A Sonic Atlas of English Language is a 131-page book the bulk of which consists of a list (to quote the title page) of “8000 English words, organized by the relative audio frequency of each word, from the highest to the lowest pitched: A potential reference for spatial acoustics and sound design, the study of hearing loss and speech, musical and lyrical composition, sound art and poetics.” You can download the pdf here; Trevor, who sent it to me, added the following quote from I know not what source:
Those interested in the musical or “phonographic” qualities of language will find the Atlas useful. All language has a frequency, and the relative frequencies of English words can be organized to follow musical concepts: For example, and as pointed out by author Shane Butler, the following words will tend to be spoken from high to low sound frequencies: beat, bit, bet, bat, and bought, something confirmed by the calculations in the Atlas. The Atlas also enables other, almost endless combinations of words to be organized by their relative sound frequencies. In addition, the Sonic Atlas also enables authors to assemble work out of particular phonemic patterns. For example, the Austrian poet Ernst Jandl, developed a German sound poem almost entirely composed of words with the “sch” and “f ” (/ʃ/ and /f/) sounds, while American poet Kenneth Goldsmith’s poetic work “No. 111” focuses on the “er” (/ɜr/) sounds in American English words. These are just a few of many examples of writing that the Atlas might cultivate. (p. 84)
It’s not anything I’m likely to use, but I’m sure there are those who will be eager to download and play with it. Thanks, Trevor!
Very strange. First of all, the “frequencies” are based on a formula. Provide the damn formula! What exactly is meant by pitch or frequency is not defined. This is far from obvious, especially for the consonants. Frequency ranges are absurd, they run from high 2.5-5kHz to low 175-750Hz. No way there is a systematic difference of about 10 times no matter what is meant by frequency. And it won’t hurt to provide at least some evidence that it has anything to do with reality.
Language has a lot more than *one* frequency. Indexing a bunch of words by some single personal dimension seems a crankish effort. Create a phonological database and look up words according to whatever criterion you actually want, not this dude’s formula.
Yeah, it did seem a bit crankish, but since I know nothing about these matters (phonetics was never an interest of mine), I thought I’d put it out there to see what better-informed folks thought.
Language of course has one frequency, f0. Well, there’s also the first through third formants, but f0 is what most people mean when they talk about the pitch of someone’s voice. That said, I have no clue what Gissen is doing. He mentions absolute and relative frequency together, but only gives absolute frequency ranges in the list itself? And his mention of “laryngeal tones” doesn’t exactly lend credibility to his project.
I know very little about phonetics but his description of how he created the atlas seems very dodgy. He says he gives each phoneme a frequency — but doesn’t say how he determined that number. And then he says that by writing each word as a sequence of phonemes he derives an average frequency for the word.
But we can say any word with a rising or falling or neutral intonation, so the notion of a unique average frequency is absurd. And even if you accept his numbers, what’s the point? It seems like saying that the average color of neapolitan ice cream is a sort of pale dirty red.
Huh? It’s just a list, no maps/charts, nothing geographical. In what sense “Atlas”?
From the blurb (Afterword — Trevor’s quote is also from there):
” We use reference materials such as dictionaries and thesauri to define and categorize words and their meanings and anthropological references and surveys to explore language’s regional and cultural particularities.”
There’s nothing exploring regional particularities. Do all dialects in the U.S. exhibit this particular frequency ordering?
@D.O. Provide the damn formula! It’s described last para p. 82, sorta.
Most of Gissen’s “experimental writings” http://davidgissen.org/ seem to be about architecture/landscape/environment. (I tried to sample a few, but they seem straight out of “Pseud’s Corner”, I came out in a rash.)
More from the Afterword:
“While the Sonic Atlas is an admitted linguistic curiosity, and will most likely be enjoyed by those simply interested in the particularities of English language, it nonetheless was developed with some disciplinary concerns in mind.
This study of the sound frequencies of English words emerged as part of a broader study that examined how language is physically impacted by spaces and environments. The ability to hear over distances and within spaces is governed by various acoustical and auditory phenomenon such as volume and reverberation, but also by pitch. Higher sounds produced by the human voice do not travel easily in space.”
So if you’re going to a whispering gallery, take the list with you.
Roughly, what he does is assign each phoneme a frequency based on something like the centroid measure, which is common measure in phonetics. For the centroid, take a frequency spectrum, and find its weighted average. For example, voiceless fricatives have just high frequency noise, so they rate high. For vowels, the location and relative intensity of the formants determine the centroid frequency.
You then average this over the phonemes in each word. So sibilant-rich words like sixths will be at the high end; words with high back vowels and semivowels, like woo, which have a low F₁ and F₂, will be at the low end.
People can hear these differences, subjectively. He’s trying to quantify it. It’s a harmless curiosity. Clearly he’s not much of a phonetician, but whatever.
If he’s having fun, I certainly have no objection, it’s harmless fun. It’s just unaesthetic to me. If he had ordered vocal sounds by pure perception, his sonic intuition of their height, I would dig that. (Perception of abstracts in sounds is interesting.) Or if he wants to devise a principled formula, sure. But halfassing it, neither intuition nor good phonology, makes me sad for him. He’s happy though.
The science of the list looks like it is based on an audiogram (a hearing test) and with the phonemes averaged, just as “Y” says above. The list of homonyms throughout this thing is fun.
Generally, I don’t respond to critics of my work, especially when the critics are anonymous! But thank you for posting the pdf of my Sonic Atlas. The author “Y” is correct when they describe how I calculated each word. As for some other comments, I’m on the fence as to the utility of listing average frequencies at the bottom of each page, but they relate to audiograms, and what is described below.
In sound mixing and acoustic design, artists might calculate the average frequencies of certain tracks to test equipment and spaces, and it can be done with words. Here is a link to a movie — https://vimeo.com/210049498 that shows the ordering of three words in the Atlas. It uses a computer produced voice and that can be found here: https://www.naturalreaders.com/ and Adobe tools for frequency analysis.
Thanks again, and I am pleased that my work inspired so much interest from your readers.
Some of us are obviously pseudonymous, other not so obviously; still others operate under their own names. But few or none are actually anonymous.
Yes, I’ve been languagehat online since 2002, and am far better known by that moniker than by my real name. At any rate, thanks for dropping by, and I’m glad you liked the post!
Dear David, I apologize for calling you “not much of a phonetician”. You do use technical terms carelessly, but I hope my phrasing did not give offense.