Going gaga over phonetics

(Nancy Friedman's recent post reminded me of the McGurk Effect.)

A long time ago I started to doubt that [m] and [n] could possibly be different sounds. I would close my mouth and make the mmmmm sound then I would put my tongue on the alveolar ridge. That's really not as dirty as it sounds. I would open my mouth trying to make the transition so smooth that the change in sound was imperceptible. I figured I was probably succeeding.

My parents were worried about me.

One of the reasons I went into linguistics was my fascination with stand-up comedians. Especially those that did impressions and impersonations. As a kid I was also taken by ventriloquists. Not kidnapped -- just amused.

It all seemed like some sort of magic trick. How could they say [p]s and [b]s and [m]s and even [f]s without using their mouths? The [f]s were easier for me to accept because air quickly forced between both lips sounded a lot like the puff between lips and teeth. But those others had me stumped.

In fact there is a difference between [m] and [n] and it has to do with the amount of space allowing the sound to resonate in your mouth. But it's a slight difference and it's mostly context and the interaction with surrounding sounds that allows us to perceive it.

And ventriloquists can count on our eyes to influence what we hear as well. [added later: That may seem counterintuitive because the ventriloquist isn't showing the correct articulation. But in the ventriloquist's case we're expected to ignore the performer's face and concentrate on the dummy's mouth movements. That in combination with ambiguous articulation and reliance on clues such as familiar words and likely interpretations. E.g.: We rarely hear such clusters as thl- so in a word like thlag we are more likely to hear it as flag especially in a sentence in which we expect to hear flag.]That's the McGurk Effect (the influence of visual cues on aural information). Watch this video. (downloaded from Arnte's sound site)



What sound is he making? Even when Buffy tries really hard to listen 'objectively' and even after she has listened while looking away and agreed that he's saying [baba] Buffy cannot but hear it as [dada] while watching. Even if she's just looking at his eyes.

The man you see in the video is articulating [gaga] but most people hear it as [dada]. We 'hear' [dada] because the phonetics of [dada] (alveolar stop) are closer to [baba] (bilabial stop) but look more like [gaga] (velar stop). So our perception splits the difference between the sound we hear on the audio track and the image we see in the video track.

The following isn't actually the McGurk Effect because the sound actually is changing as you hear it. But this is one of those tricks I used to do when I was a kid playing around with sounds. I figured that since I said [w]s by almost closing my mouth I could approximate that same effect using my hands. I would hold my mouth in an exaggerated 'aaah' position and use my hands instead of my lips to create a 'labial' approximant as an onset and a coda.

I've never performed this for anyone. And since no one reads this blog I'm willing to post a quick video of it.



Buffy's worried about me.

Broadcast standard.

We just drove through Iowa and spent an evening in Des Moines with friends.

It's true what they say. I didn't hear a single accent the whole time I was there.

Except for a young boy about a year old who was practicing his linguolabial stops. I think they were voiced because he wasn't aspirating them. They sounded a lot like [b(ə)].

A linguolabial is marked by a subscript 'birdie' (or 'seagull') diacritic. The coronal articulation is noted by the [d] and its labial placement gets the diacritic [   ̼]? The IPA doesn't recognize a [b̼] because the [b] would be a bilabial and the linguolabial doesn't use the lower lip. The 't' and 'd' are the most reasonable coronals to use because symbols like 's' 'n' 'θ' 'ð' 'ʃ' 'ʒ' etc are fricatives not stops. The other coronal stops 'ʈ' and 'ɖ' would make little sense because they are retroflex symbols. Why would a symbol used for a retroflex be used in a symbol for the opposite of retroflex?

Approximating season

In a comment a while back Nancy Friedman mentioned some voting options she found on the Overheard in New York site. She wrote

Also note that you can vote for the Overheard quotes. Thumbs Up, Thumbs Down, and WTF? need no explanation. But I had to Google "Alsome," which turns out to be the way the REALLY cool folks say and spell "awesome." (Well, spell it anyway. I'm not sure the two words are pronounced very differently. Hey, Linguist Guy, whaddya think?)

I think some people might pronounce it with an [l] or more likely one of the darker L's--either a velarized alveolar lateral approximant [ɫ] or the velar lateral approximant [ʟ].

English commonly uses the dark-l in a coda position and the light-l in an onset. The usual transcription of the dark-l indicates that velarization is a secondary articulation. [l] is an alveolar consonant and is articulated with the tip of the tongue against the alveolar ridge. Velarization means that in a secondary articulation the back of the tongue is also raised giving it a 'dark' sound. Compare the first and second L in little. Or the L in light with the L in call. The difference is indicated by the tilde that runs through the symbol [ɫ]. [liɾɫ̩] [lait] [kaɫ]

The velar lateral approximant is primarily velar and so the raised back of the tongue (the dorsum, think the dorsal fin on the back of a fish) is the primary articulation. The tip of the tongue isn't the main indicator of place. Although this is not a standard phone in English it does occur in some accents and some pronunciations as when a coda L is left markedly open. In fast speech the L in a phrase like 'all of them' might remain completely dorsal without any alveolar articulation.

Consider how close to 'awvem' that can sound. Now consider that some dialects will pronounce L's like W's. Ever heard someone say 'widow' instead of 'little'? It's a common early pronunciation among children because the sounds are similar acoustically. I know one child who pronounced 'flower' like 'wallow'--a complete reversal of the [l] and [w] approximants. Tho this might have been metathesis instead of an articulation issue.

Just earlier today I heard 'saw' pronounced before a vowel like 'sawl'. It was overheard only once so I'm not sure if it was a velar lateral or a velarized alveolar lateral, but it was definitely an approximant. I've heard it before. I asked a friend if she hears this a lot around here in Nebraska. She rolled her eyes and gave an exasperated "Ugh. Yes."

Before a vowel it's very much like I sawr 'im leaving which we would expect to hear in some northeastern American dialects. The approximant makes a nice onset for the following syllable.

I imagine also that to those who are used to hearing [ɑ] in 'awesome' the northeastern closing and rounding and raising to [ɔ] could be interpreted the same as a velar approximation.

Whatever the process there are plenty of likely explanations for the awesome/alsome alternation.

A Flat Tree Will Get You Nowhere

So Daniel wants to preserve the clarity of some pronunciation distinctions while Buffy would like to support my claims because she loves me and I'm always right.

I'll agree with Buffy. But I'll still address Daniel's issue. It's a fair challenge he poses to my analogy -- as I pose articulation with football.

I bring issue to your analogy. Movement, opening a door, running a football, picking up a saucer, is a personal and individual task. . . . I think speech should be interactive.

And yet we can reconsider the various tenors of the simile and say that in each of those activities there is a goal just as in speech there is the goal of reasonably unimpeded transference of an idea. So now we look at each performance whether in a social setting or in an empty cafe or through a lonely doorway and we again pose the question: "Why would adjustment and continuity be valued in one and derided in another?"

So when we're communicating with another, clarity is important. That awkward [t] might be important when stressing writer or whiter. Yes, there are many ways to effectively communicate, including some "tripping" language, depending on our audience, but efficient speech doesn't mean effective speech.

This is certainly true. The most efficient speech, without regard from the Speaker for the ability of the Hearer to understand, would be something like a *sigh* or an open mouthed "uuhhhh" probably corresponding to exhalation. But this is not what I defend when I use one example from a common American dialect that voices and flaps intervocalic dentals. In that example I defend a particular allophone that is quite clear to all Americans. And lest we say that the flap is at the top of a slippery slope we have to look at the company changes often keep.

American English does elide some sounds. In 'laboratory' many American speakers syncopate the second vowel saying [ˈlæ.brǝ.to.rɪ] while some British dialects would elide the penultimate vowel saying [lǝˈbɔ.rǝ.trɪ]. Any readers who know this to be false about BrE please let me know. This may be my over-eager application of a pre-sonorant/post-schwa elision rule.

If we consider again the words 'rider' and 'writer' we find that the vowel change very ably serves the goal of distinction. So while the purely phonemic underlying form differs only by the voicing of the middle consonant the surface form in AmE differs by vowel quality instead. Here is where I bring back my analogy with a running back. One color commentators will say Man that Emmitt Smith is a hardworking back. Look how he just pushes up the middle and keeps turning his legs when he hits the line and another commentator says But he's nothing like Barry Sanders. Look at how far he runs before he even gets to the line. If they counted lateral yards Sanders would have the rushing record by far! Then the first would argue that Smith did more work by being a receiver as well.

If we consider the pre-sonorant/post-schwa elision rule at work in both BrE and AmE pronunciation of the word 'watery' we see tit-for-tat surface forms. In some BrE dialects 'watery' is pronounced as two syllables because of schwa deletion: [wɔː.trɪ]. Because the schwa will not elide after a flap the AmE surface form remains trisyllabic: [wa.ɾǝ.rɪ]. We find the same comparison of AmE and BrE 'flattery': [flæ.trɪ] & [flæ.ɾǝ.rɪ]. It goes back and forth and becomes an argument about which types of distinctions are the more virtuous to preserve.

There is a larger arbitrating factor. When the allophones and pronunciations in question are part of a dialect (as opposed to an idiolect) the ability of the Hearer to understand is already attested. When a dialect recognizes and favours one distinction we use preposterous logic to say that it is out of consideration for the audience that we discard its phonology.

Okay -- Next time: Buffy's careful efficiency. It's related to this.

Speaking Trippingly

Buffy has a funny habit of pronouncing words both carefully and "efficiently" at the same time (I'll give examples in my next post). I say efficiently because I'm trying to remain neutral on dialectal judgment. One of my professors likes to use the phrase "lazy pronunciation" to describe phonological alternations, but I don't like the implications of "lazy."

Steven Pinker makes an excellent case against the view that speech markers are the result of lazy articulation. In The Language Instinct he first points out that the same people "who are derided for dropping g's in Nothin' Doin' are likely to enunciate the vowels in pó-lice and acci-dént that pointy-headed intellectuals reduce to a neutral "uh" sound" (180).

Add to that the diphthongization of the vowels in "tin can" -- leading the words to become disyllabic instead of monosyllabic -- and how can we call it lazy when some words are pronounced with units that are more marked and others with a greater number of units. Northern [kæn] is more simple than [kæ:jǝn]. So the Southern American accent is doing more work -- quantitatively.

Pinker also makes the argument that assimilative articulation is completely natural. He draws on the analogy that when you reach for a saucer and a coffee cup you grab the edge closest to you and you put your hand in the grasping position before you touch the cup.

Let's look at another example. I'll make the analogy from walking through a door to the post-stress intervocalic voicing in American English [t] > [d,ɾ]. Imagine walking towards a glass door that you know opens by pushing. Imagine further that you are familiar with this door and know that it opens easily without a strong push. When you approach the door you see there is no one on the other side and you put your hand forwards to contact the door and push it aside while your stride continues. In my observation I have seen that even when the door opens towards the "walker" and even when a knob has to turn to unlatch the door there is still often one foot in motion so that the stride of the walker is barely interrupted -- at most it slows just a bit.

Is this continuous movement of the feet evidence of slovenly or slurred motion? Would we say that these people are slaughtering the beauty of graceful and careful walking? On the contrary. By anticipation and adaptation this shows balance and agility. How odd it would look if before every door the walker stopped -- put out a hand -- placed palm against handle -- closed fingers around handle -- pulled door -- stepped through door...etc.

And yet we argue this as the very virtue of a good speaker. When I say "water" with a strict [t] between the vowels I find myself feeling stilted and guilty. That interruption of voice is just as unnecessary as the flat-footed stop in front of a door. In my dialect there is a wonderful allophone corresponding the voiceless and voiced alveolar stop: the alveolar 'tap' or 'flap.' This is not a simple and easy phone. It's a very quick and light lingual articulation. It requires great precision to avoid sounding like the lateral [l] or the voiced fricative usually represented orthographically as 'z' or 'zh'. To argue that this allophone is a product of laziness is like arguing that a running-back is lazy because he runs around a linebacker instead of walking right up to him and trying to push him over.