Lexical Ambiguity in Sentence
Comprehension
By R. A. Mason & M. A. Just
Brain Research 1146 (2007) 115-127
Presented by Tatiana Luchkina
Background
• Lexical ambiguity demands the reader to
select one Mg & retain the possibility of using
another Mg at the same time;
• This ability is correlated with the memory
capacity of the reader + the frequency of the
ambiguous W Mgs;
• Correct analysis requires inhibition of the
alternative Mg, erroneous analysis – creates a
garden path effect;
Lexical Ambiguity
• Biased: a word’s Mgs are asymmetric in
frequency, e.g.,
• This time the ball was moved ….
….because it was always so well attended
• Balanced: two equally likely Mgs, e.g.,
…the cell looked small….
because it was piled high with supplies
Ball
Garden path effect
Cell
Multiple Mgs maintained
Q: Which ambiguity type takes longer to process??
Previous studies of Lexical Ambiguity
• Majority were grounded in behaviorist
tradition, measuring:
• Reading times/response times
• Eye movements
• Cross-modal priming effects
How about brain activity during the processing
of lexical ambiguity?
Ambiguity & the brain
Left inferior
frontal
gyrus
Left
temporal lobe
de-contextualized categorization
semantic analysis; tasks
Left inferior
temporal cortex &
right inferior frontal
cortex
contextualized
semantic
processing;
• Left Hemisphere – rapid fine semantic coding
(only relevant Mgs);
• Right Hemisphere – slow coarse semantic
coding (activates a broad spectrum of
meanings);
Left inferior (red) & superior (green)
frontal regions become active while
processing ambiguity
Study
• Brain imaging (fMRI) used to measure brain activity
during the reading of Ss with lexically-ambiguous
words vs. matched control words;
• Rationale: to measure brain activity when ambiguity
occurs in early (biased) or late (unbiased) selection
of meaning;
• How: brain responses to the processes of ambiguity
resolution relate to individual differences in working
memory capacity
Experiment:
• 12 right-handed volunteer college students;
• Stimuli: 36 sentences (ambiguous vs. control)
with the target appearing before any
disambiguating context;
• Sentences presented on the screen 1 W at a time,
at a normal reading rate; A yes/no
comprehension Q followed;
• Cerebral activation measured using blood
oxygenation level contrast;
Dep & Indep Variables: degrees & areas of cortical
activation; individual’s working memory capacity;
ambiguity types;
Findings
• Lex. Ambiguity evokes extra processing due to
generation, retention, selection of multiple
meanings and coherence monitoring;
• Reading of ambiguous Ss activated left inferior
frontal gyros more than reading of control Ss;
• Biased condition only produced additional
activation clusters in inferior/superior frontal
regions of both hemispheres;
• Activation in the right hemisphere - spillover of
processing to help resolve secondary Mgs;
Biased Ambiguity activates Right
Inferior Frontal Region (circled in red)
Balanced Ambiguity – left inferior
frontal region active only
Findings, cont’d.
• Activation patterns were correlated with the
reading spans of the subjects, which reflected
their working memory capacity;
• Readers with lower reading spans use right
hemisphere, esp. right inferior frontal area, to
resolve ambiguities & maintain multiple
meanings during disambiguation;
• Bilateral extra activation occurs selectively & is
minimal in the high-span subjects;