Declarative-Procedural Model
The Declarative-Procedural (DP) Model, proposed by Michael Ullman (2001), applies the neuroscience of memory systems to language. The model posits that language depends on two distinct brain memory systems — declarative memory and procedural memory — and that L1 and L2 rely on these systems differently.
The Two Memory Systems
Declarative Memory
- Located primarily in the medial temporal lobe (including the hippocampus)
- Stores facts, events, and arbitrary associations
- Supports the mental lexicon — stored words, irregular forms, and memorised chunks
- Learning is explicit and relatively fast — a new word can be learned in a single exposure
- Accessible to conscious awareness
Procedural Memory
- Located primarily in the frontal cortex and basal ganglia
- Stores skills, habits, and sequential/hierarchical patterns
- Supports the mental grammar — rules for combining morphemes, words, and phrases
- Learning is implicit and gradual — requires extensive practice and exposure
- Not readily accessible to conscious awareness
The Lexicon-Grammar Distinction
In L1, the division of labour is clear:
- Lexicon (declarative) — irregular past tenses (went, broke), idioms, collocations, and individual word meanings are stored as wholes
- Grammar (procedural) — regular inflection (walk → walked), syntactic rules, and morphological composition are computed by the procedural system
This accounts for the double dissociation observed in brain-damaged patients: damage to temporal lobe structures impairs irregular forms and vocabulary while sparing regular grammar, and damage to basal ganglia/frontal structures impairs regular grammar while sparing vocabulary and irregular forms.
Application to SLA
The key insight for SLA is that L2 learners — particularly those who begin learning after childhood — initially rely heavily on declarative memory for both vocabulary and grammar:
- L2 grammar starts declarative — adult learners memorise grammatical rules as explicit facts ("add -ed for past tense") rather than acquiring them through the procedural system
- With sufficient exposure and practice, some grammatical knowledge may shift to the procedural system — this is the process of proceduralisation
- The shift is never complete for most learners — many aspects of L2 grammar remain declaratively mediated, explaining why even advanced learners often process L2 grammar more slowly and with more conscious effort than L1 speakers
This has a direct parallel with Skill Acquisition Theory (DeKeyser, 2007), which also proposes a shift from declarative to procedural knowledge through practice. The DP model provides the neuroscientific substrate for this shift.
Relationship to Implicit and Explicit Knowledge
The DP model maps onto the Implicit vs Explicit Knowledge distinction:
| Declarative | Procedural | |
|---|---|---|
| Type of knowledge | Explicit — consciously accessible | Implicit — not consciously accessible |
| Speed | Slower, effortful | Fast, automatic |
| Flexibility | Can be applied to novel contexts | Context-bound, requires practice |
| Error monitoring | Supports monitoring and self-correction | Operates below awareness |
Individual Differences
Ullman (2016) proposed that individual differences in L2 attainment can partly be explained by differences in the relative strength of the two memory systems:
- Learners with strong declarative memory may compensate for weak procedural learning by memorising more forms and rules explicitly
- Learners with strong procedural memory may develop more native-like automatic processing
- Sex differences in memory system strength may contribute to observed differences in L2 attainment patterns
Teaching Implications
- Early L2 instruction benefits from explicit presentation of rules (engaging declarative memory), followed by extensive practice to promote proceduralisation
- Vocabulary learning — both L1 and L2 — depends on declarative memory and benefits from spaced repetition, elaborative encoding, and contextual exposure
- Grammar proceduralisation requires massive amounts of meaningful practice — the shift from knowing a rule to applying it automatically is slow and effortful
- The model supports a PPP-like progression (present → practise → produce) as compatible with how the brain moves from declarative to procedural representation