Phonies Yike Phonowogy
Posted by Neal on July 14, 2004
A couple of posts back, I was talking about Doug’s pronunciation of /l/, and how I couldn’t figure out a natural-seeming rule that would explain why he sometimes pronounced it as [y], and sometimes as [w]. Writing the post got me so curious that I sent a message to the Phonies, an informal research and discussion group in the Ohio State Linguistics Department. I asked them what they could make of it, and Phony Jeff Mielke rose to the task, offering an explanation that made more sense than anything I could think of.
It all has to do with “clear” and “dark” l, two different ways that /l/ is pronounced in English. Clear l, also known as non-velarized l, is the variety you get at the front of stressed vowels, for example like and luck. Dark l, aka velarized l, is what you get at the end of a syllable, as in elephant or Neal. Whether you’re making a clear or a dark l, the tip of your tongue is touching the roof of your mouth (or for some speakers, is between their front teeth). The difference between clear and dark is that for the dark l, the back of your tongue is raised up near the roof of your mouth, too–to the soft part of the palate called the velum (hence the name ‘velarized’). This is a difference that you don’t ever notice until you hear a speaker whose l’s just don’t sound quite right. For example, if you ever hear Snigdha Prakash on NPR, listen to how she says dollar. She’s pronouncing it with a clear l, and it sounds a little off.
So far, the clear and dark l distribution in English parallels Doug’s [y]/[w] distribution: He’s realizing clear l as [y], and dark l as [w]. But what about the consonant clusters, where he sometimes has [y] and sometimes has [w]? Jeff notes that except for those that start with [s], the clusters all start with either labial consonants ([p,b,v,f]) or velars ([k,g]). As it turns out, [w] is both labial and velar: you round your lips to say it, and you also raise the back of your tongue. So if you say an /l/ after a labial or velar consonant, chances are that your lips or tongue will still be in position for that consonant, giving your /l/ sound some of the properties of a [w]. So given a choice between [y] and [w], it’s not surprising that Doug’s /l/ should end up as [w] in these clusters, but not in the clusters beginning with [s]. As Jeff sums it up, “Doug was producing [w] in the environments where the adult /l/ sounds most like [w]: when it’s dark l or when it’s a coarticulatorily velarized/labialized clear l, and producing [y] where it sounds less like [w].”
And as a bonus, Jeff adds a falsifiable prediction: “Doug might have produced /l/ as [y] in two environments we don’t see in the data: word-initially before an unstressed vowel, and after [the sh sound] but I can’t think of any words like that that I would expect a 3-year-old to say.” In other words, we would expect Doug to have pronounced a name such as Latanya as Yatanya, and a word like schlep as shyep. But he never said either of those words (at least, not within my hearing), so I guess we’ll never know for sure.