Krashen’s Monitor Model has been extremely influential, largely as a result of its easy applicability to teaching practices. It makes a short list of memorable claims, and was fairly revolutionary when held up against the pedagogical norms of its day. As a scientific model of acquisition, however, it comes up short.
The Monitor Model is composed of five hypotheses:
- The Acquisition-Learning Hypothesis
- The Monitor Hypothesis
- The Natural Order Hypothesis
- The Input Hypothesis
- The Affective Filter Hypothesis
Each of these makes very specific claims that are easy to apply, but fairly trivial to challenge. Although they are all worth examining, I’ll take up just the first two here.
The Acquisition-Learning Hypothesis posits that there is a fundamental difference between “Learning” and “Acquisition”, and that learning cannot lead to acquisition. We can easily apply this to classrooms to discourage teachers from using grammar drills, which is basically a good thing. As a model of L2 acquisition, it has the major flaw of positing a strong position—learning never leads to acquisition—which flies in the face of the experience of many language learners. He may not be wrong, but he certainly needs to provide some strong evidence to convince those of us who have seen a lot of improvement from overt study, and that evidence is lacking.
The Monitor Model posits a role for overt Learning. According to Krashen, overtly studied knowledge is applicable only in monitoring one’s own production for errors, and only when given sufficient time, focus, and knowledge of the rules. Here again we have an idea with clear application to the language classroom: if a teacher wants the students to attend to the accuracy of their speech, they need time, focus, and instruction. As a model of acquisition, it simply fails to account for all of the available data. Many of us are able to use learned knowledge to slowly form sentences, rehearse them in what Baddeley called the “articulatory loop”, and produce them as needed. It also fails to account for the application of learned knowledge in receptive language use, as Gregg points out in his 1984 take-down of the Monitor Model. It even conflicts with the Acquisition-Learning Hypothesis, since the Monitor is clearly a tool for advancing acquired language, yet it is informed by “Learning”. Even within Krashen’s own model, Learning has a role in advancing Acquisition.
The Monitor Model shows one thing extremely clearly: the popularity of any model among laypeople is at best only tangentially related to its basis in evidence. Certainly we would reject the Monitor Model out of hand if it failed to describe the majority of what we see in reality, and to its credit, it does describe an awful lot of what we see among language learners.
Krashen’s model owes most of its success to its applicability. An idea that can be put into practice tends to spread, whereas less practicable ideas only find traction with the true believers. Richard Dawkins compares ideas to biological genes, calling them “replicators”, and like genes, ideas spread or die out based on their ability to create copies of themselves, resist other ideas, and avoid extinction.[i]
Compare the spread of Christianity to the spread of Jainism, for an example of idea replicability. “Pray, believe and be saved” is much easier to put into practice than “Never kill, not even insects”. Likewise, “Give a lot of input, cut back on grammar exercises, and reduce student stress” is an idea with greater spreading power than the more subtle idea of “Each technique has its place, and teachers must learn to employ a wide toolbox with sensitivity to the learning styles and needs of his students.” The Monitor Model is easy to put into practice, so teachers use it. The idea replicates.
Krashen’s Monitor Model is very self-contained. “Accept these five ideas, or accept none of them,” he says. Each hypothesis supports the other and forms a neat little package. This leaves very little room to alter or adapt the model, insert other ideas into it, or abandon certain aspects in favor of others. The Monitor Model resists other ideas and avoids extinction.
None of which gets at the actual purpose of a scientific model. A model, in the scientific sense, is meant to explain as much of the available data as is possible, in as parsimonious a manner as is possible. For all its flaws, phonology’s Optimality Theory is much more scientific than the Monitor Model. It succeeds in explaining large swaths of the data available on speech sounds, and it does so with a minimum of rules. In syntax, X-bar theory similarly explains a great number of linguistic processes in a minimal manner. The Monitor Model, on the other hand, seems to start from an approach and work backward toward a model, selecting data that fits. Simply looking at available evidence, it’s hard to claim that learning functions exclusively in monitoring; rather, the Monitor Hypothesis seeks patch a hole opened by the Acquisition-Learning Hypothesis. This doesn’t make it wrong, necessarily, but it does make it unscientific.
[i] Richard Dawkins, The Selfish Gene, Ch. 11. Oxford Univesity Press, 1976