Reading Ambiguous Words
description
Transcript of Reading Ambiguous Words
Reading Ambiguous Words
Sara Sereno
in collaboration withPaddy O’Donnell
Why ambiguous words?
• Ambiguous words have 1 form and 2 meanings:
• Is only the context-relevant meaning selectively accessed, or, are all meanings accessed (regardless of context) with selection occurring at a later, post-lexical integration stage?
• The timing of contextual constraint - early or late - has implications for the architecture of language processing...
• Understanding how ambiguous words are processed tells us about how words in general are processed.
CRICKET = or cf.
Distributed hierarchical visual processing in the primatelexical human
features
letters
word forms
meanings
higher-levelsemantics
syntax
Distributed hierarchical visual processing in the primatelexical human
Measurement
• In order to specify when higher-level processes affect lower-level processes, one needs to accurately measure the processes of interest.
• In word recognition, perceptual and cognitive events occur on the millisecond scale.
+ =?
(Sereno & Rayner, Trends in Cognitive Sciences, 2003)
But, when is access?
• A word frequency effect [ HF < LF ] is used as a marker or index of successful word recognition or lexical access.
The sore on Tam-Tam’s was swollen.(HF) back(LF) rump
• The word frequency effect represents the differential response to commonly used high-frequency (HF) words vs. low-frequency (LF) words that occur much less often:
• But, what does frequency have to do with ambiguity?
BANK
Dominant:“money”
Subordinate:“river”
Biased (polarised): Dom >> Sub
Balanced: Dom ≥ Sub
BANK
ambiguous unambiguous controls
BRIMEDGE
“river”
“money” “edge”
“brim”
MEANING
FORM
Dom
Sub
HF
LF
EM ambiguity studiesDuffy & Rayner (1986) x xDuffy, Morris, & Rayner (1988) x xRayner & Frazier (1989) x xSereno, Pacht, & Rayner (1992) x x xDopkins, Morris, & Rayner (1992) x xRayner, Pacht, & Duffy (1994) x - switchSereno (1995) x xBinder & Morris (1995) x - switchBinder & Rayner (1998) x xBinder & Rayner (1999) x xRayner, Binder, & Duffy (1999) x xWiley & Rayner (2000) x xKambe, Rayner, & Duffy (2001) x - switchBinder (2003) x - switch
Contextsentence paragraph
Control wordHF LF amb
ERP studySereno, Brewer, & O’Donnell (2003) x x x
EM ambiguity studiesDuffy & Rayner (1986) x xDuffy, Morris, & Rayner (1988) x xRayner & Frazier (1989) x xSereno, Pacht, & Rayner (1992) x x xDopkins, Morris, & Rayner (1992) x xRayner, Pacht, & Duffy (1994) x - switchSereno (1995) x xBinder & Morris (1995) x - switchBinder & Rayner (1998) x xBinder & Rayner (1999) x xRayner, Binder, & Duffy (1999) x xWiley & Rayner (2000) x xKambe, Rayner, & Duffy (2001) x - switchBinder (2003) x - switch
Contextsentence paragraph
Control wordHF LF amb
ERP studySereno, Brewer, & O’Donnell (2003) x x x
Sereno, O’Donnell, & Rayner
The moon cast an eerie light as Sister Margaret hurried up the unlit road. She had heard tales about the vampire. Although she did not believe them, Sister Margaret was still cautious. So when she was out alone at night, she wore her habit and carried a stake.
The moon cast an eerie light as Sister Margaret hurried up the unlit road. She had heard tales about the vampire. Although she did not believe them, Sister Margaret was still cautious. So when she was out alone at night, she wore her habit and carried a stake.
habitcrossshawl
AmbHF (form)LF (meaning)
Single Fixation Duration
250
260
270
280
290
1
Word Type
Fixation Time (ms)
AmbHFLF
Amb HF LF
Conclusions• Ambiguous words (with prior context supporting the
weak, subordinate sense) are simultaneously:HF forms fastLF meanings slow
• The present data support a top-down account, with early (lexical) selection of the contextually appropriate sense.
• A strict bottom-up account, with later (post-lexical selection), predicts increased difficulty (both meanings would need to be integrated at least half of the time).
• Future lexical ambiguity studies should use both the word-form (HF) and word-meaning (LF) controls.
Fixation Time
240
260
280
300
320
340
360
FFD SFD GD TT Nxt +2wd
AmbHFLF
(Sereno & Rayner, Trends in Cognitive Sciences, 2003)
Emotion words
Valence
Arousal
+ ve
– ve
Lo Hi
peace love
bored fire
Neutral controls: hotel, farm
MEASURE
Normal reading
TASK
fixation duration (as well aslocation and sequence of EMs)
TIME RES.
GOOD
POORhemodynamic imaging: fMRI, PET
electromagnetic imaging: EEG, MEG
various word tasks
ms-by-ms
seconds
various word tasks
Standard word recognition paradigms (± priming, ± masking):
naming
categorisationlexical decision RT
~500 ms~600 ms~800 ms
~250 ms
MEASURE
Normal reading
TASK
fixation duration (as well aslocation and sequence of EMs)
TIME RES.
GOOD
POORhemodynamic imaging: fMRI, PET
electromagnetic imaging: EEG, MEG
various word tasks
ms-by-ms
seconds
various word tasks
Standard word recognition paradigms (± priming, ± masking):
naming
categorisationlexical decision RT
~500 ms~600 ms~800 ms
~250 ms
HF vs. LF activation
0
0.5
1
1.5
2
2.5
3
3.5
4
1 2 3 4 5 6 7 8 9 10
Time
Activation
HF
LF
t