Posted by Neal on January 2, 2008
The December 2007 issue of Language arrived while we were packing for our trip to visit Mom and Dad. I glanced at the contents and some abstracts, figuring I’d read more when we got back. The first article: “Positional neutralization: A case study from child language,” by Sharon Inkelas and Yvan Rose. When I looked at the abstract, I realized this article couldn’t wait for us to get back from our trip; it would have to go in my carryon bag for the plane. It was about a child they referred to as E, who from the ages of about one and a half to three years exhibited … lateral gliding.
Lateral gliding, you say?
Yes, lateral gliding! Lateral is the phonetic term for /l/ sounds, and glide is a term referring to vowel-like consonants such as [y] and [w]. (They’re also known as approximants.) Lateral gliding, then, is the pronunciation of /l/s as [w]s and [y]s. Sound familiar?
Maybe you remember a few years ago, when I wrote about how Doug pronounced his /l/s until he was about six years old. Sometimes he’d say them as [y], sometimes as [w]. (BTW, I should mention that I’m using a corrupt version of the International Phonetic Alphabet here. Technically, the consonant y sound is written as [j]. The j sound, meanwhile, is written [ʤ]. But for consistency with the posts I’m linking to here, and to lessen confusion for my nonlinguist readers, I’m representing the y sound as [y].) I was dissatisfied with my own analysis of the rule describing when Doug would produce a [y] and when he’d produce a [w]; another linguist had a better one, but I was naturally curious about what a paper in a scholarly journal would have to say on the subject. This wasn’t academic; it was personal.
To sum up Inkelas and Rose’s description of E’s /l/ pronunciation, E would pronounce /l/ as [y] in prosodically strong positions, and as [w] in prosodically weak positions. I gathered from the paper that a prosodically strong position is the beginning of a word or the beginning of a stressed syllable (or both); weak positions include the beginning of an unstressed syllable that isn’t at the beginning of a word, and at the end of a syllable. This strong-vs.-weak rule almost describes what Doug did, too. He would pronounce lucky, with /l/ at the beginning of both the word and its stressed syllable, as yucky. As an example of /l/ at the beginning of an unstressed, noninitial syllable, Doug would pronounce elephant as ewephant. An example of /l/ at the end of a syllable: he would pronounce bottle as bottow.
What about consonant clusters with /l/ in them, i.e. /kl, gl, pl, bl/, and /sl/? This is a prosodically weak position for the /l/, not being at the beginning of the syllable. E would most commonly just eliminate the /l/ in these clusters. When he didn’t, if the cluster was [kl] or [gl], he would almost always use a blending strategy, pronouncing the cluster as the ch [ʧ] or j [ʤ] sound, respectively. However, he would attempt an /l/ when it came in a [pl] or [bl] cluster, and when he did, it came out as a [w], just as with other prosodically weak positions. An example Inkelas and Rose give is [bwæ̃kət] blanket. This is also like what Doug did, saying bwue for blue. To some extent, Doug’s pronunciation fit the strong-vs.-weak rule even better than E’s did, since he didn’t have the blending rule getting in the way for his /kl/ and /gl/ clusters, turning them into [ʧ] and [ʤ]. He pronounced them as [kw] and [gw], as in Cwifford and Gwen for Clifford and Glen.
However, Doug’s pattern departs from E’s when it comes to the one exception that I wondered the most about. For /sl/ clusters, unlike all other clusters, Doug would pronounce his /l/ as [y] instead of [w]; for example, syeep for sleep. So after reading Inkelas and Rose’s article, I still think Jeff Mielke’s analysis, involving clear and dark /l/, is the better candidate. But you know whose language is (or at least was) consistent with the strong-vs.-weak rule, even for /sl/ clusters? Doug’s friend’s little sister, who pronounced sleep as sweep.
So why would Doug’s rule for producing /l/ hinge on the clear/dark distinction, while E’s, and Doug’s friend’s little sister’s, involve this prosodic strong/weak distinction? This is the most interesting part of Inkelas and Rose’s paper.
You see, before E began to exhibit lateral gliding, he exhibited velar fronting, a very common phenomenon in young children. That is, his velar stops (that is, [k] and [g]) would sometimes be realized as the corresponding alveolar stops ([t] and [d]), closer to the front of the mouth. When, exactly, would this fronting occur? Right: in prosodically strong positions. And why should this happen? As it turns out, phoneticians have determined that “consonants in prosodically strong positions … show larger amplitude in their articulatory gestures than those in other positions” (p. 724). In other words, the tongue makes a bigger movement in these positions. Moreover, in very young children, the tongue fills more of the mouth. These two facts together mean that if a very young child makes a bigger-than-normal tongue movement to make a [k] or [g], they’re liable to end up with not just the back of the tongue on their palate, but the front of it, too, and end up with [t] or [d].
So what does all this have to do with lateral gliding? Inkelas and Rose speculate that E simply extended a rule that was working well for him for velar sounds to help with his /l/s, too. He started with a clear-vs.-dark rule, with the clearest /l/s (i.e. those at the beginning of a word) pronounced as [y], and the darkest /l/s (i.e. those at the end of a word) pronounced as [w]. But since the clearest /l/s occur in a strong position, and the darkest /l/s occur in a weak position, for the cases where it’s a little harder to call an /l/ as clear or dark, E just made all /l/s in a strong position into [y], and all those in weak positions into [w].
This hypothesis is consistent with Doug’s case: He never exhibited velar fronting, so he never had a strong-vs.-weak rule to extend to his /l/s. Instead, he stuck with the clear-vs.-dark rule, and just had to make the clear-vs.-dark call for every /l/ he heard, even the tough cases like consonant clusters and the beginnings of unstressed syllables. This hypothesis also suggests a prediction: If I ask Doug’s friend’s dad about the little sister, it will turn out that she had velar fronting as well as lateral gliding. It’s too late to call him tonight, but stay tuned.
In the meantime, I have to say that I’m ashamed of myself as a linguist parent. According to Inkelas and Rose, “The data from E were gathered in a naturalistic, diary setting primarily by his mother, a trained phonologist.” I made a few observations in my linguistic notebooks about Doug and Adam’s language acquisition, and have even written up some of those observations here. But did I keep a daily record of it? Nooooo. Heck, I can’t even remember what the last thing was that I wrote in either of their baby books. E’s mother, however, was able to note that at the age of two years, three months, and zero days, E still had velar fronting, but that by two years, three months, and five days, it had completely disappeared. I tip my hat to the unnamed phonologist mother of subject E, not only for her diligence and keen ear, but for being willing to turn over this gold mine of linguistic data to the two other phonologists who wrote it up and published it in the official journal of the Linguistic Society of America.
UPDATE: I talked to Doug’s friend’s parents, and they weren’t quite sure if their daughter had exhibited velar fronting. The mom said it seemed like she had, but she couldn’t think of any examples right offhand. The dad recalled that she used to call him “Dag,” which is an example of the opposite of velar fronting (known as coronal backing). Furthermore, it’s happening in a prosodically weak position, not a strong one: She said Dag, not Gad. So far, then, Doug’s friend’s sister neither supports nor undermines the hypothesis that her kind of lateral gliding is primed by velar fronting.