The Comprehension Hypothesis is consistent with empirical research. The Skill-Building Hypothesis is not. The Comprehension Hypothesis allows immediate gratification, that is, interesting and comprehensible input from the beginning. The Skill-Building Hypothesis is a delayed gratification approach. Nevertheless, researchers continue to search for evidence for skill-building, and practitioners are reluctant to provide more comprehensible input in their classrooms.
A Delayed Gratification Hypothesis
We have made a serious error in language education: We have confused cause and effect. We have assumed that students first need to consciously learn their "skills" (grammar, vocabulary, spelling), and that only after skills are mastered can they actually use these skills in real situations. This assumption, the "Skill-Building Hypothesis," insists on delayed gratification. Only after hard and tedious work do we earn the right to actually enjoy the use of language.
The Alternative: Comprehensible input
There is an alternative. It hypothesizes that "skills," or mastery of the components of language, is the result of one particular aspect of language use, comprehensible input. It claims that grammatical competence and vocabulary knowledge are the result of listening and reading, and that writing style and much of spelling competence is the result of reading. The Comprehension Hypothesis does not require delayed gratification. It claims that we can enjoy real language use right away: we can listen to stories, read books, and engage in interesting conversations as soon as they are comprehensible. The Comprehension Hypothesis, in fact, insists on pleasure from the beginning, on acquirers obtaining interesting, comprehensible input right from the start. The path of pleasure is the only path. The path of pain does not work for language acquisition. I have referred to the Comprehension Hypothesis as the Input Hypothesis in previous writing, a term that I do not reject. But "Comprehension Hypothesis" appears to be more precise—it is comprehension that counts, not simply input. Smith (1975) made this clear in the title of his book, Comprehension and Learning, pointing out that they are closely related: In order to learn anything (using the term "learn" here in the more general sense, not as contrasted with "acquisition"), we must first understand it. Once we have understood it, we have learned it.
The evidence for this alternative hypothesis is strong. It has been shown that comprehensible-input based methods are very successful when compared to methods based on skill-building; this research includes beginning and intermediate foreign language teaching, and the consistent positive impact of free voluntary reading (Krashen, 2003).
Problems with the Skill-Building Hypothesis
There are serious problems with the Skill-Building Hypothesis: The effects of deliberate, direct skill-based instruction are very weak and fragile. Studies claiming to show a positive effect for grammar study show only that grammar study makes a limited contribution to competence: Subjects in all of these studies have been experienced "grammar learners," are given extensive training, and make only modest progress on tests that focus them on the target form, which are usually given immediately after the treatment (Krashen, 2003; Truscott, 1998). In terms of theory, the conditions for Monitor use (Krashen, 1982) are met in these studies.
The systems involved (grammar, spelling, vocabulary, etc) are too complex to be consciously learned.
Numerous cases exist of "acquisition without learning," cases of people who have reached very high levels of competence without skill-based instruction. There are, however, no known cases of high levels of proficiency without comprehensible input.
The skill-building hypothesis is an "output" hypothesis, that is, it demands that students produce language in order to acquire it. Actual output, according to skill-builders, serves two functions: (1) it exposes our errors, which can then be corrected, and corrections are supposed to lead to better rules, and (2) repeated output is supposed to help us solidify or "automatize" our knowledge of rules. But the amount of output we produce, either in speech or writing, is far too small to account for more than a small fraction of what we eventually acquire. In addition, correction is infrequent and studies show that it has either no effect or a weak effect, with its impact only evident in studies in which students are able to focus on form on the posttest (Krashen, 2002; Truscott, 1996), that is, when the conditions for the use of conscious Monitor are met.
An alternative to both the Skill-Building and Comprehension Hypotheses is the Comprehensible Output Hypothesis, which claims that language is acquired when we produce it, fail to communicate our meaning, and then try again, eventually succeeding in communicating by using a form that is correct. This hypothesis suffers from findings showing that few instances of comprehensible output actually occur: There are few instances in which language acquirers fail to communicate and then re-formulate their message in a way that brings it closer to the correct target language form. Only one experimental study (Nobuyoshi & Ellis, 1993) has attempted to demonstrate that comprehensible output is effective: Despite the authors' claims, it did not. One of the three subjects in the study failed to make any gains, only one subject made a significant improvement, and it was quite possible that the "improvement" was due to increased Monitor use of a previously learned rule (see Krashen, 2003). The Comprehensible Output hypothesis was originally formulated as a supplement to comprehensible input (Swain, 1985), but there is no evidence that it plays even a small role in language acquisition.
Several combination approaches have been proposed. One "balanced" method insists that methods for beginners should be based on skill-building, and that communicative activities can be introduced after the beginning stage, in order to solidify or reinforce the skills that were learned. This is really a pure skill-building approach that denies the possibility of language acquisition and that assumes that all linguistic knowledge must result from skill-building instruction.
Another "balanced" method prescribes communicative activities from the beginning, but claims that comprehensible input alone is not enough: it needs "supplementation." The usual form supplementation takes is additional output and/or grammar activities.
Mason (2003) provided a direct demonstration of the inefficacy of output and correction as supplements to comprehensible input: In her study, three groups of adult EFL students participated in an extensive English reading program for three semesters. One group wrote brief summaries of what they read in Japanese, another wrote their summaries in English, and a third wrote summaries in English that were corrected, and they then rewrote the summaries. Those in the second and third groups had requested the treatment they received. Mason reported no differences in gains on three different measures of English among the three groups, and concluded that the group that wrote their summaries in Japanese was the most efficient, in terms of amount of English acquired and the total time devoted to English.
Some conscious knowledge of grammar can be of use in editing, in filling small gaps left by acquisition that even very advanced second (and first) language users seem to have. It appears to be the case, however, that there are severe limitations on the learning and use of this knowledge (Krashen, 1982).
The kinds of supplementation that can have a strong impact on language development are those that help students get more comprehensible input (e.g. discussion of books students may find of interest) or make input more comprehensible (e.g. provide background information). In other words, what will work are activities that deal with the cause of language acquisition and not the effects.
Despite this evidence, the major focus of current research is to continue to search for ways to demonstrate the effectiveness of the skill-building approach, a desperate search, in my view. There has been little interest in seeing the effects of increasing the quality and quantity of comprehensible input, even though many pay lip-service to the value of comprehensible input, claiming to support the "comprehensible input + supplementation" position. As a result of this negligence, I suspect that we have not even come close to tapping the potential of comprehensible input.
Why is this true? I discuss here only two of the possible reasons.
The ruthless capitalist argument
It could be the case that researchers are defending their own economic interests. They continue to search for a role for grammar not because they believe in it but because they have sold out to big publishers who make profits from grammar-based materials. I have no evidence that scholars have been deliberately dishonest, but the potential for conflict of interest exists.
The grammar-lover argument
Another reason for the determination to find a major role for grammar is the fact that so many researchers find the study of grammar fascinating. I think this is true: I know this from personal experience—I love grammar too. I enjoy learning about grammatical systems, and I get a feeling a deep satisfaction from successful monitoring as well as successfully applying a grammar rule to my output. Unlike some others, however, I have realized that I am a member of a tiny minority and that most people get their pleasures elsewhere.
Barriers to Using CI-based Methods
Even if practitioners are interested in using CI-based methodology, there are barriers to using it in the classroom.
The students made me do it.
Skill-building is the "common-sense" folk theory of language development, and it is reinforced by the fact that it is used in nearly all foreign and second language classes and is the basis for nearly all materials. Although skill-based teaching is not effective, students simply blame themselves for their lack of progress. When asked, adult students insist that they want all their errors corrected (Cathcart & Olsen, 1976), many feel that the study of grammar is very important (research reviewed in Krashen, 1994) and that we learn to speak another language by speaking it. It is of course difficult for teachers to resist this pressure, especially when doing communicative activities is sometimes perceived to be non-professional and a sign of ignorance of grammar. We must, however, realize that it is our professional responsibility to teach according to our convictions about how people acquire language. As Smith (1986) put it, engineers do not consider public opinion on how to build bridges, nor do surgeons allow the public to tell them how to perform operations.
Both a short- and long-term solution to this problem is to provide information to students on how language is acquired. This will justify methodology, provide an interesting topic for sheltered subject matter teaching, and give students the tools to continue to improve after the course is over. At a minimum, students should be informed that the skill-building hypothesis is in fact a hypothesis, not an axiom, and that other hypotheses exist.
The curriculum/text made me do it.
It is likely that many language teachers work in situations where the established curriculum is not in agreement with their personal view of how language is acquired. These teachers have several options: The first is simply to go along with the curriculum, suffering silently, or complaining only to one's peers. From my observations, this appears to be the most frequent reaction. Second, one can "close the door" and secretly do what one thinks is best. This may profit one's current students, but the current curriculum and the skill-building hypothesis receives undeserved credit. "Closing the door" thus perpetuates and strengthens the dominance of the skill-building approach. In addition, the publishers make the profit from unused texts while teachers spend their own money on supplementary materials.
The only constructive option is to be honest with our students and attempt to inform the public.
The tests made me do it.
Most language tests are based on the skill-building hypothesis; they test grammar, vocabulary, spelling, etc. It seems obvious to many people that the best way to study for these tests is to study grammar, vocabulary, spelling, etc. The research, however, tells us differently: Students in classes with more comprehensible input do better on such tests than those in traditional classes.
The best way to help students prepare is to provide massive amounts of comprehensible input in class, and provide the means for them to obtain comprehensible input outside of class (see below). It is quite possible that some direct instruction (e.g. test-taking strategies) may be helpful, but it remains an empirical question just how much and what kind of instruction is best.
The lack of real-world input made me do it.
As many have commented, there is a profound difference between second and foreign language education; in the former, there is plenty of input outside the classroom but in the latter there typically is not. For this reason, some teachers have opted for skill-building over comprehensible input, claiming that there is not enough time for "the real thing." But comprehensible input is more efficient, according to method-comparison studies mentioned earlier.
Also, note that for the beginner the situation is identical: outside world input is not comprehensible anyway. The real difference between the foreign and second language situation is at the intermediate level.
We can't reproduce the second language informal environment, but we can do much better, and the Comprehension Hypothesis gives us a clear idea of what to do: Foreign language students need better libraries, libraries filled with books, magazines, comics, as well as audiotapes and videotapes. It should be possible for second language acquirers to spend a great deal of time reading books and magazines, watching TV shows and films they are really interested in, and that are comprehensible. Such a facility should be open to the public, to make it possible for anyone to get comprehensible input in the second language of their choice whenever necessary or desired.
Even if rich sources of comprehensible input are unavailable for the EFL student, this is still not a valid reason for employing a method that is incorrect. Presenting and reinforcing a false view of how language is acquired will only make language acquisition unlikely (or extremely inefficient) when input is available.
Cathcart, R., & Olsen, J. (1976). Teachers' and students' preferences for correction of classroom conversation errors. In J. Fanselow & R. Crymes (Eds.), On TESOL '76 (pp. 41-53). Washington, DC: TESOL.
Krashen, S. (1982). Principles and practice in second language acquisition. New York: McGraw-Hill.
Krashen, S. (1994). The pleasure hypothesis. In J. Alatis (Ed.), Georgetown University round table on languages and linguistics (pp. 299-322). Washington, DC: Georgetown University Press.
Krashen, S. (1994). The input hypothesis and its rivals. In N. Ellis (Ed.), Implicit and explicit learning of languages (pp.45-77). London: Academic Press.
Krashen, S. (2002). The comprehension hypothesis and its rivals. In Selected papers from the Eleventh International Symposium on English Teaching/Fourth Pan-Asian Conference (pp. 395-404). English Teachers Association/ROC.
Krashen, S. (2003). Explorations in language acquisition and use: The Taipei lectures. Portsmouth, NH: Heinemann.
Mason, B. (2003). A study of extensive reading and the development of grammatical accuracy by Japanese university students learning English. Ed.D. Dissertation, Temple University, Osaka, Japan.
Nobuyoshi, J., & Ellis, R. (1993). Focused communication tasks and second language acquisition. ELT Journal, 47 (3), 203-10.
Smith, F. (1975). Comprehension and learning. New York: Holt Rinehart Winston Smith, F. (1986). Insult to intelligence. Portsmouth NH: Heinemann.
Swain, M. (1985). Communicative competence: Some roles for comprehensible input and comprehensible output in its development. In S. Gass & C. Madden (Eds.), Input in Second Language Acquisition (pp. 235-266). New York: Newbury House.
Truscott, J. (1996). The case against grammar correction in L2 writing classes. Language Learning, 46 (2), 327-69.
Truscott, J. (1998). Noticing in second language acquisition: A critical review. Second Language Research, 14 (2), 103-135.
Stephen Krashen is currently Emeritus Professor of Education at the University of Southern California. Krashen is best known for his work in establishing a general theory of second language acquisition, as the co-founder of the Natural Approach, and as the inventor of sheltered subject matter teaching. His most recent books include Explorations in Language Acquisition and Use (Heinemann), Condemned Without a Trial: Bogus Arguments Against Bilingual Education (Heinemann), and Three Arguments Against Whole Language and Why They are Wrong (Heinemann). He also holds a black belt in Tae Kwon Do and was the 1978 Incline Bench Press Champion of Venice Beach California.