The Vowel Space

[@DrGeoffLindsey]

To try everything Brilliant has to offer-free-for a full 30 days, visit brilliant.org/DrGeoffLindsey/. The first 200 of you will get 20% off Brilliant’s annual premium subscription.

[@Somebodyherefornow]

what if you make the rounding into a third dimention, similarly to adding a z for the blue cone in CIE diagrams does that vowel chart exist?

[@beepboop204]

🙃🙃

[@yura2424]

0:05 I want to do like you at 0:05 I downloaded the Praat program from the description, but it does not work this way. It seem to be missing the audio files. Could you help to set it up, so it looks like yours?

[@Ggdivhjkjl]

Could you release a video on how to use Praat please?

[@amelali6299]

Dr Geoff Lindsey, could you please tell me what is the purpose of vowel space quality? Hope you reply as soon as possible

[@rosehipowl]

I studied linguistics at university and I can't believe I was taught "wrongly"! Your chart makes a lot more sense to me - I hope it gets more widely adopted. I love your channel because it really has made me think differently about the things I was taught, like that analogy to the colour spectrum really made links in my brain that I had never thought of before!

[@Ggdivhjkjl]

For some years now, universities have perpetuated established teachings instead of being places of free inquiry. So it's not surprising you were indoctrinated into predetermined ideas instead of being encouraged to explore new ones.

[@Unbrutal_Rawr]

​@@Ggdivhjkjl This is a pile of holier-than-thou bullsh*t. All systems of knowledge are based on perpetuation, no system of knowledge is possible by making students invent the entire science from scratch. All institutions for instruction and disseminating knowledge operate with predetermined ideas by definition. This is not instead of encouraging people to explore new ideas - old ideas is the foundantion that anyone who wants to explore new ones must have firmly established. The vowel space diagram is a basic linguistic tool of description which is required for any linguist to be able to discuss their subject. Teaching it to students is no more and no less indoctrination in old ideas than teaching them algebra, or Venn's diagram. When you teach someone the vowel space diagram, you need to explain what exactly its symbols and dimensions represent. You cannot teach them the diagram with no explanation and then expect them to explore new ideas about what it all stands for. The point of this video was that the diagram can be made to more accurately account for reality if we assume that its dimensions represent something different than originally envisioned by its inventors, or many of its current users, namely sound frequency instead of tongue position.

[@marioluigi9599]

Get a refund. Lol

[@yellowest_ball]

I learned all of this on Wikipedia. I've never trusted school.

[@flamingspinach]

Same here! I clicked on this video expecting to get a refresher on what I'd learned at university but got something completely different and better :) I had known about formants before, but never really explored the connection (or rather, the lack of connection) to the traditional quadrilateral IPA vowel chart.

[@ksiistoyiiwa]

This is the best explanation of vowel quality I have ever heard - thank you, Dr Lindsey!

[@gagadaddy8713]

Thank Dr. for the great presentation! It is the best illustration for linguistic topic analog with science and math I have ever seen. I have pinned the IPA chart on my wall for years, but I never got an idea what exactly the meaning of x-axis and y-axis there and why schwa is in the center. This video answer all these queries which slept in my mind for ages. Love it!

[@banafritz]

Btw isn't Azzurro just azure?

[@lunasophia9002]

I'd like to point out how uncommon it is for an academic open source project like this to be so actively maintained. What often happens is a student (typically someone working on a graduate degree) will work on something as part of their education, finish their degree, and it never gets updated again. But praat has a history going back more than 18 years (the github repo only goes back to 2005, and even then they'd passed version 4). Quite impressive!

[@DrGeoffLindsey]

Stunning.

[@DroolRockworm]

Then how have they not been able to make it work in real time yet?

[@lunasophia9002]

@@DroolRockworm Like most projects, lack of interest / time.

[@haukur1]

@@DroolRockworm Unfortunately, I think stubbornness on part of the developer(s) plays a role.

[@blerst7066]

The analogy between colors and vowels was perfect. I have no idea why I haven't seen anyone explain like this before.

[@margett__]

As someone who's currently working on bettering their English pronunciation, the tongue position explanation never quite made sense to me. Thanks to your brilliant video, I can finally understand why. I find it so much easier to trust my ear and try and replicate sounds this way. Somehow my tongue and lips find their way by following the information my ears have gathered. Truly fascinating stuff.

[@sirknight4981]

Actually I never thought about that. How do we know how to replicate sounds with our mouth after just hearing them? I always found it strange when musicians talked of playing by ear. So many little miracles go unnoticed in the day to day.

[@PC_Simo]

@@sirknight4981 Yes. The brain be a wondrous thing, indeed. 👂🏻🧠👄👅😌

[@madisondampier3389]

As a choir singer, being able to replicate a sound I hear is just natural (as long as it's physically possible), I have so much experience with singing by ear that any sound I hear is immediately associated mentally with the position and movements my mouth and throat would need to take in order to produce it, I am pitch perfect! And this video is a phenomenal demonstration of this vocal space, even though the acoustics of our mouths is a highly complicated mess in reality, we can still use approximations of its shape to describe the differences between distinct sounds in our vocabulary

[@thedamned8202]

​@@sirknight4981I suppose we don't completely lose our ability to imitate sounds after our childhood

[@sirknight4981]

@@thedamned8202 Astonishing that we should have ever been blessed with it in the first place 😅

[@m.germano]

As a native Brazilian, I missed the NASAL set of vowels ã (nasal A - não) and õ (nasal O - aviões), as well as the unwritten (but nonetheless spoke) ˜u (nasal U - muito), ˜i (nasal I - quinto) and ˜e (nasal E - mente). It would probably require another 3rd (or even 4th dimension hahaha) vowel space representation. I should state that I LOOOOOVED your video. It's so instructive that I keep watching it repeatedly. Maybe, you are planning to create a video about the TONES of vowels like its used in Mandarin or Cantonese

[@DrGeoffLindsey]

Thank you! Of course you're right that the video doesn't cover every aspect of vowels, particularly the "diacritic" features such as nasalization, voice quality and tone. Nasality has complex effects on resonance, affecting both the frequencies and bandwidths of the resonant peaks, so we couldn't simply add a dimension like F1/2/3.

[@mskiptr]

That's something I'm still wondering about. In my native language - Polish - we have two 'nasal vowels': ⟨ą⟩ and ⟨ę⟩. These are sometimes transcribed as /ɔ̃/ and /ɛ̃/, but also as /◌ŋ/, /◌ɲ/ and even /◌w̃/ (the last one is something I am yet to fully understand tbh).

[@thecviperspective5882]

​@@mskiptrIt's because the nasalization comes in in the second half of the vowel in Polish and Portuguese compared to French where the vowel is fully nasalized from start to end.

[@artugert]

​@@DrGeoffLindsey As for tone, my understanding is that it is a movement of the throat. I first heard this from Stuart Jay Raj, and as a speaker of Mandarin Chinese, I can verify that tones are produced in the throat. I would be very interested in seeing this explained more precisely.

[@rodrigohag]

@@thecviperspective5882 in Portuguese we have fully nasalized vowels from start to end, like ã (pão), õ (limões), an/am (andam), en/em (mentem), in/im (quindim), on (ontem), om (homem), un (unha), um (1)...

[@PC_Simo]

For me, as an associative synesthete, who always saw vowels as more colourful (and ”round”), and consonants as more black-and-white (and angular), this fills so many gaps and answers so many questions. Thank you. 😌👍🏻

[@PC_Simo]

@Winged Jupiter Thank you for your interest, and your comment 😌👍🏻. Now that I think about it, I see vocalisations as sort of less saturated pastel versions of the colours of the corresponding vowels; and, in terms of shapes, as sort of polygonal, angular shapes, where the vertices/angles have been rounded. That’s an interesting question 🤔. I’ve never really given it much thought, since my native language (Finnish) lacks vocalisations. Thank you for the question, too. 👍🏻

[@danielbrockman7402]

I think by vocalizations Winged Jupiter simply meant vowels. In other words what colors are the vowels? What was it you thought vocalizations meant (the phenomenon that Finnish lacks)?

[@PC_Simo]

@@danielbrockman7402 Phenomena like ”the dark L” (a.k.a. ”L vocalisation”), which means kind of ”fusing” the ”L”, in words, like: ”Soul”, to the preceding vowel; and thus, pronouncing the ”L”, itself, like something of a vowel. So, in general: pronouncing consonants more like vowels.

[@PC_Simo]

@akotov OK. Yeah; well, I see vocalisations and semi-vowels as more colourful, amongst the consonants, for example, ”J” (which, in Finnish, represents the palatal glide: [j]; not the English-style voiced post-alveolar affricate: [d͡ʒ]), as dark green, and the glide: [w] (which I moreso associate with the symbols: ”Ʋ”/”ʋ”, than ”W”/”w”), I view as maroon. Vowels, I see as very colourful; like ”E”, I see as yellow, ”I”, I see as green, ”A” I see as (the original, non-CMYK-)magenta. Generally, I tend to see back vowels as having cold colours, and front vowels as having warm colours. Compared to a regular old consonant; like: ”K”, which I view as, basically, ranging from black to graphite grey (pencil grey); these are definitely more colourful. Senses are, indeed, connected; and scientists pretty much agree that everyone’s a synaesthete, in the very beginning of our lives; and it’s been proven that this synaesthesia can be re-triggered via psychedelics (surprise, surprise!). Yeah; that’s, pretty much, what’s been happening. Given this new information, like new sounds and letters; I then consult my inner synaesthete, for ”his” judgement, given the input, and there’s the answer. Since I mostly *_DON’T_* physically see the letters in these colours (projective synaesthesia), it might take a bit longer for the ”answer” to pop up, and maybe require information of both the symbol and the sound it represents. However; the synaesthesia (at least, in my case) could be used to make predictions and guesses about phonetics, like the continuum-like nature of vowels being hinted at, by their ”roundness” (and, to an extent, their colourfulness). Thanks for the questions. It’s interesting, even for me, to ponder these things, consciously. 😃👍🏻

[@PC_Simo]

@akotov Also; as for diphthongs, I tend to see them as combinations of the colours of their two vowel components (and sometimes, in an intermediate colour).

[@Nikedemos]

I can see Dr Lindsey is trying to beat the record for the most dynamically changing thumbnail! We must be on version 9 by now :D

[@SiddharthS96]

Really interesting video! The comparison with colours reminded me of the terms used in Sanskrit: the vowel letters are called svara varna, which means resonant colours.

[@Rjkooljay2]

"The Vowel Space" *yawn "The Vowel Space by Dr Geoff Lindsey" *click

[@DrGeoffLindsey]

I would pin this comment but my sponsors might not like it.

[@Kris_not_Chris]

this really explains to me better why [ø] and [œ] always sounded rhoticized to me, and why [ʌ] sounds much closer to [ə] than it seems like it should

[@LjL-Videos]

In English, [ʌ] may sound very similar (or identical) to [ə] simply because it is, depending on the dialect you speak. This is covered in other videos by Lindsey.

[@indus7841]

As a math guy. I like that you're representing vowels as a point in 2d space. Very clever my friend.

[@notwithouttext]

3d if you include er and even more if you include other variations

[@mertatakan7591]

​@@notwithouttextDo you mean rhotacization?

[@notwithouttext]

@@mertatakan7591 yes, but also rounding: rounded vowels have a slightly lower f3, distinguishing them from the normal centralized vowels.

[@LeelooMinai]

It's interesting how a-u-a-i-u glide at 6:29 is interpreted by the brain (at least mine:) as "how are you" - it fills the missing sounds automatically to match the most probable phrase.

[@MrSlowrolla]

I have a degree in linguistics, majoring in phonetics (graduated 2013). I've since moved into a totally unrelated professional field, and sadly lost touch with a lot of concepts and ideas of phonetics, my formative passion. I must thank you for reigniting a nostalgic fire within me with this fantastic video! ❤️

[@vanderkarl3927]

Learning this feels a lot like when I learned that elementary school lied to me about red, blue, and yellow being the primary colors

[@taylormanning2709]

Dr. Lindsey, you may not have known this application of vowel-space, but brass musicians use this as a pedagogical tool to describe ways of modifying their tone. Unfortunately, many of them reject the science of their art, but many of us understand this is a very effective way to develop a way to communicate about an essential part of what we do. Thank you for providing a video that I can send to my musical friends; as long as their minds are open, they will benefit greatly from your work. Sincerely - thank you

[@БулатМиннуллин-р8щ]

Those "голубой" and "синий" were scarily accurate

[@SupGaillac]

It would be fun to see an animation of the great vowel shift on this diagram, with blobs representing a rough position of each vowel at a given time, slowly moving around :) Excellent video, which makes some much more sense to me than IPA diagrams on Wikipedia !

[@gameXylinder]

5:25 🤣 I burst out laughing at the "bluuu" 😂 You are both educational and hilarious!

[@cruztijerina9855]

This is mindblowing and somewhat intuitively understood by many but not articulated (clearly) by anyone until now. Thank you Dr. Lindsey !

[@amelali6299]

Could you please summarize the purpose of vowel space quality?

[@robertjenkins6132]

When I fist started learning the IPA vowel symbols a few years ago, I could only distinguish them with minimal sets or minimal pairs, e.g., golf _putt_ ([ʌ]) versus _put_ ([ʊ]), although slowly I became better at distinguishing the sounds out of context, and at understanding tongue position and rounding. The bottom and lower-right corner were particularly difficult: a, ɑ, ɒ, ɔ. The latter three are merged into one phoneme in my American accent, and the first one, [a], I think is mainly only used as first element in the personal pronoun "I" diphthong, [aɪ]. I wanted to form a minimal set that would cover all the symbols used in the transcription of English, but I think that such a set would require pronouncing some of the words with a different accent, since they are not all distinct phonemes in my accent. I think "can, Khan, con, con (British accent), co(r)n (r-dropping accent), cone" could cover [æ] though [oʊ], if the words are pronounced appropriately.

[@DrGeoffLindsey]

Ladefoged used heed, hid, had, head etc.

[@pierreabbat6157]

I have a set, which uses some words with extra sounds: Beale, bill, bale, bell, ballot, Baal, ball, boll, bowl, bull, Boole, bulk.

[@columbus8myhw]

@@DrGeoffLindsey I like b_t, as in: bat bet bit bot/bought but bait beet bite boat boot (and even Bert). I _am_ missing a 'put' vowel, and to an extend a 'bide' vowel (as I pronounce 'bite' with a noticeably different vowel).

[@zak3744]

@@columbus8myhw Would the word "kibbutz" give you the missing vowel you are after?

[@columbus8myhw]

@@zak3744 I think I've mostly heard it said with a more Israeli "kee-BOOTS", using the [u] vowel, but I could see someone with an Americanized (or Yiddishized?) pronunciation say "[kɪˈbʊt͡s]" I guess. Accent on the second syllable only though.

[@leolucas6232]

I’ve thought about this concept for so many years. The idea of a 2-dimensional space where certain points equate to different vowels. This is awesome.

[@XGD5layer]

The historic chart did that to some degree by feel, this one just maps 'f1' and 'f2' qualities (and 'f3' for the 3d version)

[@tkmmusician]

It would be very interesting to discuss these ideas as they relate to classical singing and vocal technique!

[@Fafner888]

Absolutely, it's well known that some vowels are more conducive to beautiful or resonant voice production than others (for that reason, some opera singers use what's called "covering", for example changing vowels (in the sung text) such as \a\ and \u\ into \o\). I can imagine that science can aid singers at training their voices better and improving singing techniques. Unfortunately opera pedagogy is even less consistent and scientific than phonetics...

[@caboose202ful]

@@Fafner888 I believe the 'singer's formant' is relatively well described, so I wouldn't be surprised if research related to that has also gone into covering.

[@shirou9790]

Finally I understand why e.g. [œ] and [ʌ] sound so similar to me, even though I thought they had to be drastically different according to the usual charts (the former being anterior with rounded lips, the latter being posterior with unrounded lips)!

[@timgerk3262]

6:30 As an Iowan, I feel heard. It's been a long road, expressing our all-vowel nature in four glyphs.

[@elizabethfreed472]

This remains the best explanation of vowels I've ever heard, and I got my degree in Linguistics from a respected university. I come back and rewatch every few months. Thank you, Dr Geoff!!!

[@timolson4809]

How do the nasal vowels work in this chart? I’m in an a cappella group and vowel quality is considered quite important, and I think a vowel gamut chart would be cool to have. And I’m out last rehearsal we were talking about making a vowel a little bit nasal to sound brighter at the end of a song, and I’m wondering how this would be expressed on the vowel chart?

[@matthewbartsh9167]

+1 Good question.

[@kered13]

I don't know, but I found these lecture slides that discuss the acoustics of nasal vowels: www.ling.upenn.edu/courses/Fall_2008/ling520/week8.pdf

[@vaheakli4551]

I think it would be easily represented with adding another dimension into that diagram. But being not very important for most of cases (like for colours, that are kinda better represented in 3 spaces without this added part for blue) it haven't find its way into. But sadly 4+ dimensional diagrams are not comprehensible as well as 2D ones.🧐

[@alexandrashvydun8726]

they're not represented on the vowel chart, but the IPA does have a diacritic for them en.wikipedia.org/wiki/Nasalization#:~:text=In%20the%20International%20Phonetic%20Alphabet%2C%20nasalization%20is%20indicated%20by%20printing,nasalized%20equivalent%20of%20%5Bv%5D.

[@muttlanguages3912]

A bit like putting a hat on the end of your trombone. Still the same note but sounds different, or something. Probably not the best analogy.

[@7MPhonemicEnglish]

👍🙌 I'm so glad that you're willing to turn tradition on it's ear. This was very educational. I've been thinking about the vowel space and how to identify ideal vowels that are easily differentiated from the others, such as might be done if inventing the ideal language from scratch. I never thought of comparing it to color space though. Another vowel that the traditionalists leave out of all the vowel charts in addition to the /ɹ/ is the "dark L" with it's radical tongue formation. The tongue is amazing. I had no idea what it was doing on the backside to effect sound.

[@DrGeoffLindsey]

Have you looked at my vocal organs video, where I likened the tongue to a badger?

[@7MPhonemicEnglish]

@@DrGeoffLindsey No, I'll check it out. Thanks.

[@7MPhonemicEnglish]

@@DrGeoffLindsey Done. Yes, that is an excellent video as well. MRI is a dream tool. The side angle doesn't show the tongue's ability to narrow & widen, although the narrowing may just be how it elongates. It seems to me to be a necessary aspect for the Dark L sound. Here's the video you spoke of: kzbin.info/www/bejne/iYeug2Z7iMl0oZo

[@matthewbartsh9167]

+1 " I've been thinking about the vowel space and how to identify ideal vowels that are easily differentiated from the others, such as might be done if inventing the ideal language from scratch." Wow. What an interesting idea. How about writing the vowels, or every letter, using colors?

[@7MPhonemicEnglish]

@@matthewbartsh9167 I think that methods which condense the written form of our language, reduce character count and ease of writing, are worth exploring. Changing from one colored pencil to another several times per word would be cumbersome though. Computers give us more options but we're not quite ready to give up hand-writing.

[@lock_ray]

It always felt like the way vowels are described in the IPA were a bit archaic/imprecise. It's so satisfying to see it described as a proper moduli space. I think there's probably something to be said about the topology of consonants too, even if much of it ends up being discrete, as suggested, it feels like some of the more R or L like sounds bleed together somewhat. It would be interesting to apply the same notion of formants there...

[@comradewindowsill4253]

no, you're on to something. mostly, there''s a system of fixed referents along with ways to record deviations from there, which kind of annoys me. who decided that retracted s was the weird thing and not 'normal' s? and there's different levels of retraction... obviously eventually you just get to 'sh', but it's much more a matter of degrees than is ever acknowledged.

[@digital_matt]

I saw a post somewhere that said "consonants are like a staircase, vowels are like a slide". This is the perfect expansion of that concept

[@confusedowl297]

Your reconceptualizations of phonetics concepts have helped me understand phonetics a lot better and also figure out why traditional ways of understanding it are often so confusing to me

[@travelingonline9346]

If you plot F1 and F2 on x and y axes, reverse the orientations (i.e. right to left and top to bottom) and do the plot logarithmically, then the picture looks nearly like the vowel space(s) postulated by the 19th century phoneticians. That is there is a correspondence between high and low tongue position and "high" and "low" vowel sounds, and the same is true for "front" and "back". So telling students that e.g. French [i] is higher than Englisch [i] does make sense. The point is - as you have done at the beginning of the video that vowels come in gliding ranges and that students must readjust their vowel ranges to speak another language properly.

[@ebey1525]

From one linguist to another, thank you for such a great video!

[@DrGeoffLindsey]

🙏🙏🙏

[@Jlaaag]

Incredible. Just a few days ago i was trying to come up with my own two-axis vowel map, not being satisfied with the one found on the Wikipedia page for the International Phonetic Alphabet. My best attempt was considering "up-down-ness" of the tongue and use as one axis, and "lips-with" as the other. A separate time i thought i had landed on the idea that Ah / Ee / Ooh were the three primary vowel sounds that can be mixed to form all the rest - seeing the space mapped on your graph is very satisfying to see. Now I'm considering approxmants and semi-vowels, Y and W, and wondering if they get any representation here. Awesome to see R included on this graph since I consider the sound might be deserving of vowel status.

[@matthewbartsh9167]

"Now I'm considering approxmants and semi-vowels, Y and W, and wondering if they get any representation here. " I second that.

[@QDWhite]

Finally, it makes sense! So many books I’ve read just talk about “high vowels” and “low vowels” that are simply produced “high in the mouth” or “low in the mouth” but that gave me me no intuitive understanding about how one is different from another. I feel like you’ve finally answered the question I’ve been asking for years.

[@danielbriggs991]

Every once in a blue moon, a work of genius comes out. This video among them: not only for the clarity of the theory itself, but also for the clarity and succinctness with which it's presented.

[@DrGeoffLindsey]

Thank you!

[@wingedarr0w]

Wow this is amazing, especially your clickable IPA vowel chart. I am trying to learn Hungarian from English, and plotting the vowels this way helps so much. Knowing that ű is like 'e in eat', and ő is like 'ei in eight' BUT with the tongue in a different place helps significantly, because we don't really have those sounds in English. I am also glad that you somehow decided to make videos and post them on youtube. In reality I would never get, or even be curious enough to go see a linguistics lecture, but on here I can learn about such an interesting topic basically in 20 minutes on the other side of the world.

[@alexandrashvydun8726]

10:45 i love how going from /i/ to /u/ accentuates the natural overtones

[@falquicao8331]

7:40 Small correction: only some dialects of Italian distinguish between open o and closed o. In other dialects, especially in the South, that difference disappears. I remember years ago hearing a debate between Northeners over the correct vowels to use in "loro" and "l'oro" which, to my southerner ear at the time, sounded exactly the same.

[@kellensanna]

I always found the vowel chart taught in school a bit strange. This explains why. Thank you and i really hope to see a fleshed out chart with three formant vowel spaces

[@AntipodePolyglot]

Your pronunciation of the words "синий" and "голубой" was flawless! I run a KZbin channel about phonology in Russian and also another channel about linguistics in English.

[@Tynach]

I'd just like to correct one minor point about the chromaticity diagram: That's an 'xy' chromaticity diagram, based on the XYZ color matching functions. XYZ was first devised before they could determine exactly what the response curves for the three cone types were, so they ended up with an approximation that turns out to be a *linear combination* of the actual three cone cells' response curves. Because the 'Y' curve was calculated to map out how much each wavelength contributes to the perception of luminosity, XYZ is still commonly used because its analog for 'green' works very well as a quick way to calculate 'brightness' in linear space. However, if you want a chromaticity chart that has a nice 'U' shape that's not at a weird 45° angle, one can convert XYZ's color matching functions to the LMS space (which stands for 'Long, Medium, Short', and REALLY represents the amount each type of cone responds to each wavelengths) first, and then make an 'lm' chromaticity diagram. It's also worth noting that your three absolute primaries are Always going to lie outside of the range of physically possible values, because they absolutely Must contain all possible answers within the triangle they create. For example, in an 'xy' chromaticity diagram, the 'horseshoe' shape's boundary contains individual wavelengths, but all wavelengths are within x and y coordinates between 0 and 1, and the three primaries are located at (0, 0), (1, 0), and (0, 1). As a humorous aside, the actual chromaticity coordinates for the LMS space's primaries (as defined by CVRL data that forms the CIE's LMS standard), when charted on an xy chromaticity diagram instead of an lm chromaticity diagram, are (0.73840145, 0.26159855), (1.32671635, -0.32671635), and (0.15861916, 0.0) for L, M, and S, respectively. The presence of coordinates greater than 1.0, and negative coordinates, shows just how far off the XYZ space is from matching the cones' own actual spectral responses.

[@Tynach]

All of that said, I've been wondering about how to map out vowels in my head for YEARS, and been dissatisfied and confused by all the diagrams and whatnot I've seen until now. Having an understanding of colorspaces and chromaticity diagrams and whatnot, and having a 'vowel space' explained to me in a way that directly relates to that, is FANTASTIC. I don't know if it'd be helpful for most people, but it was certainly helpful to me! So thank you SO much for making this video.

[@DrGeoffLindsey]

Am I right that you're talking about the kind of diagram here? jila.colorado.edu/~ajsh/colour/primary.html I used the CIE because it's easy to find and widely referenced on the internet. Not that I really know the details!

[@Tynach]

​@@DrGeoffLindsey Well jeez, I had planned to link to that exact page, but remembered that the last time I tried to link to an external site in a comment my comment would get deleted almost immediately, so decided not to. That's exactly the sort of thing I mean, yes! I'm amazed you found (or already knew about) that page. I ran into it only because I ran into that site when looking up stuff about black holes and worm holes, and that was several years ago. Though, in my own renderings on Shadertoy (I tried to link them just now, and the post got deleted; I think KZbin considers external links to be spam unless you're the uploader), I render it with a D65 whitepoint to stay true to the sRGB colorspace of the output. It's worth noting that the author of that page had trouble with the white point, but only because he didn't rescale the LMS values to treat (1, 1, 1) as the equal energy white point.. Though, to be fair, this DOES make the colors a little more balanced, and the 'U' shape less stretched. So this is merely personal preference. One thing that page does get wrong, is what it says about color blindness. The reason that protanopes don't see things in psychedelic aquamarine is because they aren't missing L ('red') cones; rather, their L cones have mutated to behave like M ('green') cones. The result is that they receive the same signal from both types of cones, rather than one being completely absent. However, this is a common misconception about color blindness, and it's actually a handy misconception when it comes to working out how to simulate color blindness, so it's not that big of a deal.

[@Tynach]

@@DrGeoffLindsey Oh yeah, one more little tidbit: CIE actually has two different standards for chromaticity charts. The other one is the "CIE u'v' Uniform Chromaticity Scale", and is produced by first performing some weirdly formed stretching and scaling operations to XYZ. It's meant to be more 'perceptually uniform'. Interestingly enough, while it too is rotated at about 45° (much like the xy chromaticity diagram), the resulting shape more closely resembles an lm diagram. I don't know if this is coincidence or not.

[@eljestLiv]

This video is fantastic. I never was able to figure out the four-sided vowel diagram, it always made more sense in my head as a triangle shape. Thanks for reassuring me that I had the right idea!

[@CrysolasChymera2117]

As a lover of languages in general, that understands the complexities of phonology and grammar not just in English but in other languages as well, and that is interested in conlanging as well: I LOVE THIS KIND OF CONTENT! Edit: Your pronunciation of other languages sounds amazing :)

[@DrGeoffLindsey]

Thank you!

[@Skiddla]

wow, it would be amazing when learning a new language to be able to speak into a microphone and get feedback on the vowel sound frequency you make vs what your target frequency is. this would also be amazing for impressions of certain accents. powerful stuff

[@whycantiremainanonymous8091]

6:30: I'm fine, thank you. And how are you?

[@twodollars4u]

Great, well written video on a very interesting subject. I remember reading about IPA's use of tongue position as the defining characteristic of vowel sounds and being confused that it didn't sound accurate. The formant system seems like a better way of explaining.

[@ScarfmonsterWR]

This would've been so useful when I was taught English pronunciation. The tongue charts just don't make any sense to me, and honestly every time I was told to pronounce the vowel as the chart says, it was worse than however I pronounced it before. The tongue chart also makes understanding how vowels relate between different languages much harder than it needs to be.

[@EmblemParade]

Fascinating. My assumption has always been that every speaker figures out their own tongue-and-lip tactics to make sounds, which is why we all sound a bit different despite sharing a recongizable pronunciation type. Children are not taught how to talk, they do it through imitation and trial-and-error, so it seems inevitable to me that we must be somewhat idiosyncratic in our approach. I also don't think phonotactics can be ignored in any analysis, as glides and clusters provide the context for individual vowel sounds.

[@teshn1229]

I find it fascinating that we become attuned to, differentiate between and learn to reproduce different sounds with very similar parameters in English (and likewise for native speakers of other languages). I don't know if the research has been updated, but one example is voiced stops taking less than 30ms for the voicing to start, and voiceless stops taking greater than 30ms. (Kessinger & Blumstein, (1997), in Ballard, Maas & Robin, (2007)) (Forgive the (old) referencing 😅)

[@teshn1229]

In comment on your comment, the video is looking at vowels in isolation, in a 'pure' sense, although it could be useful to map vowels in a variety of consonant contexts. Although I've also been taught that sounds can impact more than the sounds that immediately surround them.

[@teshn1229]

@@skyworm8006 from what I understand yes. I probably shouldn't have put the parentheses.

[@comradewindowsill4253]

@@skyworm8006 yeah, there's plenty of vowel-only words in a lot of languages. for example, russian /i/, means 'and'. and there's a few others, but those I'd need ipa symbols for. there's another 'and', for when there's a change of subject involved, then there's 'about'... these two have very similar but distinguishable A-like sounds. I'm sure that there's languages with longer words which are fully comprised of vowels, too.

[@yura2424]

This channel is gold!

[@resourceress7]

I know that eventually you will upload human-edited captions. But I have to tell you that the auto captions on the computerized vowel simulation are golden. "hi" "hey" "oh" "wait" "Oh yeah" "Oh yeah" "How are you" "How are you"

[@notwithouttext]

timestamp: 6:16 yeah they are great

[@SillyWillyFan47]

About 10 years ago I invented a colour-coded Scrabble-like word game using sounds (colours) rather then alphabet letters. And I'm astonished now (2024) to see that I chose the colours (by myself) exactly along the lines of 3:13 spectrum graph.

[@kelcben]

Thanks! Love your expertise, rich and delightful videos, and the sense of humor threading through them. I've been fascinated by all this all my life (Enfield, UK, war bride mother, Newark, NJ, father). Although it's not a phonetic issue, I've often thought to ask you your view on double-is. Today I was reading "SMART Speech" and there it was (I admit I was dismayed to see). Page 42: "A corollary of this is, as I discussed in Chapter Two, is the expectation that..." It has always irked me a little and I suppose it will now do so a bit less. Congratulations and best wishes for your continued success.

[@arcuscotangens]

Thank you so much for this video. It will be very helpful once my son starts talking. I can't wait to teach him his formants.

[@sahidcm]

I'm a English teacher and this is really useful to me. I intend to reference it for when my students want to know more about vowel distribution in the charts, and understand vowel differentiation better. Brilliant! Thanks Dr. Geoff!!

[@amj.composer]

Your content is absolutely insane [Just realized my comment may be ambiguous, I mean that in a good way]

[@DrGeoffLindsey]

Insane comment, thanks!

[@AirKIng74]

Every time I see a new video by you I feel like I walk away with a new and more profound understanding of the way phonetics works. I clicked on this video the second I saw it and I'm glad I did, this will be extremely useful for the conlang I'm working on.

[@averynhiell]

This is so interesting, thank you! I have vowel-colour synaesthesia and the vowels are actually on a colour spectrum for me: [i] is red, [e] yellow, [ɛ] light yellow, [a] white, [ɑ] light grey, [ɔ] grey, [o] black, [u] dark blue, [y] purple. I also noticed that a great part of my vowel-colour associations are the same as shown at 6:55.

[@DrGeoffLindsey]

I have a kind of synaesthesia between vowel *letters* and colours. For me, A/a is red, I/i is black, etc.

[@resourceress7]

This is much clearer than the phonetics textbooks or explanations I ever received in linguistics classes. Thanks!

[@anj000]

You can find color palettes made on the basis of some particular image or series of images. Like it takes colors present on the image and compiles a set of for example 6 main colors present, or a set of colors scaled proportionally to amount of particular color on the image. I would love, LOVE, to see something like this but for languages. Take a language, make a palette of this language in vowel space and then convert it to color space. Some languages would be more red than others, others would be more saturated than other. I think it would be really cool.

[@Bebeu4300]

I really think some of your approaches are much better than the normally used ones. Everything from an objective, data-based vowel chart to your more accurate list of english phonemes. I wish more people would embrace these, as a lot of things feel outdated, and as if you have to learn that it isn't what it seems like because the same system has been used for too long and is no longer accurate, or better technologies have come along.

[@teshn1229]

And surely it would be much easier to update the IPA than something like the English spelling system 😂

[@numbersix8919]

Actually AO is still used in Japan for the color green, especially for vegetation, although it actually means the color blue. People understand which according to context. MIDORI or GURIN ("green") are always perfectly understandable, but not always in the vernacular. Bonus tip: MURASAKI (purple) is slang for soy sauce (SHOYU) in sushi chef slang.

[@yura2424]

9:41 This is a very useful demonstration

[@hoi-polloi1863]

Hey Doc! I really like your vowel charts, your approach makes sense. I'd recommend you chart the vowel inventories of various languages on your chart and look at how the spacings fall out. I'm reminded of how languages with only three vowels tend to pick a,i,u at the very corners of the chart.

[@PC_Simo]

Yes; like Inuktitut. I guess they strive for maximal clarity 🤔.

[@bten-pow2095]

Very informative and eye opening - thank you

[@telescoper]

Wow! In 20 minutes, for the first time ever, I understand vowels AND color space! Brilliant!

[@shmoola]

Wow, thank you! I always thought it was a 3 dimensional space: 2 from tongue and 1 from lips. Having the chart displayed on the space with 2 frequencies makes a lot of sense.

[@aramiscalcutt]

It seems to me that even though it might not be scientifically/anatomically accurate, the old-fashioned rhomboidal chart is still useful, especially when teaching basic phonology to beginners because it makes room for clear distinctions among the low vowels (æ, a, ä/ɑ̈, ɐ, ɑ, ɒ, ʌ), which is usually the most difficult part of elementary vowel transcription for a layperson in my experience. The more accurate diagram doesn’t make it easy to explain the differences. (It also doesn’t help that discussions about phonology almost never distinguish between a and ä).

[@carlotapuig]

In 20min I have learnt more about vowels than in 5 years at uni. I'm not joking.

[@lucasteganox9049]

From a language learning standpoint, this is very useful. Although some things about Vowel Production are leaved out the table(in the basic form of the chart I mean), it's way more intuitive to give yourself an idea about how to produce a certain Vowel. Pd: For the sake of representing "F3" itself, a third dimension might not be necessary. You could color the symbols in the chart based on a color scheme that represents "F3". For example: blue means low F3 while grean means high F3. It might be useful to do that to keep the handiness of 2d representations without leaving F3 unrepresented. Truly, thanks for unconstraining our mind with new ideas and giving us power with new tools, in such an accessible way ☺️

[@stephanberger3476]

We are working on a new course book for non-Frisians to learn our language. We want to have some IPA in it, mainly because of the difference between two 'oh-sounds' where Dutch just has one. Your page will be of great help, thanks so much!

[@billywydola3388]

I studied phonetics independently as research for building a fantasy alphabet for a fantasy world I am building and I'm glad I found this video after doing my own research but it affirms most of the decisions I made when grappling with some of the inconsistencies you spoke of. Specifically, your point about rounded and unrounded vowels being mingled at times in some of the formerly established systems. I found it convenient to isolate the approximates made when (in English) the consonants "Y", "W", and "R" interact with vowels, although I had to make "R" a special case as it seems to be found on a range that could stretch from Y->R->L and gets fuzzier as it goes on, so I stopped it at "R" and "L", but "R" has a lot more vowel interaction. "W" and "Y" became reduced to accents to be place atop a vowel to mark a consonant emphasis going into the vowel, such as in the words "yes" or "wet", or to be placed below a vowel to change the sound of the vowel in "span" to "spine" (approximately! for the lower "Y" accent) or "caw" to "cow" (approximately! for the lower "W" accent). For a word like "lower", one would make a determination on whether the "W" accent, should be below the "o", above the "e", or both in some cases. Ultimately, this worked great because I need no convention for combined and inconsistent vowel sounds as in a word like "read", it would be the approximate for "R", followed by the vowel in "it" with a lower "Y" accent, followed by the stop for "D", for "read" as in the present tense, or for the past tense, pronounced like the color "red", it would be the approximate for "R" followed by the vowel in "red" followed by the stop "D". In other words, I only need short vowel sounds and all diphthongs are accents or an appropriate number of vowel symbols.

[@poahnfiap]

I'm curious if this vowel space covers all values in every language. For example I speak Portuguese which has nasal vowels- is nasality yet another dimension in that space or just a binary modifier?

[@DrGeoffLindsey]

Great question, and not easy to answer simply (at least not by me). Nasalization affects not only resonant frequencies, but also the shape of the spectrum (roughly = the loudness of the different resonances) and the bandwidth of the resonances. Opening up the nasal cavity has the effect of damping (roughly, muffling) the resonances. Of course there were limits on what I thought I could cover here.

[@Unbrutal_Rawr]

Nasality is another articulatory-based descriptor, it means some of the airflow is directed through the nasal cavity by lowering the soft palate. Since the degree and timing of this opening can be modulated, nasality is not a binary modifier at all, and phoneticians sometimes talk about the former or the latter part of the vowel being nasalised, and degrees of nasality differing as well. Nasalising a vowel doesn't have a single fixed effect on the vowel, but modifies existing formant frequencies and adds new peaks, chaning the acoustic spectrum. The most noticeable effect is that nasalised vowels are acoustically centralised, especially on the vertical F1 axis.

[@poahnfiap]

Fascinating! Would love a video covering this more in depth. I'm sure like there are many rabbit holes in the vowel space you could delve into. Thanks for a great intro!

[@Jack_Simpson]

To my understanding, the first vowel formant is the main factor is coloring vocal registers when singing. [u] and [i] turn over low and [a] turns over high. Matching pitch with that formant is how you get the sopranoey, operatic sound. That’s why sometimes their vowels sound strange.

[@vladimir520]

Thank you SO much for shedding light on the topic of vowels, your video could not have come at a better time for me! I've been stuck for a couple weeks now on the first chapter of Ladegofed's Course in Phonetics, because I thought I didn't really get vowels and nothing sounded quite right for me, but now, seeing how difficult a problem vowel recognition is even for phoneticians, I feel like I can relax now and try to gain a better understanding through the use of modern technology, spectrograms and the like. Thank you again for your concise, documented and entertaining explanations on phonetics, your videos have been helping me a lot as of late!

[@andrewanderson6121]

Sometimes it seems like I can taste words which makes sense considering the physical activities and their location in the vocal apparatus. Geoff always sets me thinking. That you Geoff!

[@elizabethfreed472]

This video completely reshaped my understanding of phonetics. I studied linguistics at uni (about a decade ago now), and was taught the tongue-position understanding of vowels. I could never seem to feel the "tongue high point" that others did, and ultimately vowels made little sense to me due to their continuum-like nature. Watching this video, I "got" vowels for the first time! The color chart, the frequencies, the "tubes" of throat and mouth, the interplay between i/u/a at the extremes/corners... Even the retroflex "ɚ"with its lowered frequency. It finally made sense. It's really like you said with pre-Galilean astronomers: observations were good, they just didn't have the right lens of understanding why. I have watched your videos for a long time, Dr Lindsey, and this one really helps me. Thank you so much for offering well-founded, easily-accessible linguistic education!

[@kmanz60]

I would pay to watch you interview an impressionist in a podcast format… super interesting stuff!

[@elmukaltino]

Dr Lindsey, I discovered your channel couple of months ago and I’m hooked. You have such a gift of explaining phonetics in such an engaging and visual way! Definitely the best channel on the topic on KZbin. Every analogy you draw is just spot on! Watching this video I’m just astounded about the color space/vowel chart similarity that has never crossed my mind before even though I’m more than familiar with both concepts. Oh and your Russian pronunciation was flawless! Best of luck to you and your channel!

[@JonathanSharman]

6:29 I'm fine, thanks for asking!

[@nurtiny1024]

The vowel of upside down “m” is similar to the “eu” sound, 으 in Korean (Also present in Amharic and other languages) it’s when I was first introduced to the specific letter They also have 의 which is pronounced like the mix between u and i, or pronounced like “eh”

[@notwithouttext]

yeah it is, it's [u] but unrounded, and on his video on bts he shows how koreans can have a hard time saying [w] after a consonant becaue of this

[@aaronk5405]

A collaboration with Ken Bozeman on Vocal Acoustics would be incredible! The relationship between pitch and vowel choice, how many singers alter vowels from their speech for aesthetic reasons (I'm looking at you, "r" vowel), or how many singers delay a diphthong's second vowel for the resonance that a higher 1st Formant would have, would be an amazing resource. Thanks for all your content!

[@StefanRotenberg]

Dr Geoff, you are the best. Seriously, you have taught me so much, and it feels so much more "scientific" when you say it than most of the instructions I've received or read in other places. Seriously, listening to you is like reading fine, clear prose.

[@DrGeoffLindsey]

Thanks so much, Stefan!

[@drayyoutube5843]

5:27 “And BLOO...” That part honestly makes me laugh out loud.

[@drdca8263]

This is great! It reminds me of a question I have: If we used speech synthesis to make something that sounds like a vowel but where the 1st and 2nd formants are in a combination outside what can be comfortably made with one’s mouth, how weird would that sound?

[@notwithouttext]

new alien conlang phonology idea

[@comradewindowsill4253]

@@notwithouttext yeah, what sorts of formants would a creature with a vocal tract more like a bird's sound like? or just one with a much longer mouth or throat, even?

[@CuriosityCore101]

I've recently started trying to learn the IPA and this makes so much more sense to me than the way the IPA explains vowels. Thank you!

[@eliotbaez]

Thank you for this video! I've been working on a voice synthesizer project and have been stuck looking for a way to understand how vowels are formed. This was a huge eye-opener in terms of a starting point for further research. The two-cylinders approximation of the vocal tract was especially fascinating. I'm definitely embarking on a rabbit hole of research after this.

[@mustardsfire22]

I think it's really cool how you always seem perfectly willing to buck the established trends in favor of more practical and widely useful techniques and characterizations, Dr. Lindsey. I think it feeds our natural curiosities as humans to try to figure out something new. We're easily bored which might be why most education systems bore students. We want to feel like we're figuring something out rather than walking the same road. I think your videos do an excellent job of showing where we are, presenting your own observations and experiments, and suggesting paths for future research by those more able or willing. Brava.

[@cmyk8964]

I still find the vowel trapezoid useful in the same way as TAB score on a guitar sheet. It doesn’t represent the resulting formants, but it helps you learn what configuration of the noisemaking apparatus should make the correct formants.

[@whatisthisayoutubechannel]

Yeah, it’s not like we can manually adjust formants when we speak, so the traditional chart is better at helping production (and IMO easier for beginners to understand)

[@ashedtogether]

Fun fact: The so called 'red receptor' in the eye actually peaks in the yellows, perhaps a tiny bit on the green side (~575nm) and it's response curve is pretty close to the green receptor's (peaks ~535nm). There's a lot of overlap with in the two receptors curves and both receptors respond to light from red through to cyan. Seeing red to green colours requires precisely comparing the level of response by both receptors. I guess this is why red-green colour-blindness is so much more common than other forms - if either receptor was responding in any way inconsistently the brain wouldn't be able to do that determination.

[@jeff__w]

7:58 “…but it can be more challenging to hear vowel contrasts that your native language lacks.” My friend, a Korean, insists on pronouncing _beach_ as the exaggerated “beeeeach” to make sure he distinguishes between the “long e” and “short i” sounds. “It’s _okay,”_ I tell him, “you can shorten up that vowel length just a bit,” but he still persists. Too risky, I suppose. (Or he’s just trying to annoy me. 😂)

[@_volder]

At first, I misread this new vowel graph because I didn't realize that it was losing one of the three parameters of information by putting everything in just two axes of the graph. It looked at first like the horizontal axis represented where the vocal tract is the most constricted: tongue on the left, lips on the right, both or neither in the middle. But when I looked at it more, especially the middle area, I realized that: ►Lip-rounding actually flips back & forth along the horizontal axis, implying that if you smoothly increase F2 you get alternating rounded & unrounded sounds. ►The horizontal axis also runs between front and back vowels, so this one axis has two different real parameters to try to represent at once, implying that those are connected. ►The graph also has a rounded-unrounded pair separated vertically instead of horizontally (schwa & rounded schwa), implying that roundedness/unroundedness is connected to F1. In the Jones & Wells F1/F2 graphs that inspired this one, rounded and unrounded vowels are scattered among each other randomly, but the statistically small number of them and the process of spreading them out evenly by somebody familiar with the old vowel map resulted in an illusion of a correlation between roundedness and either F1 or F2 in different cases (more often F2). Lip-rounding is definitely a real thing that's needed to properly describe vowels, but it's not correlated with F1 or F2, even though it got graphed as if it were. A graph which only uses two parameters to convey information just can't accurately handle three parameters of actual information. A third thing is needed to specifically represent lip roundedness & unroundedness. But it doesn't need to be a third geometric dimension. On the old vowel graph, that was done by putting each pair right next to each other with a dot between them to show that they both belong on that same point in the grid, which is what this graph undoes by spreading each pair apart. If we graphed the vowel space as areas with borders to separate them instead of dots with lines connecting them, then each area could just as well have two symbols in it for a rounded-unrounded pair instead of a single symbol. Another option would be to make two separate F1/F2 graphs and put them side by side: one for rounded vowels and one for unrounded vowels. Even different colors would work if you could count on the thing always being displayed in the right colors. But the third real-world parameter just can't be left unaccounted-for without making it look as if it were somehow based on the other two (the horizontal & vertical axes).

[@Unbrutal_Rawr]

Your old, misread understanding that you've unhelpfully spammed under many comments to this video makes no sense to anybody familiar with vowel articulation. But in this comment you show that you still don't understand what this video is about. The video proposes that the horizontal axis accounts for both tongue advancement and lip rounding, and that the parameter that both of these contribute to is the length of the vocal tract in the horizontal dimension, increasing which results in a lower F2 frequency. Lip rounding adds length to the vocal tract when you can't move the tongue any further back. This is why any unrounded vowel is shown further to the left than any rounded vowel - the horizontal length is shorter, and its F2 higher. No pair of rounded-unrounded vowels can occupy the same space on the diagram because adding lip rounding in fact increases the length of the tract, which results in a lower F2. If F2 is identical for two vowels differing in rounding, then tongue position must necessarily differ. You seem to be someone who's so absorbed in their own thoughts, and so preoccupied with sharing them, that they can't properly hear other people. I would suggest you focus on understanding other people and avoid getting stuck in your own thought process or sharing its results at every opportunity. Because as it stands, you seem to have a lot of trouble meaningfully engaging with other people's thoughts.

[@ferociousfeind8538]

5:25 petition to swap the red and green response axes, so we can call it "green" and then something like "rad"

[@DrGeoffLindsey]

The Japanese for red is "aka"! I was going to put it in the video, but I need to be succinct and meet deadlines...

[@موسى_7]

This may be your best video yet. Vowels aren't confusing me anymore.

[@Horhne]

What a fantastically clear explanation! I consider myself a linguist and love to delve into new languages but constantly come up against the vowel sounds of a given language. I have tried researching linguistics but always found the charts to be incomplete leaving me with more questions than answers. I am delighted that my frustrations had some genuine basis; and now I have a greater insight into this area of linguistics. Thank you so much. By the way I am grappling with learning European Portuguese at the moment where the vowel sounds are very 'fluid' soo can relate very easily to your demonstration near the start of the video.

[@altitudinem6038]

This is very interesting! I’ve always had this almost synesthesia-like relationship with vowels and colors. /ɑ/ is red, /i/ is yellow, and the /ε/ between them feels orange!

[@matthewbartsh9167]

Awesome vid. One of Dr. Lindsey's most meaty vids ever. There is *so* much here. Some of the technology referred to could be very useful to me in my hobby of coding educational audio files. I'm not very good at coding, so I might need help from a coder, or from AI. By a coincidence I have long been interested in both phonetics and the color sphere and so on. I never thought they could be linked, though. Liked and tweeted out, and I am going to watch this again, many times. I think I'll share it a couple more times, too. Thank you so much, Doctor.

[@MogaTange]

We should just replace all the colours with bl and their corresponding vowel. White? No, bluh. Blue? No, blu. Cyan? Blo.

[@erikesswein4608]

This is something i noticed with throat signing, where you have to generate the overtone frequencies by gliding between vowels