Exploration in Language Acquisition and Use : the Taipei Lectures

Chapter One:

PRINCIPLES OF LANGUAGE ACQUISITION

Stephen Krashen
University of Southern California, Los Angeles (U.S.A.)

PART 1: THEORY 

The following five hypotheses are the core of current theory on language acquisition. They are presented here as a summary without supporting evidence. For much more detail, see Krashen (1981, 1982, 1985, 1994a), Krashen and Terrell (1983). 

The Acquisition-Learning Hypothesis 

The acquisition-learning hypothesis claims that we have two independent ways of developing language ability. 

Language acquisition is a subconscious process, while it is happening, we are not aware that it is happening. Also, once we have acquired something, we are not usually aware that we possess any new knowledge; the knowledge is stored in our brains subconsciously. The research strongly supports the view that both children and adults can subconsciously acquire language. Also, both oral and written language can be acquired. 

In non-technical language acquisition is sometimes referred to as "picking up" a language. When someone says, "I was in France for a while and I picked up some French," it means he or she acquired some French. 

Language learning is what we did in school. It is a conscious process, when we are learning, we know we are learning. Also, learned knowledge is represented consciously in the brain. In non-technical language, when we talk about "rules" and "grammar," we are usually talking about learning. 

Error correction helps learning. When we make a mistake and someone corrects us, we are supposed to change our conscious version of the rule. If a learner says, "I comes to school every day," and a teacher responds with "No, it's 'I come to school'," the learner is supposed to realize that the -s doesn't go on the first person singular. As we shall see, error correction and conscious learning are very limited. 

The Natural Order Hypothesis 

The natural order hypothesis claims that we acquire the parts of a language in a predictable order. Some grammatical items, for example, tend to be acquired early, while others come later. 

The order of acquisition for first and second languages is similar, but not identical. It has been established, for example, that the -ing marker in English, the progressive, is acquired fairly early in first language acquisition, while the third person singular -s is acquired later. The third person singular may arrive six months to a year after -ing. In adult second language acquisition, the progressive is also acquired early, but the third person singular may never come. It is common to hear people who speak English as a second language very well, and yet have not acquired the third person singular. 

The order is not exact: Not every acquirer proceeds in exactly the same order. But the variation among acquirers is not extreme. There clearly is an "average" order of acquisition. 

There are three amazing facts about the natural order phenomenon, and I will list them in their order of amazingness: 

1. The natural order is not based on any obvious features of simplicity and complexity. Some rules that look simple (e.g. the third person singular) are acquired late. Others that appear to linguists to be complex are acquired early. This presents a problem to curriculum designers who present rules to language students from "simple" to "complex." A rule may seem to be simple to a linguist, but may be late-acquired. 
2. The natural order cannot be changed. It is immune to deliberate teaching. We cannot alter the natural order by explanations, drills, and exercises. A teacher can drill the third person singular for weeks, but it will not be acquired until the acquirer is ready for it. This explains a great deal of the frustration language students have. 
3. One might suppose that the solution to our problems is simply to teach along the natural order: We need only find out which items are naturally acquired early and teach those first, etc.. The third amazing fact is that this is not the solution: The natural order is not the teaching order. But I am not yet ready to tell you why. 

The Monitor Hypothesis 

The Monitor hypothesis attempts to explain how acquisition and learning are used. Language is normally produced using our acquired linguistic competence. Conscious learning has only one function: As a "Monitor" or editor. 

Here is how it works: We are about to say something in another language. The form of Our sentence pops into our mind, thanks to our subconsciously acquired competence. Then just before we actually produce the sentence, just before we say it, we scan it internally, inspect it, and use our consciously learned system to correct errors. 

We can also use our conscious Monitor to correct sentences after we have produced them; this is called "self-correction." (Of course, we also self-correct, or edit, using our acquired system, or our "feel" for correctness. The Monitor hypothesis claims that conscious learning has only this function; it does not contribute to our fluency.) 

While the Monitor can make a small contribution to accuracy, the research indicates that acquisition makes the major contribution. Thus, acquisition is responsible for both fluency and most of our accuracy. 

It is difficult to use the Monitor. In order to use the Monitor successfully, three necessary conditions must be met: 

1. The acquirer must know the rule. This is a very difficult condition to meet. Research linguists freely admit that they do not know all the rules of any language. Those who write grammar texts know fewer rules than the linguists. Language teachers do not teach all the rules in the texts. Even the best students don't learn all the rules that are taught, even the best students don't remember all the rules they have learned, and even the best students can't always use the rules they do remember: Many rules are too complex to apply while engaging in conversation. 
2. The acquirer must be thinking about correctness, or focused on form. This is hard to do. It is hard to think about both form and meaning at the same time. 
3. The acquirer must have time. For most people, normal conversation doesn't provide enough time for the use of the Monitor. A few language experts can Monitor while conversing, but these are very advanced acquirers who only need to Monitor an occasional rule here and there, and who have a special interest in the structure of language. 

The research shows that Monitor use is only obvious when all three conditions are fully met. For most people, all three conditions are met only when we give them a grammar test! 

The Monitor is weak, but it is not useless. Some conscious knowledge of language can be helpful. Acquisition does not, typically, provide us with 100% of a language, there is often a small residue of grammar, punctuation and spelling rules that even native speakers do not acquire, even after extensive aural and written comprehensible input. In English, these can include the lie/lay distinction, the its/it's distinction, and spelling demons such as "separate," and commitment" (how many t's?). Because our standard for written communication is 100%, these aspects of language need to be learned, but they make up a small part of our language competence. 

We pay a price for the modest amount of accuracy we get from Monitoring. Some research shows that when we focus on form when speaking, we produce less information, and we slow down (Hulstijn and Huilstijn, 1982). This can seriously disrupt conversation. Some people "over-Monitor" and are so concerned with grammar and accuracy that speech is slow and painful to produce as well as to listen to: 

“The major, who had been the great fencer, did not believe in bravery, and spent much time while we sat in the machines correcting my grammar. He had complimented me on how I spoke Italian, and we talked together very easily. One day I had said that Italian seemed like such an easy language to me that I could not take a great interest in it; everything was so easy to say. 'Ah yes,' the major said, 'Why then, do you not take up the use of grammar?' So we took up the use of grammar, and soon Italian was such a difficult language that I was afraid to talk to him until I had the grammar straight in my mind." (E. Hemingway, In Another Country.) 

The best advice is, I think, to use the conscious Monitor when it does not interfere with communication, when we have time, as in the editing phase of writing. 

The Input (Comprehension) Hypothesis 

The input hypothesis attempts to answer the most important question in the fields of language acquisition and language education: How does language acquisition occur? The evidence strongly supports a simple hypothesis. We acquire language in only one way: When we understand messages, when we obtain "comprehensible input." We acquire language, in other words, when we understand what we hear or what we read, when we understand the message. 

In recent years, I have used the term "comprehension hypothesis" to refer to the input hypothesis. "Comprehension" is a better description - mere input is not enough, it must be understood. This term also allows me to honor the lineage of the input/comprehension hypothesis: The idea is certainly not new with me. In the field of second language acquisition, James Asher , Harris Winitz, and Robins Burling proposed similar ideas years before I did, and in the field of literacy Frank Smith and Kenneth Goodman had proposed that we learn to read by reading, by understanding the message on the page. 

Comprehensible input has been our last resort in language teaching: We have tried everything else: grammar rules, repetition drills, computers, etc.. The input hypothesis claims, however, that comprehending messages is the only way language is acquired. There is no individual variation in the fundamental process of language acquisition. 

The input/comprehension hypothesis can be restated in terms of the natural order hypothesis. Let us assume a very simple version of the natural order hypothesis, that we acquire the rules of a language in a linear order: 1,2, 3 .... The question of how we acquire language can be restated as: How do we move from rule 3 to rule 4. from rule 987 to 988? More generally, if "i" represents the last rule we have acquired, how do we move from "i" to "i+1", where i+1 is the next structure we are ready to acquire? 

The input hypothesis claims that we move from i to i+1 by understanding input containing i+1. We are able to do this with the help of our previously acquired linguistic competence, as well as extra-linguistic knowledge, which includes our knowledge of the world and our knowledge of the situation. In other words, we use context. (For a more detailed discussion of the role of context, including the issue of what happens when context is "too rich," and no linguistic processing is necessary, see Krashen, 1999a.) For beginners, pictures are a tremendous help in making input comprehensible, as well as the body movements that are at the core of Asher's Total Physical Response method. 

Now that we have some idea of the input/comprehension hypothesis, I can share two mystical, amazing facts about language acquisition. First, language acquisition is effortless. It involves no energy, no work. All an acquirer has to do is understand messages. Second, language acquisition is involuntary. Given comprehensible input, and a lack of affective barriers (see below), language acquisition will take place. The acquirer has no choice. 

In a theoretical sense, language teaching is easy: All we have to do is give students comprehensible messages that they will pay attention to, and they will pay attention to them if the messages are interesting. 

Corollaries of the Input/Comprehension Hypothesis 

If the input hypothesis is correct, the following corollaries are correct: 

Talking is not practicing 

The input hypothesis maintains that speaking does not directly result in language acquisition: talking is not practicing. If you practice your French out loud every morning in front of the mirror, your French will not improve. Rather, the ability to speak is the result of language acquisition, not a cause. 

Speaking can help language acquisition indirectly, however, First, it can result in conversation, and conversation is an excellent source of comprehensible input. What counts in conversation, however, is what the other person says to you, not what you say to them. I suspect that speaking can help in another way: It can make you feel more like a user of the second language, like a member of the "club." I return to this argument a bit later, in the section on the Affective Filter. 

Given enough comprehensible input, i+7 is present 

The second corollary states that if we provide students with enough comprehensible input, the structures they are ready to acquire will be present in the input. We don't have to make sure they are there; we don't have to deliberately focus on certain points of grammar. If this corollary is correct, it means the end of grammatically based language teaching. 

Before discussing this, it is important to emphasize that grammatical accuracy is an important goal. What we are discussing is how to attain this goal. I am arguing that comprehensible input is a better way of developing grammatical accuracy than direct instruction in grammar. 

We all remember grammatically based classes. Students focus on one rule at a time, the idea being to "master" one rule and then move on to the next. It simply doesn't work. I will discuss four problems with the grammatically-based syllabus, problems that I think are unsolvable for the grammatical syllabus, but that comprehensible input solves with ease. 

1. What if a student misses class one day? If the class is based on grammar, the student has  missed the rule of the day." If the class is based on comprehensible input, however, there is no problem. Every class will contain a rich supply of grammar and vocabulary, and there will be plenty of chances for the student to get comprehensible input containing i+1. With grammar-based teaching, the student gets only one chance, unless review is constant and extensive. With comprehensible input, there are many many chances. 
2. Even though we all acquire language in the same way, there is individual variation in rate  of acquisition. Some students in a class will progress faster than others. Individual variation in rate is especially likely in second language classes; some students get more input outside of class than others. If the "rule of the day" is the past tense, some students may have already acquired it, and some may be nowhere near ready. With comprehensible input, everybody is covered, even though i+1 may be different for different students. We need not know exactly where each student is in his or her developmental path; all we need to do is to provide a great deal of comprehensible input.
3. In order to teach grammar, a teacher has to know grammar, and this is a task that is getting harder every day. With each new discovery, with each new grammatical rule, each new rule of sociolinguistic competence, the curriculum gets more and more complex. And it will never end. But if instruction is based on comprehensible input, this problem disappears. If comprehensible input is plentiful, students will absorb the rules teachers and  good authors use, whether teachers consciously know the rules or not, whether linguists have discovered them or not. 
4. The final problem with grammar teaching is the most serious: It's boring. It is very hard to say things that are interesting and comprehensible when your hidden agenda is the relative clause. But if instruction is based on comprehensible input, all we need to do is to present messages that are interesting and comprehensible, and grammar will take care of itself. As most teachers know, this task is difficult enough. 

We can now return to the third "amazing fact" about the natural order hypothesis. With comprehensible input-based language teaching the syllabus is not based on the natural order. If the arguments presented in this section are correct, the syllabus is not based on any grammatical order. Rather, students will acquire the language in a natural order as a result of getting comprehensible input. 

The Affective Filter Hypothesis 

The affective filter hypothesis claims that affective variables do not impact language acquisition directly but prevent input from reading what Chomsky has called the "language acquisition device," the part of the brain responsible for language acquisition. If the acquirer is anxious, has low self-esteem, does not consider himself or herself to be a potential member of the group that speaks the language (see Smith, 1988 for discussion of this last factor), he or she may understand the input, but it will not reach the language acquisition device. A block, the affective filter, will keep it out. 

The presence of the affective filter explains how two students can receive the same (comprehensible) input, yet one makes progress while the other does not. One student is "open" to the input while the other is not.