Week 6: Syntactic Sentence Processing

Question 1

How does complexity effect sentence processing?

Answer 1

Early transformational grammar proposed that sentences requiring more transformations (e.g., passives, negatives, object‑relative clauses) are harder to process. 

Question 2

subject relative and object relative clauses:

Answer 2

(10.5) The boy who chased the dog ran home -> subject relative 

(10.6) The boy who the dog chased ran home -> object relative 

Question 3

Evidence for complexity effecting sentence processing

Answer 3

Sentence–picture matching: passives and negatives took longer.

Transformation tasks: more complex transformations → longer response times.

Memory tasks: sentences closer in transformational structure were more confusable and harder to remember

Question 4

Critiques of experiments measuring how sentence complexity effects processing

Answer 4

These tasks don’t reflect natural comprehension.

People rarely transform sentences explicitly.

Exposure frequency (e.g., actives > passives) also matters.

Question 5

What is the clausal hypothesis - how does this relate to how we process sentences?

Answer 5

The hypothesis: clauses are the basic units of analysis in language comprehension

Clause = a group of words in a sentence that includes a verb 

The argument is that we chunk sentences into clauses in order to process them

Question 6

How do click-location experiments provide evidence for the clasual hypothesis?

Answer 6

Participants given a printed version of a sentence. They then listened to the sentence and a click was inserted at a point that was in the middle of the word/ not a clear boundary. The participants then had to mark on the written sentence where they heard the click.
Finding = Clicks were mis‑located toward clause boundaries, suggesting perceptual segmentation into clauses.

Question 7

Critiques / counter argument of the clausal hypothesis

Answer 7

Distinguish clausal structuring (segmenting) from clausal processing (processing only at boundaries).

Word‑monitoring studies show processing is incremental, not restricted to clause boundaries.

“Sentence processing does not need to wait until major structural boundaries but can take place in a cumulative way as a sentence is heard.”

Question 8

What did findings show in regards to the clasual hypothesis versus cumulative facilitation for normal, anomalous and scrambled prose?

Answer 8

Normal prose = syntactically and semantically well formed

Anomalous prose = syntactically well-formed but has little meaning

Scrambled prose = neither syntactically nor semantically well-formed but consists of real words 

Normal prose shows cumulative facilitation; Anomalous prose also shows some, but scrambled prose does not -> in sentences that are both syntactically and semantically well formed as more of the sentence is heard words can be responded to more rapidly. This shows the sentence is being processed as it is heard rather than just at clause/ syntactic boundaries

Question 9

How do explicit syntactic markers help with sentence processing? What experiments show this?

Answer 9

Explicit markers (e.g., that, who, inflections) facilitate processing but these are often left out in English language.

 Findings:

Phoneme monitoring where participant have to find certain phonemes in the sentence: faster when markers are present.

Eye‑tracking: complementiser ‘that’ reduces ambiguity (e.g., John knew (that) the answer was wrong).

Markers are frequent, short, and provide anchor points.

Question 10

How do prosody and punctuation support sentence processing

Answer 10

Prosody and punctuation also act as structural cues:

Prosodic phrasing helps only when syntactically motivated.

Punctuation and line breaks influence interpretation and reading difficulty. 

Can make it more difficult/ slower to read a sentence if there is a line break in an unnatural place

Question 11

Discontinuous constituents

Answer 11

Discontinuous constituents (I.e. phrase the belongs together syntactically but gets spilt apart by other material in the sentence e.g. Rang …Extra information about who… up) are hard to process.

Question 12

Structural preferences in readers

Answer 12

Readers show structural preferences, sometimes leading to misanalysis and the need to re-read when they realise their ‘prediction’ is not correct

Question 13

Garden‑path sentences
What are they?
What can be added to help with them?

Answer 13

induce an initially plausible but incorrect parse. 

Example:

“The horse raced past the barn fell.” Readers initially treat raced as a past‑tense verb, not a reduced relative.

Adding explicit syntactic markers can help us figure out what it means

“The horse which was raced past the barn fell.”

Question 14

The sausage machine model (Frazier & Fodor):

Answer 14

The sausage machine is a parser it analyses sentences according to their syntactic structure 

It builds a syntactic tree known as phrase marker incrementally.

The parser is deterministic (one structure at a time) -> contrast with parallel processing models where multiple interpretations can be entertained at the same time

It prefers simple structures and avoids leaving material unattached.

Question 15

List the two key parsing strategies

Answer 15

Late closure
Minimal attachment

Question 16

Late Closure

Answer 16

Attach new material to the current clause/phrase whenever possible.

In the sentences below people generally prefer the first one as it obeys late closure -  ‘the driver’ is attached to the first clause.

Question 17

Minimal Attachment

Answer 17

Build the structure using the fewest new nodes. In other words have the simplest structure possible. 

The first sentence is easier because it doesn’t have the extra S node

Question 18

Evidence for late closure and minimal attachment comes from..

Answer 18

Eye tracking

Eye‑tracking shows longer fixations and regressions when these preferred parses fail.

Question 19

Right Association strategy

Answer 19

New material attaches to the lowest possible node, unless other strategies (minimal attachment and late closure) override it

Question 20

Syntactic category ambiguity + strategies to over come it

Answer 20

Many words are category‑ambiguous (e.g., walk = noun/verb).

When structure doesn’t force a category, the parser may:

Delay attachment until disambiguation → faster reading during ambiguous region, slower at disambiguation (portion of the sentence that follows and make the ambiguous parts clear)

Or (in alternative models) build multiple structures in parallel and then choose

Eye‑tracking evidence supports delayed attachment in some cases

Question 21

are parsing strategies universal?

Answer 21

Parsing strategies are universal, but their relative weighting differs across languages depending on cue reliability.

Question 22

Difference in languages for sentence processing

Answer 22

English speakers rely on word order.

German speakers rely more on animacy.

Italian speakers rely on agreement morphology.

Question 23

How different languages respond to relative clause attachment ambiguity

Answer 23

English prefers Late Closure (attach to the most recent noun).

Spanish prefers high attachment (attach to the first noun).

With three‑noun structures, both languages prefer Recency.