Join me for another installment of my “Conversations with Robots” series.
While working on the ear training features for the FrankenTongues app, I stumbled across a reference to the Automatic Language Growth (ALG) model of language learning, and the moment I read it, I had to stop everything I was doing to investigate.
Because it resonates loudly with my own views on how we learn languages.
◇◆◇◆◇◆◇◆◇◆◇◆◇◆◇◆◇◆◇
The Traditional View
Current wisdom about language acquisition goes something like this:
Children have an innate facility for learning language; they come out of the womb hardwired to hear patterns in the noises around them, mimic the most common sounds with their mouths, infer meaning for them, and thereby join the world of richly communicating humans.
Then, as they grow up, that facility wanes until, by our teens, any quest to acquire new languages becomes hampered by the ossified library of phonemes we can still pronounce, compounded by an increasingly tin ear for foreign phonemes, and a rigidity of lexicon.
We all know this principle. We hear about it all the time. Researchers have studied its component phenomena; measured them five ways from Sunday, and the results have been spoken: Older people simply lack the neural flexibility to acquire new languages well.
But according to ALG, that may not be true.
Cracks In The Logic
Consider the conclusion that adults can no longer make the mouth shapes necessary to pronounce foreign words the way a native speaker does. Yet somehow, despite this crippling limitation, adults can learn the accent of a foreign language. Whether it’s an actor learning accents to broaden their skills, or a jerk at the local bar mocking an immigrant waiter, adults seem perfectly capable of picking up a foreign accent if they want to.
And what is a foreign accent? It’s when you pronounce the words of one language using the habitual phonemes of another. You know, by making those “impossible” foreign mouth shapes.
Not everybody has an interest in exercising this skill, of course, but enough of us do that it seems to undermine at least part of the standard theory.
My suspicion is that the “inability” measured by researchers wasn’t actually a problem making the sounds. They were measuring something else.
The Drunkard’s Clue
So how do those drunkards manage to acquire foreign accents in the first place? They don’t take university courses; they don’t read books of theory; and they certainly don’t study transcripts encoded in IPA.
According to many dialect and accent coaches I’ve seen: what they do is listen. And listen. And then listen some more. And after they’ve listened enough, and begun to recognize the rhythms and sounds being spoken, they begin to copy them. Not because they’ve now learned to make the sounds - but because they’ve learned to hear them.
Furthermore, as many of those coaches point out, trying to learn from text instead is a dangerous minefield, because adults instinctively read the phonemes with the unconscious rules of their native language instead of the target language. They practice pronouncing things wrong over and over again in their heads.
For example: When learning English, Brazilian Spaniards pronounce the word “difficult” as “thee-fee-col”, not because they can’t say “dih-fih-kult”, but because “thee-fee-col” is how those letters are pronounced in Brazilian Spanish.
So if you want to learn to speak a foreign language well, get your head out of novels, magazines, text books, and newspapers. Listen to the language instead, in radio, movies, videos, and music. Tune your ear first, and your tongue will follow.
Exactly the way babies do it.
The Theory and the Rebuttal
In the 1980s, Dr. J. Marvin Brown was the first to formally describe what is now called the ALG model. His work was based on his experiences teaching adult students to speak Thai. Students who followed his program of intensive listening - and listening only - for the first year, went on to acquire fully native sounding, natural Thai, even though some were studying quite late in life. But those who interrupted their listening regimen and tried to speak too soon, or to read, were no better than students who were taught the traditional way.
Unfortunately, his theory was not well received by language researchers. Not because they studied it and found it flawed, Brown felt, but because their study protocols insisted on evaluating students’ progress throughout the trials. And how do they test a student’s facility with a language? By getting them to read it, testing their vocabulary, having them translate sentences, etc.
But how many babies are tested for their language proficiency? Most kids are given at least a year before parents expect even simple words. And even then, nobody cares about pronunciation or grammar. That doesn’t come until later. Children are free to simply experience their native language - and play with it - for years before we burden them with any other expectations.
So it seems possible to me that, in this case, science may have backed itself into a corner. They’re looking for a way to teach adults to acquire foreign languages fully, but the tools they use to evaluate their methods are cutting them off from replicating what even babies can do. Is the inflexibility of adult language learners an artifact of ossified language skills? Or is it an artifact of the unnatural teaching methods that only adults are subjected to?
“Science has turned its back on our natural toolset for acquiring a language, and instead tries to force it in with a science-shaped hammer. Then they blame the students when the hammer causes damage.”
My own practice
I came across ALG too late in my current journey to take full advantage of it. I’ve been speaking and translating Norwegian in my head for almost a year now, so I’ve already spoiled the ALG effect. But it has altered the way I’m developing my speech skill, which is still my weakest dimension in norsk. I’m putting a much bigger emphasis on passive listening.
So far, it seems to be working. I’m hearing more than I used to. More of the rhythms and cadences. I’m understanding more too. But I’m not going to speak it out loud any more. Not yet. I’ve devised a simple test that will tell me when I’m ready.
I’m waiting until I can speak English with a Norwegian accent.
