (church modes) is strongly influenced by the overtone series, as all but the 4th of a major scale are within the first 16 partials, which cover 4 octaves from the fundamental. Most western instruments work within that 3rd and 4th octave above the fundamental--wind instruments of course manipulate the overtone series directly--so it's easy to see where the 7-note scale came from.

My understanding is that the 12-note chromatic octave came from combining the 7-note scales of different fundamentals so that (more advanced) modulation was possible, not from some sort of a priori decision to cut an octave into 12.
posted by LooseFilter at 12:51 AM on March 17, 2009

I still think the most reasonable explanation is that the 12-tone system came about as a practical necessity, a kind of emergent property of the kind of music people were making around the Renaissance and Baroque periods, and it stuck around out of inertia. This won't satisfy the prescriptivists, though.

Yeah, this. I avoid prescriptivists as much as possible. Dogma in creative art-making is baffling to me.
posted by LooseFilter at 12:55 AM on March 17, 2009

« Older Gar's revenge   |   They paid what to who? Newer »

This thread has been archived and is closed to new comments