The Declarative-Procedural (DP) Model, developed by Michael Ullman and colleagues from approximately 1997 onward, proposes that the two major memory systems known from cognitive neuroscience — declarative memory and procedural memory — underlie distinct components of language knowledge. Declarative memory stores words, meanings, and factual/memorized forms (including irregular verb forms); procedural memory stores combinatorial grammatical rules that generate regular forms. The model makes distinctive predictions about first language acquisition, second language acquisition, and language deficits in clinical populations.
In-Depth Explanation
Declarative memory (the “mental lexicon”)
Declarative memory is explicit, consciously accessible, involves the hippocampus and associated medial temporal lobe structures, and stores learned associations between forms and meanings. In language, it underlies:
- Vocabulary knowledge — word meanings, phonological forms, and mappings
- Irregular morphological forms (e.g., “went,” “taught,” Japanese iru/aru distinction)
- Any memorized form that is stored as a whole unit rather than computed via rule
Declarative memory is flexible — it can associate novel forms quickly — but may be slower to retrieve in real-time production at early learning stages.
Procedural memory (the “mental grammar”)
Procedural memory is implicit, operates largely below conscious awareness, relies on basal ganglia-frontal circuits, and underlies the automatic processing of sequential and rule-governed operations. In language:
- Regular morphological inflection (add -ed, add -s)
- Syntax — building hierarchical phrase structure
- Phonological sequencing rules
Procedural memory is slower to form but, once established, extremely fast and automatic in operation.
Predictions for L1 vs L2
A distinctive prediction of the DP model is that children and adult L2 learners rely differently on the two systems:
| Learner type | Declarative | Procedural |
|---|---|---|
| Native speakers (post-critical period) | Irregular forms, vocabulary | Grammar, regular morphology |
| L2 learners (adult onset) | Both grammar AND vocabulary — even rules may be stored declaratively initially | Develops with experience and automatization |
The model predicts that adult L2 learners will tend to store grammatical rules declaratively at first (treating even regular forms as memorized patterns), and will develop procedural automatization for grammar only with extensive practice. This aligns with findings that explicit grammar knowledge gradually becomes automatized — bridging declarative and procedural systems.
Clinical evidence
Ullman’s group drew substantial support from clinical populations: children and adults with Specific Language Impairment (SLI) show selective deficits in regular morphology (procedural system) while irregular forms are relatively intact; Alzheimer’s patients show the opposite pattern (declarative deficits, regular morphology preserved). This double dissociation supports the model’s neurological claims.
Relationship to practice and automatization
The DP model provides a theoretical account of why repetitive, meaningful practice leads to automatization: initially declaratively-mediated grammatical knowledge gradually shifts to procedural processing. This has implications for instruction — suggesting that explicit grammar instruction (declarative) can be a precursor to implicit, automatized use (procedural) with sufficient practice.
History
The Declarative-Procedural model emerged from Michael Ullman’s collaboration with Steven Pinker and others in the 1990s on the Words and Rules model of morphology. Ullman’s 2001 paper “The neural basis of lexicon and grammar in first and second language: The declarative/procedural model” (Bilingualism: Language and Cognition) provided the comprehensive L1/L2 theoretical statement. The model drew on a growing neuroscience literature on hippocampal vs. basal ganglia memory systems and their independent roles in cognition. Subsequent years saw applications to SLI, Parkinson’s disease, autism, dyslexia, and adult SLA. The DP model competes with and complements other dual-system models, including the work of Nick Ellis on explicit/implicit learning and the Adaptive Control of Thought-Rational (ACT-R) model of skill automatization. It remains an active and cited framework in neurolinguistics and applied linguistics.
Common Misconceptions
- “Declarative = explicit and procedural = implicit.” While there is overlap, the mapping isn’t perfect. Declarative memory is consciously accessible but can operate rapidly after practice. The key distinction is neural substrate (hippocampus vs. basal ganglia), not just conscious awareness.
- “The model means grammar and vocabulary can’t interact.” The model is about different memory systems for different aspects of language knowledge, not a rigid separation. In naturalistic acquisition, both systems operate continuously.
- “Adult L2 learners can only ever store grammar declaratively.” The model predicts that with sufficient practice, procedural automatization of grammar is possible for adult L2 learners — though the timeline and ceiling may differ from L1.
Social Media Sentiment
The Declarative-Procedural model is rarely discussed by name in language-learning communities, but its implications surface constantly in debates about explicit grammar study vs. immersion, and about whether studied rules ever become “natural.” Arguments that “you need to drill grammar until it’s automatic” map onto the procedural memory concept. The model provides theoretical backing for approaches that combine explicit instruction with massive practiced use — a middle ground between pure grammar study and pure immersion.
Last updated: 2026-04
Practical Application
- For grammar instruction: Initial explicit study builds declarative knowledge; subsequent massive practice (reading, writing, speaking, listening) is required to drive automatization into procedural memory.
- For Japanese learners: Japanese morphological patterns (verb conjugations, て-form, conditional forms) may initially be stored declaratively (learned as individual paradigm entries) and only become procedurally automatic after thousands of encounters. Sentence mining and extensive reading serve this automatization process.
- Irregular forms: Japanese irregular verbs and irregular polite forms (する→します、くる→きます) are stored declaratively — flashcard-style memorization is appropriate for these by the model’s prediction.
- Patience with automatization: Even if an explicit rule is “known,” fluent procedural deployment takes extended time. This explains the gap between “knowing the grammar” and “being able to use it.”
Related Terms
- Explicit Learning
- Implicit Learning
- Automaticity
- Cognitive Load Theory
- Input Processing
- Procedural Knowledge
See Also
- Sakubo – Japanese Study — SRS app for building Japanese vocabulary and grammar through spaced repetition, supporting both declarative and procedural knowledge development.
Sources
- Ullman, M.T. (2001). The neural basis of lexicon and grammar in first and second language: The declarative/procedural model. Bilingualism: Language and Cognition, 4(2), 105–122. — foundational statement of the full model for L1 and L2.
- Ullman, M.T. (2004). Contributions of memory circuits to language: The declarative/procedural model. Cognition, 92(1–2), 231–270. — detailed neural substrate analysis across language components.
- Pinker, S. & Ullman, M.T. (2002). The past and future of the past tense. Trends in Cognitive Sciences, 6(11), 456–463.01990-3) — Words and Rules context from which the DP model emerged.