Skill acquisition theory, the missing link between knowing and doing in your Spanish classroom.

Robert DeKeyser borrowed the most useful framework cognitive psychology had on offer (John Anderson's ACT framework) and applied it to language learning. The result explains why your students can pass a grammar quiz on Monday and produce broken Spanish on Friday.

The three stages

John Anderson's ACT framework (The Architecture of Cognition, Harvard 1983; ACT-R, 1993) is the dominant cognitive theory of how humans learn skills. The framework identifies three stages every skill passes through, from cooking pasta to playing piano to driving stick.

Declarative knowledge

The stage where you know facts about the skill. The student knows that Spanish preterite for regular -ar verbs ends in é, aste, ó, amos, asteis, aron. They can recite the chart. They can fill in a worksheet. They cannot yet produce the forms in conversation. The knowledge is explicit and slow.

Procedural knowledge

The stage where you can use the knowledge to act, with effort. The student forms preterite verbs in a sentence but needs three or four seconds, often pausing, often with visible cognitive load. The knowledge is being applied. It is not yet automatic.

Automatized procedural knowledge

The stage where the action runs without conscious effort. The student produces preterite forms in real-time speech without thinking about the ending. The mental resources are now available for what to say, not how to say it.

Why this matters for language

DeKeyser's argument (Practice in a Second Language, Cambridge 2007; Skill Acquisition Theory, in VanPatten and Williams 2015) is that most US Spanish classrooms move students through stage one but rarely all the way through stage three. The student learns the rule. The student does controlled exercises. The student passes the unit test. And then production tasks reveal that the procedural knowledge never automatized.

The reason is dose. Anderson's data on skill automatization across many domains shows that automatization requires extensive, varied practice over time. The classic ACT-R learning curves require hundreds of trials, distributed across sessions, with feedback. A typical Spanish 2 unit on the preterite might involve thirty to fifty production trials total. That is not enough to automatize. It is enough to procedure.

The crucial transition is from procedural to automatized knowledge, and it requires the kind of extensive, varied practice that classroom instruction rarely provides.

What automatization actually requires

DeKeyser's integration of Anderson into SLA produces four conditions for automatization. All four must hold.

1. Declarative knowledge must be in place first

You cannot automatize a procedure you do not have. The explicit teaching of the form is not optional. This is one of the points where DeKeyser parts company with the strong-input camp. Skill acquisition theory says the rule comes first, then the practice, then the automatization.

2. The practice must be communicative

Mechanical drills automatize mechanical drills. Spanish students who do hundreds of fill-in-the-blank items automatize the fill-in-the-blank skill. They do not automatize the underlying communicative use. The practice must look like real communication: meaning-focused, with a real reason to use the form.

3. The practice must be sustained

Across days. Across weeks. Across topics. The same form practiced in many different communicative contexts builds the generalization that drilling in one context cannot produce. This is why the cumulative-review approach from the spacing article matters here too.

4. The feedback must be calibrated

Errors during the procedural stage need correction (Lyster's prompts work well here). Errors during the automatization stage should be allowed to pass for meaning's sake. Over-correcting an automatizing student pulls them back into declarative attention and slows the automatization.

The pedagogical implications

Stop assessing only declarative knowledge

The unit test that asks students to fill in preterite endings measures declarative knowledge. It does not measure whether the procedure has automatized. A student with high declarative scores and low automatized production is your future Spanish 3 dropout. Build production tasks into every unit assessment.

Plan for the dose

If you want students to automatize the preterite by the end of the year, you need to estimate the trials. Roughly: the form needs to be used in real communicative production many hundred times across the year, in dozens of different contexts. That is not one unit of work. That is a year of recycling.

Use the procedural stage explicitly

Procedural practice (effortful, slow, conscious) is not a waste of time. It is the bridge to automatization. Many CI teachers skip this stage because it feels mechanical. The result is students who understand a lot and produce slowly. The bridge cannot be skipped.

Where skill acquisition theory fights with the strong-input camp

DeKeyser and Krashen disagree publicly. Krashen says explicit teaching of forms is not necessary because input alone produces acquisition. DeKeyser says input alone produces fluent comprehension but not automatized production. The honest reading of the evidence is that DeKeyser is closer to right for most US classroom contexts.

The reason: US classroom learners do not get the input volume that fully implicit acquisition requires. A child immersed for ten thousand hours can acquire implicitly. A student getting four hours per week with you cannot. The explicit-procedural-automatized path is faster for the time-constrained learner. The strong-input path is more appropriate for the immersion learner who has the hours.

This is not a knock on input. Input is the substrate. The explicit-procedural path is built on the input. They are not in conflict if you stop framing them as a binary.

The heritage-learner footnote

Heritage learners often arrive with automatized procedural knowledge for conversational Spanish but no declarative knowledge of the rules behind it. The skill acquisition sequence runs in reverse for them. They can do something they cannot explain. The challenge is that academic Spanish (formal register, written conventions, certain verb forms) may require the standard declarative-procedural-automatized path even when the spoken Spanish is already there. This is one of the clearest cases for tailored heritage instruction.

What this means for your Monday

Pick a form your students nominally know and reliably misuse in production. For my Spanish 3 it is the preterite-imperfect contrast. For your students it might be something else.

Audit the practice trials your students have done with that form in the last six weeks. Count actual production instances (not exposures, not fill-ins). If the number is under fifty, the procedural-to-automatized transition has not had the dose to happen. Plan the next four weeks to deliberately raise the dose. Twice a week, ten minutes of production practice on that one form, in communicative tasks. By the end of the four weeks, recount. The automatization curve takes time. The dose is what feeds it.