Dianne%20Bradley,%20Eva%20Fern

1
Prosodic weight versus information load in the
RC attachment ambiguity
CUNY GC QC

• Dianne Bradley, Eva Fernández Dianne Taylor
• Graduate Center Queens College CUNY
• DBradley_at_gc.cuny.edu
• 16th Annual CUNY Conference on Human Sentence
  Processing
• Cambridge MA ? 27-29 March 2003

2
RC Attachment Illustration
N1
N2
The plot concerns the guardian of the
prince who was exiled from the country for
decades
RC
La trama es sobre el guardián del
príncipe que fue exiliado del país por
décadas
3
RC-Length Attachment Accounts

• Informativeness Hypothesis
• e.g., Hemforth and colleagues
• Implicit Prosody Hypothesis
• Fodor and colleagues

4
RC-Length RC-Content

• who was exiled
• who was exiled from the country for decades
• Long RC has more lexical content, hence more
  informative than Short, necessarily?
• Informativeness per se may influence the
  preferred attachment

5
Informativeness RC-Length

• Informativeness Hypothesis
• Hemforth and colleagues
• cf. Relativized Relevance (Frazier 1990)
  Predicate Proximity (Gibson et al. 1996)
• The likelihood that modification applies toan
  element central to the assertion increases with
  the modifiers informativeness
• Length Effects
• N1 an argument of the main verb, cf. N2

6
Prosody RC-Attachment

• Implicit Prosody Hypothesis
• Fodor and colleagues
• Attachment preferences in silent reading
• In silent reading, a default prosodic contour
  is projected onto the stimulus, and it
  may influence syntactic ambiguity resolution
  (Fodor 1998, 2002)

7
Prosody RC-Length

• ? Alignment Optimally, prosodic phrase breaks
  and syntactic phrase breaks edge-align

? If RC is separately phrased, N1 attachment
(but not N2 attachment) satisfies
alignment ? Reconfigure syntax accordingly
8
Why Phrase RC Separately?

• Balance1
• N1-of-N2 RC
• Balance2
• (S) V N1-of-N2 RC
• RC-Privilege
• RC

9
Extra Weight in Remote Left Context?
e.g., The unusual plot concerns

• Informativeness
• Balance1
• N1-of-N2 RC
• Balance2
• (S) V N1-of-N2 RC
• RC-Privilege
• RC

• PREDICTION
• Irrelevant
• No change in N1 attachment
• Irrelevant
• No change in N1 attachment
• Balance point moves leftward, e.g., N1 of-N2
  RC
• Less N1 attachment
• More likely, N2 RC
• More N1 attachment

10
Roadmap

• Behavioral data More N1 attachment?
• RC Length
• Subject Weight
• Therefore
• Informativeness
• Balance1
• Balance2
• RC-Privilege
• Acoustic data Phrase break?
• N2 RC
• Elsewhere

?
?
?
?
?
? ?
?
?
?
11
Questionnaire Materials

• RC Length ? Matrix-Subject Weight
• RC has 1 versus 3 prosodic words
• MX has 1 versus 2 prosodic words
• The plot concerns the guardian of the prince who
  was exiled.
• The plot concerns the guardian of the prince who
  was exiled from the
  country for decades.
• The unusual plot concerns the guardian of the
  prince who was exiled.
• The unusual plot concerns the guardian of the
  prince who was exiled from the
  country for decades.
• N 36 sentence quadruples
• Assessed and adjusted to eliminate inherent
  plausibility bias favoring either N1 or N2

12
Contextualized Plausibility Test
The plot concerns the guardian of the
prince. A. The guardian was exiled. B. The
prince was exiled.

• Circle one number to record your judgment of the
  relative plausibility of Sentences A and B, in
  the context set up by the initial sentence.
• AgtgtB BgtgtA
• 1 2 3 4 5 6
  7

13
Questionnaire Procedure

• Reading comprehension test
• 36 targets, 108 fillers (13 ratio)
• Comprehension question after each sentence

Example of target The plot concerns the guardian
of the prince who was exiled from the country for
decades. Who was exiled? the guardian the
prince Example of filler The sneaky burglars
took all the stereo equipment but overlooked the
computer system. What was stolen? the stereo
the computer
14
Questionnaire Participants

• N 44, Queens College students
• US English speakers
• Language-history questionnaire,
  non-nativespeakers excluded/replaced
• Rejected/replaced for errors gt 15 in fillers

15
Results Preference Questionnaire
Relative Clause Length F1(1,40) 24.95,
plt.001 F2(1,32) 30.12, plt.001 Matrix
Subject Weight F1(1,40) 5.51, plt.05
F2(1,32) 9.43, plt.01 Interaction F1 lt 1
F2 lt 1
16
Roadmap so far

• Behavioral data More N1 attachment?
• RC Length
• Subject Weight
• Therefore
• Informativeness
• Balance1
• Balance2
• RC-Privilege
• Acoustic data Phrase break?
• N2 RC
• Elsewhere

?
?
?
?
?
? ?
17
Overt Prosody Study

• In overt prosody, are sentences with heavier
• RCs and/or heavier subjects, systematically,
• likely to be phrased as N2 RC ?
• Procedure Elicited production
• Measure Duration

18
Elicited Production

• N 8 native US English speakers 5F, 3M
• N 6 ? 4 sentences,RC Length ? Matrix-Subject
  Weight
• RC 1 versus 3 prosodic words who was exiled (
  from the country for decades )
• MX 1 versus 2 prosodic words The ( unusual )
  plot

19
?
The plot concerns the guardian of the prince. The
prince was exiled.
The plot concerns the guardian of the prince who
was exiled.

• Combine two simplex sentences into complex
  sentence with N1-of-N2-RC
• Simplex sentence pair disambiguates attachment

20
?
The plot concerns the guardian of the prince. The
prince was exiled.
The guardian
The prince
The plot concerns the guardian of the prince who
was exiled.
The plot concerns the guardian of the prince who
was exiled.
The plot concerns the guardian of the prince who
was exiled.
the guardian of the princes who was exiled.
the guardians of the prince who was exiled.
the guardian of the princess who killed
himself.
the guardian of the princess who killed
herself.
21
?
The plot concerns the guardian of the prince. The
prince was exiled.
The plot concerns the guardian of the prince who
was exiled.

• RC is consistently attached to N2
• RC attached to site dictated by Late Closure
• Cards stacked against the RC Privilege prediction

22
1
The unusual plot concerns the guardian of the
prince. The prince was exiled from the country
for decades.
23
Acoustic Analysis Regions
Duration Uniform acoustic signature of phrasal
break
Wt
S
V
N1
N2
RC1
RC3
The( unusual )
plot
concerns
theguardianof
theprince
who wasexiled
( from the countryfor decades )
24
Acoustic Analysis Regions
S V
V N1
N1 N2
N2 RC
25
N2 RC
26
Region N2
RC F1(1,7) 11.46, plt.02 MX F1(1,7) 2.80,
p.138 F2(1,5) 9.96, plt.05 F2(1,5) 2.07,
p.209 Interaction MX x RC F1 lt 1, F2 (1,5)
1.62, p gt .25
27
Roadmap Destination Report

• Behavioral data More N1 attachment
• RC Length
• Subject Weight
• Therefore
• Informativeness
• Balance1
• Balance2
• RC-Privilege

Acoustic data Phrase break N2
RC Elsewhere
?
?
28

• Our thanks to
• Janet Dean Fodor
• Shukhan Ng
• CUNY Linguistics graduate students
• Fernanda Ferreira
• Research Institute for the Study ofLanguage in
  an Urban Society