#formants

A detailed discussion of pitch, intonation, and the role of technology in language description in Gizmodo. Excerpt: 

In school, we generally learn about two types of speech sounds: consonants and vowels. But I promise, there’s more! One layer of additional structure in our speech is stress. As Mike Myers demonstrates in View From the Top (2003)—“You put the wrong em-pha-sis on the wrong sy-lla-ble!”—in English, one specific syllable in multisyllabic words is more prominent than the others. Stress is one part of prosody, which is a large umbrella of speech phenomena that take place in larger domains like syllables and phrases, instead of smaller pieces like consonants and vowels.

But the real fun (if you’re me) begins when you ask how we know a syllable is stressed in the first place. The best clue is how the word interacts with intonation, the part of prosody that investigates how languages use tonal melodies.

For example, let’s say you’re at work, and someone walks into the break room and utters one of the following:

1. “There’s coffee.”

2. “There’s coffee?”

Same consonants and vowels. Same context. The first is a statement informing that there’s coffee. The second is a question, possibly where someone is surprised to hear that there’s coffee. Aside from periods and question marks, purely the domain of writing, what exactly is the difference between the two?

The most common approach to modeling intonation is by using the building blocks H (high tone) and L (low tone). A rise can be described as LH, and a fall as HL. These and longer melodies are used for one of two purposes: 1) a ‘pitch accent’ which marks a stressed syllable; or 2) a ‘boundary tone’ which marks the edge of a phrase (like a comma might do in writing).

It’s The Dress but for linguistics (from Cloe Feldman‏ on twitter). 

A couple explanations of what might be going on. From The Verge:  

The secret is frequency. The acoustic information that makes us hear Yanny is higher frequency than the acoustic information that makes us hear Laurel. Some of the variation may be due to the audio system playing the sound, Reicke says. But some of it is also the mechanics of your ears, and what you’re expecting to hear.

Older adults tend to start losing their hearing at the higher frequency ranges, which could explain why Riecke could only hear Laurel, but his eight-year-old daughter could hear Yanny. It’s a phenomenon you can mimic on a computer, he says: if you remove all the low frequencies, you hear Yanny. If you remove the high frequencies, you hear Laurel.

From Rory Turnbull on twitter: 

Here's what I think is going on. In the first syllable, there's only one major spectral peak below 2.5kHz. It has a wide bandwidth, which is consistent with an F1 and F2 very close together: an /ɑ/ (for "Laurel"). 

The higher spectral prominence dips down about halfway through the word, between the two syllables. If the lower spectral prominence is F1 & F2, then the higher one must be F3. A low F3 = /ɹ/! 

But what if we treat that higher spectral prominence as an F2, rather  than an F3? Then we have a very high F2 in the first syllable, consistent with a front vowel or approximant, e.g. /j/. The F2 stays  pretty high and the F1 gradually rises, giving a percept of /jæ/

The fall of the F2 between the two syllables is then consistent with an /n/, although we don't see the general amplitude dampening that we normally associate with nasals. The F2 rises and F1 falls again at the end, resulting in /jæni/ overall.

An earlier example of the Yanny/Laurel phenomenon: BILL BILL BALE BALE PALE PAIL MAYO

I saw on Wikipedia an article about “Formants”,  and I think it’s neat!

The Formants are like, something about the different frequency peaks in a spectrogram of speech, and different vowels have different ones. Here is a chart from Wikipedia of the different ones that different vowels have on average:

(this image available at [link] and is licensed under CC BY-SA 4.0 )

And, unless I am misinterpreting this chart, it appears to me that there is a region of sounds which is possible to produce, and which I expect would sound somewhat like a vowel, but would not sound like any vowel that people can make with their voice. Specifically, I mean the area near the bottom left of the chart, say with F2=2100Hz and F1=800Hz .

(Or at least, I imagine that people can’t pronounce these “vowels”. I could be wrong though. Maybe languages just happen to never use those vowels, despite being available. Maybe people can pronounce them, but doing so is inconvenient so language avoids them?)

I think it could be neat to do something like this. Like, maybe make a text* to speech program which supports vowels that people can’t* pronounce, but could maybe learn to recognize.

Maybe a fun conlang could be produced which people could understand easily enough, but would be physiologically unable to speak!

Oliver Niebuhr doing a presentation on “Phonetic exponents of charismatic speech”, March 1. 2018. A diagram of the vowel spaces for Mark Zuckerberg and Steve Jobs before different audiences.

How do we create different kinds of sound waves when we only have one mouth? What properties do those speech waves have? In this week's episode, we talk about resonance and formants: how different parts of a speech wave can get amplified, how that relates to how we talk, and how the sounds of vowels are influenced by our tongue and lip setups.