Definition:
The Levels of Processing (LoP) framework, proposed by Fergus Craik and Robert Lockhart (1972), holds that the durability of a memory trace is determined by the depth at which it is processed during encoding, not by how long it is held in short-term memory or whether the learner intends to remember it. Shallow processing (structural, surface-level) creates weak, transient traces; deep processing (semantic, meaning-based) creates strong, durable ones.
In-Depth Explanation
The Framework
Craik and Lockhart challenged the then-dominant multi-store model of memory (Atkinson & Shiffrin, 1968), which emphasized the location of a memory (sensory → short-term → long-term) and the time spent rehearsing it. Instead, they proposed that what matters is the type of processing:
- Structural/visual processing (shallowest): Attending to the physical form — typeface, capitalization, letter shape
- Phonological processing (intermediate): Attending to the sound — rhyming, pronunciation, syllable count
- Semantic processing (deepest): Attending to meaning — definitions, associations, context, personal relevance
The Classic Experiment
In the orienting task paradigm, participants answer different questions about words:
- “Is the word in uppercase?” → structural processing
- “Does the word rhyme with ‘train’?” → phonological processing
- “Is the word a type of vehicle?” → semantic processing
Despite identical exposure time and no instruction to memorize, participants recall semantically processed words at 2-3× the rate of structurally processed words.
Refinements and Criticisms
The framework has been refined since 1972:
- Elaboration (Craik & Tulving, 1975): Within a given depth, more elaborate processing (richer associations, more connections) produces better retention
- Distinctiveness: Items encoded in a unique or unusual way are better remembered
- Transfer-appropriate processing (Morris et al., 1977): Memory is best when the type of processing at encoding matches the type of processing at test — if you’ll be tested on sounds, phonological encoding may outperform semantic
- Involvement Load Hypothesis (Laufer & Hulstijn, 2001): Applies LoP to L2 vocabulary, operationalizing “depth” as need, search, and evaluation components
Application to Language Learning
LoP directly explains why some study methods work and others fail:
| Study Method | Processing Level | Expected Retention |
|---|---|---|
| Copying vocabulary lists | Structural | Very poor |
| Repeating words aloud | Phonological | Poor |
| Reading a definition | Shallow semantic | Moderate |
| Using the word in your own sentence | Deep semantic | Good |
| Explaining the word to someone | Deep + elaborative | Excellent |
| Sentence mining from authentic context | Deep + contextual | Excellent |
| Active recall testing | Deep + retrieval | Strongest |
Key Researchers
- Fergus Craik and Robert Lockhart — Original Levels of Processing framework (1972)
- Endel Tulving — Elaboration within levels; encoding specificity
- Batia Laufer and Jan Hulstijn — Involvement Load Hypothesis for L2 vocabulary