I'm a big fan of Stephen Krashen's research and the comprehension hypothesis. I'm garnering enough experience learning Thai that I'm now convinced that input is the only way to truly acquire a language.

Recently, I've been thinking about how to determine a level of input that provides optimally efficient acquisition. That is, how can I tell when input is at level "i+1"? The American University Alumni language school uses a very pure input approach, and they provide some guidelines. AUA instructors target 80% or greater understanding for all of their students. For AUA, "understanding" includes not only familiar vocabulary, but also the context of visual and experiential cues. For example, a beginning student playing Uno in Thai might understand fewer than 80% of the words, but most people who already know the game would understand at least 80% of the experience. After a large amount of comprehensible experience in the foreign language, words acquire meaning without resorting to translation.

I think an 80% rule-of-thumb is good, and an inclusive definition of understanding is also good. In fact, I cannot imagine a pure input approach which does not allow for understanding by contextual cues. Without contextual cues, there would be no way to "get started" without introducing translation of some sort and breaking the input model.

This definition of understanding means that all sorts of material is available. For the months of May and June, I studied primarily by watching children's movies in Thai. I avoided subtitles, since they can interfere with acquisition. By using visual and contextual cues, I was able to have 80% or greater comprehension while only following 30% of the spoken dialogue. Watching the same movies again and again, my comprehension of the spoken dialogue improved, and for one those movies it's now around 60-70% and still growing.

My vocabulary is bigger than it was in May, and a new source of comprehensible input is radio news stories. If I already know something about the subject, I can fill in gaps left by unfamiliar vocabulary. If I don't know much about the subject, I have to understand more of the vocabulary. The vocabulary is usually easiest in human interest stories, like a review of the week's top movies.

It's not always possible to understand 80% of everything. Like Tony, an enthusiastic ESL blogger, I place the highest priority on my level of interest. If I am very interested, I will listen again and again with a great desire to uncover the meaning. Even if I only understand 50%, the listening is very productive.

I am not a purist. A very pure input approach doesn't seem practical without a teacher. I own a dictionary, and I do use it. However, I strongly emphasize input over anything else, and I avoid memorization, letting words and phrases build themselves into my vocabulary naturally.

I have been listening to the Voice of America in Thai.

It's propaganda. I know. As an American who values critical thinking, I'm suspicious of any media outlet that is controlled by my government... especially recently.

But as a Thai acquisition device, VOA is awesome. The vocabulary is everyday. The content is interesting. The announcers are engaging. I often know something about the topics. Words get repeated often. Individual stories are perfect for narrow listening.

I edit the news in Audacity and listen to interesting segments over and over. As usual with Thai content on the internet, the content is a little beyond my i+1 level. But I can look up a few words.

Stephen Krashen, the American University Alumni school (AUA), and others recommend an initial "silent period" for students of a second language. During the silent period, the student is instructed to listen and understand, but not try to speak. When I registered at AUA in Bangkok, an instructor asked me many questions in both Thai and English while observing my answers, speech, and comprehension. Because of my background in Thai, I was placed at an advanced level where complete silence is not required. Consequently, I do not try to stay "silent" all the time in Thai. However, there is a long-term version of the silent period, which is a useful behavior pattern in general for language acquisition. AUA alludes to this in some of their literature, such as this paragraph from "Just Let Me Try To Say It" by David Long (italics mine).

'When will I start to speak?' Without doubt, this is the most often asked question. This is also a question that a baby or young child never asks. The easiest answer is, 'When you're ready.' Each person will be little different but overall, when a student has acquired between 60% and 70% of the new language the phonemes of the new language are firmly set. After that, it is simply a question of whether or not the word is 'there' when you need it or not. If it's not, don't worry about it. Use what's there. You'll be amazed at how well it will work for you. When a student begins speaking, it isn't that the language will immediately come out perfect, but that he doesn't have to 'remember' anything at all. He will simply think the thought and the words will be there. This is exactly how your native language works for you. The key is that because the student is drawing only from the input of Thai teachers, those things he says will correct themselves in just a short while.

Just Let Me Try To Say It, David Long

For me, "using what's there" consists of using English words, remaining silent, or changing what I'm saying. This completely circumvents a hesitancy that can occur when a student learns by translation and is forced to speak.

For the last week or so, I have been listening to a 12 minute Thai Buddhist discourse called "Living With the Cobra". It's somewhat beyond an i+1 level for me, but not by much. It's short, I have a transcript, and there are many repeated words, so with a little help from my wife, I can make sense out of it. People tell me that discourses on Buddhism are usually difficult because of specialized words and language. This particular one does have some specialized language, but it seems more accessible, perhaps because it's addressed to a westerner.

I could listen to this audio for hours and hours without losing interest. In college, I listened to a tape of the philosopher J. Krishnamurti over and over. It was all-absorbing because the subject was so interesting and he is so eloquent. By listening repeatedly with interest, I automatically reached a point where I had large parts of the tape memorized, including Krishnamurti's intonation, rhythm, and even pronunciation. This Thai mp3 is on a similar subject, and the speaker is similarly engaging. I can imagine myself assimilating this mp3 in the same way over time.

My son is almost three years old. Consequently, over the past couple of years, I have had the opportunity to observe his natural language acquisition. It has been fascinating. Several things surprised me, most notably that if he cannot pronounce or does not know all of the words in a phrase, it doesn't stop him at all. He just says the phrase, coming as close as he can to the unknown words. Over time, these "approximations" come closer and closer to the words he is imitating.

His internal model of the words and phrases is apparently far ahead of his pronunciation. If we misunderstand a word or phrase, he always corrects us and says "no", that is not what he meant. Then he repeats what he said. Sometimes, he will repeat the misunderstood word several times, while we try different guesses at what he is approximating. Once we arrive at the right word, he finally tells us "yes" -that's what he meant.

This seems very relevant for my acquisition of Thai, and at the same time it is somehow very reassuring. If a child's acquisition of language can be an accurate model for an adult's acquisition of a second language, pronunciation will come along by itself, without explicit practice, once an accurate internal model of words and phrases is established. This is consistent with my understanding of the AUA literature and other research on second language acquisition. The nice thing is that it's normal to pronounce things imperfectly at first, and it's not harmful as long as an accurate internal model of words and phonemes has been well-established by extensive listening to native speakers.