Ask a linguist: To creak or not to creak?

I got this question the other day from P.D. (who may or may not be related to me): "What do you think of the 'creaky voice' syndrome that seems to be so pervasive among young females in the U.S.?"

I had to do a bit of research before I could respond to this, and it turns out creaky voice is kind of a fraught issue that isn't completely understood. But I'll do my best!

First of all, what exactly is creaky voice (aka vocal fry)? This video gives an exaggerated example so you can hear it clearly--you'll probably recognize it once you hear it (but you can ignore the rest of her rant):




To produce creaky voice, you compress your vocal cords, which makes them slacken and vibrate slowly and irregularly. This is why creaky voice sounds low-pitched and, well, creaky. Here's a visualization of my voice saying "alphabet":

The dark vertical bars each represent a single vibration of my vocal cords. In the first "a" sound, the bars are regularly spaced and close together; this is normal voicing. During the "e" sound, you can see that the bars are more spread out and irregularly spaced--this is creaky voice. (The blue line shows my pitch; the program I used to make this image gets easily confused about the pitch of creaky vowels, which is why I apparently have no pitch there.)

It turns out that everyone uses creaky voice sometimes, especially at the end of a sentence. It's often correlated with low pitch (also frequent at the end of a sentence). I didn't really try to say "alphabet" with creaky voice; it just happened naturally, and I'm not a particularly creaky talker. And this is true of people of all ages, not just young people.

Researchers have also discovered that creaky voice in English and other languages can have a paralinguistic function: that is, it doesn't directly affect the meaning of what you say, but it does aid in communication in some way. For example, it can be used to signal the end of your "turn" in a conversation. In Australian English, creak can signify that you have low solidarity with the person you're talking to. And if you say "yeah" with creaky voice during a conversation, it can mean that you're allowing the other person to take their turn talking, or that you want them to change the subject.

So creaky voice is not restricted to teenage girls. But it hasn't been fully established who actually uses it the most. (For a longer discussion, see this post at Language Log.) This is what I've been able to glean:

• Creaky voice has to some extent been associated with men in the past (it sounds kind of tough and growly, I guess).
• There is some evidence that extensive creaky voice is used by young women on the West Coast, and possibly spread from there. But older women might use it more than men too.
• Creaky voice is perceived by many young people as informal, urban and upwardly-mobile.
• Unexpectedly, NPR interviewers apparently use a lot of creaky voice! Maybe it makes them sound calmer and more disinterested?

I also found this video of a young man using tons of creaky voice. Interesting that it doesn't strike me as annoying or exaggerated--maybe because extensive creak originated as a masculine characteristic?



So, bottom line, creaky voice has been around forever and has several conversational and social functions. So try not to let it bother you!

P.S. There are some languages (including several native Mexican languages) that use creak contrastively; that is, a word with a creaky vowel will have a different meaning than the same word with a normal vowel. Creak is a pretty versatile thing!

Talking funny: Thai tones

Statue at Wat Phra Kaew, Bangkok (photo by Vladimir Samoilov)

The interesting linguistic feature I'm going to talk about today is exotic from a Western perspective, but will be familiar to speakers of many East Asian (and also African) languages. That feature is tone. A language has tones when it uses the pitch of the voice to distinguish different words. Tones can consist of a single steady pitch, or a contour (a change from one pitch to another).

In Thai, for example, a syllable can have one of five possible tones. Here's a set of five words that differ only in their tone. You can hear these words being spoken here (just click on the words). Can you hear the difference between the tones?

1. naa21 [a person's name] -- low falling tone
2. naa51 'face' -- high falling tone
3. naa45 'aunt' --high rising tone
4. naa213 'thick' -- low falling-rising tone
5. naa32 'field' -- mid falling tone

The numbers after each word are a way of describing the tone. The number 1 means the lowest possible tone, while 5 is the highest. Each of these words has a contour tone--a changing tone--so multiple numbers are needed to describe them.

Many people who don't speak a tone language (including many students I've taught) have a hard time hearing the difference between tones, even when the tones are explicitly pointed out to them. This is normal, of course, because we're not used to listening for pitch in that way. But for speakers of a tone language, it's as easy as hearing the difference between p and t is for us.

And it's not just a matter of listening for specific pitches; we can't just say "in Thai, a low falling tone starts at 110 Hertz and drops to 90 Hertz". Why not? Well, the most obvious reason is that all people have different voices with slightly different pitch ranges. A man might produce a "high" tone that's actually a lower pitch than a woman's "low" tone. So when you meet a person for the first time, you have to be able to adjust to that person's natural pitch range. And one person's pitch range can change depending on many factors (their emotional state, if they're imitating someone else, etc.).

But worse than that, even if we're just talking about a single speaker, saying a single sentence, the exact pitch of a given tone will still vary a lot. This is because all languages use pitch for something called intonation. Intonation is the use of pitch to demarcate and emphasize various features of a sentence. For example, in English, when we ask a question, we generally put a rising intonation at the end of the sentence. To demonstrate, I recorded myself asking a question and saved the pitch track (which shows my pitch over time):

That big upward climb during the word "brownie" indicates that I'm asking a question.

Another use of intonation is to emphasize words that the speaker wants the listener to pay attention to.

In this case, the big peak on the word "brownies" marks it as the most important word in the sentence (as it should be!).

Tone languages make use of intonation too. So speakers of Thai would have to be able to take intonation into account when listening for tone.

There's yet another difficulty in the way of perceiving tone. Phoneticians have noticed that over the course of a sentence (or a longish phrase within a sentence), a speaker's pitch tends to gradually drift downward. So a word with a pitch peak (like the word "brownies" above) may still have a lower pitch than earlier words in the sentence. To illustrate this, I recorded myself reading the first line of Pride and Prejudice:

You can see that, although there are lots of peaks and valleys in my pitch (corresponding to intonational patterns), it generally decreases over time. Speakers of tone languages have to compensate for this too!

So tone seems pretty wild to us English speakers, but really it's just another example of the amazing power of the human capacity for language (not to mention the very fine control we have over our vocal cords!).


P.S. #1: Another thing I didn't mention, because I don't know a lot about it, is the fact that tones can actually mess with each other when they're adjacent!

P.S. #2: If you're wondering why there are some gaps in the visualizations of my pitch, it's because not all speech sounds have a pitch. During the sounds s, f, t, and p, among others, your vocal cords aren't vibrating; the only sound is noise caused by air bouncing around in a narrow channel (in the case of s and f) or the explosive release of air in t and p.

Talking about talking: regular mice

Last month I posted about the linguistic process of analogy--where speakers change one part of their language to be more like another part. In that post, I mentioned that in general, a given irregular pattern in a language will tend to become regular over time. This raises an obvious question: if speakers tend to eliminate irregularity, then why aren't all languages totally regular by now? Where do new irregular patterns come from? The answer is actually quite simple: they're accidental.