Most readers stumble on the final word of 'The horse raced past the barn fell.' What does this garden-path effect reveal about sentence parsing?
AReaders wait until the sentence is fully heard before assigning grammatical structure
BThe parser immediately commits to one structural interpretation and must expensively reanalyze when that interpretation fails
CReaders rely on semantic plausibility to correctly assign structure in real time, preventing such errors
DThe sentence violates universal grammar constraints and is therefore unprocessable
The garden-path effect is direct evidence that parsing is incremental and immediate, not optimal and wait-and-see. The parser applies a minimal-attachment heuristic — treating 'raced' as the main verb — before reaching 'fell,' which makes that analysis impossible. At 'fell,' costly reanalysis must occur, producing the characteristic reading disruption. If readers waited for all input before assigning structure, garden-path effects would not exist. Eye-tracking confirms this: fixation times and regressions spike at the disambiguating word.
Question 2 Multiple Choice
Center-embedded sentences like 'The reporter that the senator that the lobbyist attacked praised wrote the story' are nearly incomprehensible despite being grammatical. The best explanation is:
AThey use unusual vocabulary that readers haven't encountered before
BThe parser applies different grammatical rules to multiple levels of embedding
CMaintaining partially-built syntactic structures across multiple interrupting clauses exhausts working memory capacity
DCenter-embedding beyond one level violates syntactic rules in most languages
Center-embedding requires holding the first noun phrase ('reporter') available in working memory across two interrupting clauses before the main verb ('wrote') finally arrives. Each additional level of embedding adds another unresolved dependency that must be maintained simultaneously. This rapidly exceeds the capacity of working memory — not because anything is grammatically wrong, but because the parser's workspace has finite capacity. Right-branching sentences with equivalent complexity are easy because each clause can be resolved before the next begins, avoiding simultaneous storage demands.
Question 3 True / False
Syntax-first models of parsing (like Frazier's garden-path theory) claim that the parser initially uses most available information — including semantic plausibility, discourse context, and word frequency — to build sentence structure.
TTrue
FFalse
Answer: False
Syntax-first models claim the opposite: the parser uses only syntactic information to build an initial structure, applying heuristics like minimal attachment without consulting semantics or context. Semantic plausibility and context are consulted only after an initial structural analysis is complete. This is exactly what produces garden-path effects in semantically implausible or contextually inappropriate sentences. Interactive models challenge this by arguing that all information is used simultaneously from the start — but evidence supports a nuanced middle ground where syntax has priority, not exclusivity.
Question 4 True / False
Garden-path effects, as measured by longer fixation times and backward regressions in eye-tracking studies, provide evidence that the parser commits to an initial structural interpretation before the sentence is complete.
TTrue
FFalse
Answer: True
Eye-tracking during reading allows millisecond-level measurement of where readers look and for how long. At the disambiguating word in a garden-path sentence, fixation times spike and regressions (backward eye movements) increase — behavioral signatures of reanalysis. If the parser held multiple interpretations open simultaneously (a parallel parser), we would expect little or no disruption at the disambiguating word. The consistent disruption is strong evidence for serial commitment to an initial parse followed by effortful reanalysis.
Question 5 Short Answer
Why does the garden-path effect challenge the idea that sentence parsing is an optimal, wait-and-see process? What does it reveal about how the parser actually works?
Think about your answer, then reveal below.
Model answer: The garden-path effect shows that the parser commits to one structural interpretation immediately, before sufficient information is available to guarantee that interpretation is correct. It applies syntactic heuristics (like minimal attachment) incrementally, word by word, rather than waiting for more input. When the initial interpretation turns out to be wrong — as 'fell' reveals in 'The horse raced past the barn fell' — the parser must abandon its committed analysis and reanalyze from scratch, a cognitively costly process. An optimal, wait-and-see parser would hold multiple structural hypotheses open until disambiguating information arrived, and would therefore show no disruption at the disambiguating word. Garden-path effects prove this doesn't happen.
The deeper implication is that the parser trades correctness for speed: immediate commitment is faster and requires less memory than holding multiple parses open, but it fails on structurally unusual sentences. This reflects a general principle in cognitive psychology — the mind uses heuristics that are fast and usually right, at the cost of systematic errors in specific cases. The garden-path effect is one of the cleanest demonstrations of this tradeoff in language processing.