Competition Model

The Competition Model, developed by Elizabeth Bates and Brian MacWhinney (1982, 1989), is a functionalist, usage-based theory of language processing and acquisition. Its central claim is that learners use probabilistic cues — such as word order, animacy, agreement morphology, and case marking — to interpret sentences, and that these cues are weighted differently across languages. L2 acquisition involves learning to reweight cues from L1 values to L2 values.

Core Concepts

Cues and Cue Validity

When interpreting a sentence, listeners must determine who did what to whom. Languages provide multiple cues to help:

• Word order — in English, the first noun is typically the agent
• Animacy — animate nouns are more likely to be agents
• Agreement — verb morphology may indicate which noun is the subject
• Case marking — in languages like German or Japanese, case particles directly mark grammatical roles

Cue validity is the product of two factors:

• Cue availability — how often the cue is present in the input
• Cue reliability — when the cue is present, how often it correctly indicates the intended interpretation

Different languages weight cues differently:

Language	Dominant cue	Secondary cues
English	Word order	Animacy
Italian	Agreement	Animacy, word order
Japanese	Case marking	Animacy, word order
German	Agreement, case	Animacy, word order

Competition

When cues converge (all point to the same interpretation), processing is fast and accurate. When cues conflict — e.g., word order suggests one noun as agent but animacy suggests another — the cues compete, and the one with higher validity in that language wins. Processing is slower and more error-prone when cues conflict.

Application to SLA

The Competition Model makes specific predictions about L2 transfer and development:

1. L1 cue weights transfer to L2 — early L2 learners interpret L2 sentences using L1 cue weightings. An English speaker learning Japanese will initially over-rely on word order and under-use case markers.
2. Cue reweighting is gradual — learners progressively adjust cue weights as they receive more L2 input. The shift is driven by the statistical properties of the L2 input — consistent with Usage-Based Theory and frequency-based accounts.
3. Forward transfer and backward transfer — L1 cue weights affect L2 processing, but L2 learning can also modify L1 processing (backward transfer or attrition).
4. Difficulty is predictable — structures where L1 and L2 cue hierarchies differ most will be hardest to acquire.

The Unified Competition Model

MacWhinney (2005, 2012) expanded the framework into the Unified Competition Model (UCM), which integrates insights from connectionist learning, sociocultural approaches, and neurolinguistic research. The UCM addresses not only sentence processing but also phonology, lexis, and pragmatics, and incorporates factors like:

• Resonance — connections between L1 and L2 forms that share features (cognates, similar structures)
• Parasitism — initial L2 learning parasitises on L1 representations before developing independent L2 representations
• Entrenchment — L1 patterns become deeply entrenched through years of use, making them resistant to change in L2

Strengths

• Empirically testable — the model generates specific predictions about processing speed and accuracy that can be measured in real time (e.g., via reaction time experiments)
• Cross-linguistically grounded — the same framework applies to any language pair
• Explains Language Transfer as a natural consequence of cue reweighting, not as error or interference
• Compatible with Input Processing theory (VanPatten) — both emphasise how learners process input, though they focus on different aspects

Criticisms

• Focuses primarily on sentence-level interpretation and has less to say about production or discourse-level phenomena
• The concept of cue validity, while intuitive, is difficult to calculate precisely for real languages
• Does not address the role of explicit instruction or conscious noticing in cue reweighting