Transcript Garnsey presentation 02_23_06

Results of Eyetracking & Self-Paced
Moving Window Studies
DO-Bias Verbs:
The referees warned the spectators would probably get too rowdy.
The referees warned the game would probably go into overtime.
read slowly
Clause-Bias Verbs:
The bus driver worried the passengers were starting to get annoyed.
The bus driver worried the tires
were starting to go flat.
EQ-Bias Verbs:
read slowly
The senior senator regretted the decision had ever been made public.
The senior senator regretted the reporter had ever seen the report.
Ambiguity Effects
First-Pass Times in Eyetracking
At Disambiguating Verb
At Ambiguous NP
Plausibility
*
Plausibility
NS
Plausibility
NS
Plausibility Plausibility Plausibility
*
NS
NS
Ambiguity Effects
Total Times in Eyetracking
At Disambiguating Verb
Plausibility
*
Plausibility
*
Plausibility
NS
At Ambiguous NP
Ambiguity Effects
Self-paced Moving Window Times
At Disambiguating Verb
At Ambiguous NP
Plausibility
*
Plausibility
*
Plausibility
NS
Plausibility
NS
Plausibility
NS
Plausibility
NS
Notice direction of Plaus Effects at NP !!!
Verbs Rule! But Why?
• In these sentences, verb comes before relevant noun,
so gets a head start?
– BUT, Trueswell (1995) found same for Reduced Relatives,
where noun comes first (e.g., “evidence examined” )
• Verb Bias may be retrieved as part of recognizing verb
– While plausibility must be computed on-line for particular verbnoun combinations, which probably takes longer?
• Verbs determine how everything else in sentence
combines, so weighted most heavily?
• In English, verbs appear early in sentences, so we learn
to rely heavily on the predictive info they provide?
– So what do speakers of verb-final languages rely on instead?
Plausibility of possible noun combinations???
Correlations between Ambiguity Effect Size at Disambiguation
& Verb Bias Strength Across All Verbs
First-Pass
Times
Moving
Window
Times
Correlational Evidence
Specifically for Parallel Parsing
When Verb Bias & Plausibility provided conflicting cues
Clause-Bias Verb + Plausible-as-DO NP
As DO-Bias INcreases, difficulty INcreases
r = +.56 (First Pass)
+.47 (Moving Window)
DO-Bias Verb + Implausible-as-DO NP
As Clause-Bias INcreases, difficulty DEcreases
r = -.58 (First Pass)
-.59 (Moving Window)
So, when Plausibility cue conflicts w/ Verb Bias
- Other structure is considered
- To the extent the V is used in that other structure
Could ALL Non-Syntactic Influence be on
Reanalysis, not Initial Parse???
• Frazier (1995)
– “It may be significant that garden paths have never been convincingly
demonstrated in the processing of analysis A (the structurally simplest
one)… In such cases, if analysis A ultimately proves to be correct,
perceivers should show evidence of having been garden-pathed by a
syntactically more complex analysis even though the syntactically
simpler analysis is correct.”
• So, need to demonstrate early influence of nonsyntactic factors even when GP Model predicts no
reanalysis
– GP Model says simplest alternative tried first
– If correct, no reason to try others, so no reanalysis
– And thus no reason for non-syntactic factors to influence
parse
- Interactive models do predict “garden-pathing” in
simpler structures, if lexical &/or contextual cues
push toward more complex possible structure
- But they predict it should be smaller than in more
complex structures because …
- Structural (& conceptual) simplicity not ignored in such models
- Just not the only important factor in initial interpretation
- Simpler sentences are generally more common
- So other cues have to fight structural frequency effects
Wilson & Garnsey (2006)
Put both kinds of verbs in both kinds of structures
• Clause-Bias Verb:
The ticket agent admitted the mistake might not have been caught.
The ticket agent admitted the mistake because she had been caught.
• DO-Bias Verb:
The CIA director confirmed the rumor could mean a security breach.
The CIA director confirmed the rumor when he testified before Congress.
ALL post-verb NPs plausible as DOs
• Since DO-sentence with implausible-as-DO NP is implausible!
Length-Corrected Residual RT at Disambiguation
*
*
*
*
*
NS
*
*
NS
Convinced???
• According to GP Model, in DO-Structure Sentence
– Should never be any reason to reanalyze
– So no need to go back & reread
• So why do people reread earlier words more in
DO-Structure Sentence w/ Clause-Bias Verb?
– Because Verb Bias does influence initial interpretation,
not just reanalysis?
• Why do people
– Slow down & stay put in Clause-Structure Sentences
w/ DO-Bias Verbs?
– But reread earlier words instead in DO-Structure Sentences
w/ Clause-Bias Verbs?