Overview
This post is based on a YouTube interview by the Oxford University Linguistics Society with Professor Janet Dean Fodor (CUNY Graduate Center), focusing on the competence-performance distinction, the role of prosody in sentence processing, center embedding, acquisition, and the relationship between theoretical linguistics and psycholinguistics. The discussion is framed to align with a modular, nuts-and-bolts understanding of mental architecture, avoiding speculative theoretical claims while remaining empirically grounded.
1: Competence and Performance Revisited
The traditional competence–performance distinction maps closely onto the contemporary divide between theoretical linguistics and psycholinguistics.
Competence concerns the abstract knowledge of language: how sentences are structurally shaped and what configurations are licensed by grammar.
Performance concerns real-time language use: how speakers produce and comprehend sentences under temporal, cognitive, and articulatory constraints.
While competence seeks cross-linguistic generalizations and universals, performance must integrate phonology, syntax, semantics, and prosody simultaneously, under real-time pressure. These are not competing perspectives but distinct explanatory targets.
2: Phrase Length Effects and Psycholinguistic Evidence
A central empirical insight from psycholinguistics is the length effect:
Sentences that are grammatically well-formed can become harder to process when phrases are either too short or too long.
Processing difficulty does not track syntactic category alone, but rather linear length and grouping.
This insight opened a productive research path by shifting attention from abstract structural descriptions to measurable cognitive constraints.
3: Prosody as Mental Structure
Prosody is not merely acoustic ornamentation; it is a core component of mental representation.
Key claims:
Prosodic phrasing operates during silent reading, indicating an internalized prosodic representation.
Prosodic structure chunks sentences into cognitively manageable units.
These chunks are sensitive to length and balance, not syntactic depth.
Prosody, therefore, reflects how sentences are grouped mentally, rather than how they are hierarchically generated.
4: Center Embedding and the Syntax–Prosody Mismatch
Center-embedded sentences (e.g., clauses embedded within clauses) are grammatically valid but notoriously difficult to process.
Traditional explanations:
Memory limitations
Structural complexity
Empirical findings suggest an alternative:
Syntax allows unlimited embedding.
Prosody prefers flat, balanced structures.
Difficulty arises from the mismatch between nested syntax and linear prosodic chunking.
When prosodic balance is optimized, by adjusting phrase length, center-embedded sentences can become surprisingly acceptable.
5: Individual Differences in Sentence Processing
A striking finding is the presence of extreme individual variation:
Some speakers rapidly adapt and achieve natural prosodic phrasing.
Others fail to integrate structure even after repeated exposure.
This raises foundational questions:
Can prosodic competence be trained?
Do these differences correlate with broader phonological or cognitive abilities?
Such variation challenges one-size-fits-all models of processing.
6: The Sausage Machine and Chunking
Early work on sentence processing proposed the Sausage Machine model:
Sentences are parsed into chunks of approximately 6–7 words.
Chunking is governed by length rather than syntactic category.
This model anticipated later insights:
Chunk boundaries align with prosodic phrases.
Chunking is an active, mental process during comprehension.
Prosody, in retrospect, was always present, though initially unrecognized.
7: Prosody in Silent Reading and Beyond
Prosody appears to be ubiquitous:
Present during silent reading
Guiding comprehension
Structuring memory
This suggests that linguistic cognition may always involve an internal voice or signing equivalent, extending beyond spoken language and potentially into signed modalities.
8: Acquisition and the Poverty of the Stimulus
The poverty of the stimulus argument rests on a simple observation:
Learners acquire structures they have never encountered explicitly.
Challenges include:
Identifying what is not permitted without explicit negative evidence.
Generalizing from limited input.
Prosody may provide indirect constraints by making certain structures incompatible with production requirements.
9: Parameters as Treelets
In acquisition, parameters are reconceptualized as small structural units (treelets):
Each treelet represents a possible construction.
Language learning involves assembling these pieces incrementally.
Rather than abstract descriptive parameters, learners acquire usable structural fragments compatible with existing representations.
10: Grammar Evaluation and Subset Bias
Grammar learning is guided by evaluative pressures:
Preference for simpler structures
Subset bias: avoid overgeneration
Computational lattice models demonstrate how parameter values can be pruned incrementally, though their neurological plausibility remains open.
11: Minimalism and Psychological Plausibility
Minimalist syntax pursues evolutionary and formal simplicity, often at the expense of processing realism.
Key contrasts:
Minimalism strips away features.
Psycholinguistic models require feature-rich representations to guide integration.
Both approaches may be valid within their respective goals, but they answer fundamentally different questions.
12: Personal Touch
This section accentuates a central theme of this post:
Linguistic structure is real.
Mental implementation is constrained.
Prosody bridges grammar and cognition.
Understanding language requires respecting both the elegance of abstract systems and the practical realities of human processing.
