Reading Comprehension – Duolingo English Test
Advanced University‑Level Study Notes (MIT‑style)
Audience – Graduate students, advanced undergraduates, and high‑performing test‑takers
Goal – Understand the cognitive & linguistic mechanisms underlying reading, apply them to the Duolingo English Test (DET), and master edge‑case strategies.
1. Introduction
Reading comprehension is the cognitive ability to extract meaning, infer intent, and integrate information from written text.
The DET measures this ability through a series of short passages followed by multiple‑choice or short‑answer questions.
Unlike standard academic tests, the DET emphasizes real‑world language use, fast processing, and contextual reasoning.
2. Theoretical Foundations
| Level | What ≈ | Key References |
|---|---|---|
| Physiological | Visual‑orthographic mapping → Wernicke’s & Broca’s areas → phonological loop | Froehlich & Brown (2000), Siegler & DeLoache (2014) |
| Cognitive | Interactive‑Compensatory Model: decoding ↔ comprehension → mutual reinforcement | Kintsch & Rawson (2005) |
| Processing | Dual‑Process: shallow (lexical) vs deep (semantic) | Paivio (1971), Cowan (2008) |
| Working Memory | Cowan’s 4‑store model: sensory, phonological, visuospatial, central executive | Cowan (2010) |
| Pragmatics | Gricean maxims + Relevance Theory | Grice (1975), Sperber & Wilson (1995) |
Key Points
- Decoding ↔ Comprehension
Successful comprehension requires fast, accurate decoding so that neural resources can be re‑allocated to semantic integration. - Working Memory Capacity (WMC)
WMC predicts reading speed and comprehension depth. Strategies that reduce WMC load (e.g., chunking, using schema) improve performance. - Inference as a Cognitive Process
Inference is predictive (Top‑Down) and data‑driven (Bottom‑Up). The predictive coding framework explains how readers fill gaps efficiently.
3. The Reading Process in Detail
3.1 Decoding & Lexical Access
- Orthographic → Phonological → Semantic mapping.
- Levenshtein distance of words influences decoding speed.
- Lexicon size and frequency effects: high‑frequency words are retrieved in ~80 ms; low‑frequency in ~200 ms.
3.2 Syntax & Parsing
- Bottom‑up parsing: build syntactic tree from words → constituents.
- Top‑down parsing: anticipate structure