The CBB

So in case you haven't heard the phonological equivalent of the blue vs. gold dress:
https://www.youtube.com/watch?v=SbjnIK6VEjc
The explanation they give is that "psychologists" say that older people hear [lɑʊrl̩] and younger people [jɑniɪ] due to damage to the ear canal. I'm 24 and hear [lɑʊrl̩], but that's anecdotal. Other "scientific" explanations have to do with the pitch interfering with phone perceptions in some people. I wasn't sure that this was the answer, and I created a "Non-English Phone Hypothesis"- that it involves non-English sounds. This is based on a ten year old meme, called “Buffalaxing”, where people would take songs in other languages and subtitle them with what they heard them as in English. One heard the Hindi string [mukəla mukəbəla mukəbəla sʊbʱan allah] as “Boot the log, look at my log, look at my log, suck on my log” and another as “Hook the law, Look out! The law! Look out! The law! f*** up the law” because it involved non-English sounds. The same could be happening here. The test of the pitch hypothesis or mine is what do people with a native language other than English hear? I decided that even though I'm an undergrad, with no Liguistics-related Degree, I could test it scientifically. So, my Structure of an East Asian Language (Korean) professor heard [i:.je.jo i.jek̚̚] and a Chinese Uber driver I was taking as /i:.jai/. Now this is hardly a significant sample size, but I certainly was able to replicate (even only twice) that non-English speakers hear neither [lɑʊrl̩] or [jɑniɪ] or something even close to the two Native English-speaker options. I'm chalking this to "needs more research".

Here's a link with a slider that helps you hear both:
https://www.nytimes.com/interactive/201 ... ebate.html

Another factor that affects perception here, other than age, is the device used. A lot of people have reported differences when they try with different devices (e.g. speakers vs headphones, computer vs phone, etc.).

It seems that the sound bite is taken from a readout of the word "laurel" on a dictionary site, so "laurel" is the original word. Then, the clip was modified in such a way that it can be heard either way by different people. For reference, I took a plot of the spectrum of the originaly dictionary clip (top) and the ambiguous clip (bottom):

http://www.pabappa.com/pics/2laurels-b.png

What surprised me was that it's the people with better hearing that get the wrong word out of this ... apparentrly our brains are tuned to pick out just one interpretation of a sound when we perceive it as coming from one source ... and the "yanny" part is in the high tones, which are out of the range of normal human speech. But even the "yanny" part is just a pitch-shifted duplicate of the "laurel" clip, as can be seen by cutting out all of the low tones from either of the t wo source clips.

I havent studied phonetics but I know that even ordinary speech is composed of several different sounds concurrently, which are called formants. Somehow the "yanny" sounds override the "laurel" sounds for those of us with intact hearing.

I hear "laurel" until about -12% pitch shift, at which point it abruptly changes to "yanny". Cropping out the upper frequency bands (above 2400 Hz) also changes it to "yanny". There seems to be very little middle ground, even for the vowels ... as above, the brain seems determined to pick on one and only one possible interpretation of a speech sound.

Pabappa wrote: ↑17 May 2018 02:11 It seems that the sound bite is taken from a readout of the word "laurel" on a dictionary site, so "laurel" is the original word. Then, the clip was modified in such a way that it can be heard either way by different people. For reference, I took a plot of the spectrum of the originaly dictionary clip (top) and the ambiguous clip (bottom):

Right, so I'm proposing that part of the ambiguity is that the modification included the introduction of formants that make it sound like non-English phones are being produced, and thus non-native speakers hear neither "laurel" or "Yanni", but forms like [i:.je.jo i.jek̚̚] and [i:.jai]. Perhaps the /l/ was modified to resemble more an [lʲ] or [ʎ]. Thus, saying its the latter, native English speakers interpret [ʎ ʎ̩] as either [l l̩] or [j i].

Listening to the CBS clip, I hear both, but I heard "yanny" first.

Shemtov wrote: ↑17 May 2018 02:35

Pabappa wrote: ↑17 May 2018 02:11 It seems that the sound bite is taken from a readout of the word "laurel" on a dictionary site, so "laurel" is the original word. Then, the clip was modified in such a way that it can be heard either way by different people. For reference, I took a plot of the spectrum of the originaly dictionary clip (top) and the ambiguous clip (bottom):

Right, so I'm proposing that part of the ambiguity is that the modification included the introduction of formants that make it sound like non-English phones are being produced, and thus non-native speakers hear neither "laurel" or "Yanni", but forms like [i:.je.jo i.jek̚̚] and [i:.jai]. Perhaps the /l/ was modified to resemble more an [lʲ] or [ʎ]. Thus, saying its the latter, native English speakers interpret [ʎ ʎ̩] as either [l l̩] or [j i].

I retract the "Buffalax Hypothesis" based on more evidence; it seems I was beginning to veer into "Pathological Science", just like René Blondlot did in physics or Sapir and Whorf did in the very same science I was doing. I had an unrecognized hubris: a. I knew about the Buffalaxing meme, I remembered the meme, I thought it was beyond the scientific community's field of view, much less to make the connection b. The original research and Hypotheses were done by "Psychologists commenting on Psychophysics " and "Acousticians commenting on phonetics"; obviously Psycholinguists and Psychophysicists needed to comment, and surely "Native Language" would come up. In fact, I see a bit of Sapir-Whorfian thinking in the "Buffalax Hypothesis", though an extension of "Weak Sapir-Whorf"

Shemtov wrote: ↑17 May 2018 06:16

I retract the "Buffalax Hypothesis" based on more evidence; it seems I was beginning to veer into "Pathological Science", just like René Blondlot did in physics or Sapir and Whorf did in the very same science I was doing. I had an unrecognized hubris: a. I knew about the Buffalaxing meme, I remembered the meme, I thought it was beyond the scientific community's field of view, much less to make the connection b. The original research and Hypotheses were done by "Psychologists commenting on Psychophysics " and "Acousticians commenting on phonetics"; obviously Psycholinguists and Psychophysicists needed to comment, and surely "Native Language" would come up. In fact, I see a bit of Sapir-Whorfian thinking in the "Buffalax Hypothesis", though an extension of "Weak Sapir-Whorf"

I actually retract my retraction to a degree, and propose a "Weak Buffalax Hypothesis" based on this article's https://www.livescience.com/62583-yanny ... ained.html ending

As for what makes a person sway one way or the other after listening to this audio clip, that's anyone's guess for now.

"I'm not sure that anyone knows why some people hear it one way and other people hear it another way, but that's often the way with these visual and auditory illusions — our brains 'fill in' missing information, and how that happens seems to vary a lot from one person to the next," Oxenham said.

Bharath Chandrasekaran, an associate professor in the Department of Communication Sciences and Disorders at the University of Texas at Austin, said he doesn't know either, but he's planning to find out. He told The Verge that he is going to look for volunteers in both camps and then run tests in which he looks at their brain waves while they listen to the audio clip.

The "Weak Buffalax Hypotheses" is that native English-speakers' perception may be based partially on level of exposure to foreign languages, though not necessarily full bilingualism, and your native dialect of English may play a minor part.

So, what about those of us that just heard the horrible conflation of irritating computer quasispeech?

Except at the far ends of that slider thing, i just hear a kind of garmble, like there's two words overlapping.

Anybody else hear "yaMMy" rather than "yaNNy"? Sweetie pie says she listened and had no idea where people get this "yanny" business from.

elemtilas wrote: ↑17 May 2018 19:24 Anybody else hear "yaMMy" rather than "yaNNy"? Sweetie pie says she listened and had no idea where people get this "yanny" business from.

On one pitch change I'd heard "yannel" and after that at the same pitch I heard Laurel *shrugs*

The sequel has arrived:

https://twitter.com/GeorgeAylett/status ... 0142903300

This is a retweet of a retweet of a tweet of a Reddit post, but they are all of the same video and it's definitely real and not a hoax as some commenters have wondered. Confirmed by viewing the waveform as I did with the other clip.

My take on this one is that it works because the sounds towards the end are highly ambiguous ... the /st/ in "storm" is ambiguous with the /i/ of "needle" ... (in isolating it really just sounds like two metal pipes clanging...) , but once your mind has determined whether it's a vowel or a consonant youre hearing , the rest of the word is determined. this is why it's difficult to hear, for example "greeneeorm" or "brainstdle".

In my physics class, my teacher played a series of tones, each one increasing in frequency until he noted the point at which he could no longer hear them (he was in his late 40s). Of course we all laughed because we could hear the high-pitched tones easily.

I hear "laurel". Like, I don't even know how to begin to fathom how one can hear "yanny". With the dress, I could at least fathom the other side even though one perception came to me more obviously (for example, when I saw the brightness and contrast of the dress played with, I could see both). This on the other hand is plain unfathomable I'm 17 and I can hear all kinds of sounds my parents can't hear, but I guess in this case my hearing is sub-par (I'll try it through my hi-fi equipment and see if I can hear "yanny"---through my phone speakers and earbuds I can only hear "laurel").

Maybe differences in perception like this can explain certain anomalous sound changes that would seem to violate sound laws. Maybe some day there will be a dialect of English where the word for laurel is "yonnie". That might have been unexplainable until this showed up.

I hear "yanny" and nothing but. I have an equalizer in chrome and tested some hypotheses out. When I push the equalizer to maximum at 500 Hz or 16000 Hz and set everything else to 0db, "laurel" is the most prominent sound, but yanny is barely audible behind it.
Setting the equalizer to have 125 Hz and 250 Hz to maximum and everything below results in a state where I can actively switch.

Setting 250, 500, 1000, and 2000 Hz to max and everything below results in a "laurel" state, but the opposite results in a "yanny" state.
Setting 32,64,125 Hz to max gives me a switch state.

I think "Laurel" is the intended result, but due to processing errors, the sound has distortion in non-modal (modal = 250~1000 Hz) states, causing a "yanny" sound.

This reminds me of that optical illusion where you can see a dancer spinning either clockwise or counterclockwise and at any time your brain may "switch" the dancer's direction:

https://upload.wikimedia.org/wikipedia/ ... Dancer.gif

Tried it with headphones and my high-end amp. I still hear "laurel" by default, however when using the slider, if I go from "laurel" to "yanny", I hear "laurel" until almost halfway between the middle and the right end. At that point I can hear both; beyond that point I hear only "yanny". Then when I slide back toward the middle, I hear "yanny" almost until right at the middle (so the range between the middle and halfway to the right I will hear as either one depending on which direction I'm moving the slider). And if I listen for "yanny" in the original clip, I can sort of hear the "undertones" of it but I still cannot hear the word as anything other than "laurel". Interesting stuff :)

I'm 38, native Anglophone, and I heard [jæni]. Yes, with an [æ].

But the YouTube video turned out to be on CBS. Ewwwww!

elemtilas wrote: ↑17 May 2018 19:24 Anybody else hear "yaMMy" rather than "yaNNy"? Sweetie pie says she listened and had no idea where people get this "yanny" business from.

Not me, but I've seen 2 other people say this. It may be as much a difference in how different consonants affect neighboring vowels than a difference in the perception of the consonants themselves.

Khemehekis wrote: ↑18 May 2018 04:12 I'm 38, native Anglophone, and I heard [jæni]. Yes, with an [æ].

That's not so weird. I hear Laurel but when messing with the NY Times tool, I hear it the exact same way you do when moved to the "Yanni" end. What's weird is I can move the slide to a point on the "Yanny" side, keep it there, continue to hear "Laurel" and then suddenly without me touching anything I start hearing [jæni], and when I move it back the opposite happens! Curiouser and curiouser, when I hear Laurel (as is normal for me), the exact transcription of what I'm hearing is [ɫɑʊɹɫ̩]. So, aside from [ɹ]>[n], the shift to "Yanni" seems to going from total velar/back to total palatal/front.