Mechanics and Organics: The Power of Output (Class Eight)
Last week, inspired by Levelt’s language acquisition model, I wrote about the tension between complexity, accuracy, and fluency experienced by second language learners, and about Peter Skehan’s proposition to include lexis in the trio of characteristics of a competent speaker. The topic of this week’s class with Dr. Jim was practice, and two of the three articles we read dealt with the role of output and interaction in second language acquisition. The Levelt model of speech that I am still enamoured of re-appeared in an article by Tohoku Gakuin University’s Hitoshi Muranoi, and I will paste the model here once again for reference: recall the different roles of the Conceptualizer (conceives the utterance itself and processes it as a pre-verbal message), the Formulator (takes care of the grammatical and phonological encoding), and the Articulator (retrieves the utterance from the mind and transmits it into audible sounds). Like this:
Now notice that the model is actually cyclical, with the spoken utterance (“overt speech”) proceeding to the “speech comprehension system”, and from there travelling as parsed speech straight back to the Conceptualizer. When I first saw the Levelt model, I assumed that the speech comprehension system belonged to a second participant–a
listener. But looking at it once again after yesterday’s class, I realized that no, the listener and speaker are one and the same in this model, and that’s the beauty of how it works. The speaker hears his own utterance, checks for meaning and grammatical accuracy–that’s the monitoring part–and if seems in some way incorrect, the
Conceptualizer will get back to work on re-formulating the message. Merril Swain calls that “noticing the hole” in one’s own language production. It makes perfect sense, and we do it unconsciously all the time even as native speakers, especially when faced with the challenge of putting something very difficult or delicate into words. It doesn’t sound right to us or isn’t clear enough; we stop mid-sentence, go back and try again, creating a cycle in which output influences input. I must be getting academic-nerdy, because I find that really cool.
So output influences input. By producing imperfect speech, language learners notice the holes in their own interlanguage (that special imperfect and constantly changing working model of the language being acquired) but they can’t yet do the repair work. That’s where interaction comes in. Perhaps the learner’s interlocutor (speech partner) is a native speaker, and consciously or unconsciously happens to use just the phrase that the learner has been struggling to produce. “Aha! That’s it!” says the learner to himself, assiduously comparing the correct version to his or her own imperfect one and plotting to produce the phrase correctly next time, gosh darn it. If context allows, the highly-motivated learner will bring the conversation to a screeching halt and try out the new phrase then and there, hopefully meeting with encouragement and kind words from the patient interlocutor. This, of course, is called “noticing the gap” (Schmidt, 1990), and the process is known as cognitive comparison.
Now let’s look at another model, designed by H. Douglas Brown, Professor Emeritus at San Francisco and the author of Teaching by Principles: an interactive approach to language pedagogy. Are you ready? Here it is, and there’s a world of difference:
One day, in a rebellious moment of simultaneous frustration and inspiration, I was moved in a second language acquisition class I was teaching to create a different “picture” of language acquisition: one that responded not so much to rules of logic, mathmatics, and physics as to botany and ecology…..in a burst of wild, artistic energy, I went out on a limb to extend the flower-seed metaphor to language acquisition. (from a book chapter published as part of the Georgetown University Round Table on Languages and Linguistics in 1993)
I’m sure that Brown’s students thought their teacher was awesome and inspired. And before I proceed any further, I’m not ashamed to admit that I too have a soft spot for this kind of model: it’s warm and nurturing, it’s visually appealing, and it’s more “likeable” than a flow chart featuring boxes with names that sound like cold, linguistic super-hero robots: Formulator smashes Conceptualizer! Articulator crushes Audition! Lexicon caught in the middle! Brown’s model, on the other hand, is mellow and groovy.
But as much as the organic model appeals to my sense of aesthetics, it’s flawed for me, because, unlike the Levelt model, it’s not cyclical. Although this was a language acquisition model and Levelt’s was a model of speech production, I would argue that both the act of speaking and the act of acquiring language are cyclical in nature.
According to Levelt, learners’ own output influences their input, and vice-versa. According to Brown, the seeds of language acquisition are sowed, watered, nurtured (and all those details are all right on the mark), and finally bear the “Fruit of Performance”, or output. What’s missing is that the leaves and fruits on the “Output Tree” need to fall to the ground, turn to compost, feed the roots of the tree, and produce still more output. That could easily be taken care of by drawing a big old compost pile off to the right of the tree, but then what about the role of interaction in language acquisition? It appears that learner’s “production” doesn’t really mature until the end of the natural cycle, with no representation of interaction along the way or of attempts to use language imperfectly (hypothesizing how language works). I tried to think of how to re-work the model to include interaction, but the best I could do was to envision a flock of birds sitting on the tree branches, doing their best to act as interlocutors and spurring the buds on to full bloom. Aviary-locutors, in constant motion.
Six short months ago, I might have been satisfied with the Brown model, so I guess that my education is progressing. That’s a good feeling: I found something cool on the internet, scrutinized it, compared it with the theories I’m learning and with another model, and found it sadly lacking. Although it’s not groovy, the Levelt model still comes out on top for me because it works so brilliantly, even as a model of first language acquisition. A bit of searching on the side reveals that it has been re-designed by Kees de Bott as a model for bilingual speakers….and I think I just may have found a good article to review in a future post. Stay tuned, and thank you for reading!