1. Emotion Word Processing: Evidence From Eye Movements
crucify betray industry owl avenue festive astonished cheer blackmail hostage bus history stove hard intercourse brave trauma sinful bees drown stomach wine alert passion casino cockroach insult passage column aloof dollar Christmas sexy slaughter confused privacy yellow exercise couple admired shark surgery rejected skull errand umbrella happy life party enraged disgust cancer rape fork manner win promotion love pressure war anger elbow barrel progress desire car intimate accident fire killer salad slush curtains nude fireworks erotic evil pain controlling radiator kettle miracle treasure inspired weapon divorce panic bland violin joke valentine reunion assault surgery tumour alley sentiment aroused ecstasy glory mutilate hurricane thief elevator prairie rescue laughter flirt
Emotion Word Processing:
Evidence From Eye Movements
Glasgow
Language
Processing
Graham G. Scott, Patrick J. O’Donnell, & Sara C. Sereno
University of Glasgow
Results: Gaze Duration
Main effect of Frequency (p<0.001).
Main Effect of Emotion (p<0.01).
Significant Interaction (p<0.001).
For LF words:
Negative < Positive < Neutral
For HF words:
Positive < Negative = Neutral
Introduction
Only 2 studies have investigated the interaction of emotionality and word frequency.
Nakic et al. (2006) carried out a lexical decision task (LDT) using a 2x3 design manipulating word frequency (high and low, HF & LF) and word emotionality (‘highly’ negative, ‘mildly’ negative, and neutral). They found main effects of frequency (HF<LF) and emotion (highly negative < mildly negative < neutral), but no interaction.
Scott et al. (2006) carried out an LDT using a 2x3 design to investigate word frequency (HF, LF) and emotion (positive, negative, neutral) and found main effects as well as an interaction:
The present study
Replicate and extend the results of Scott et al. (2006) in the context of normal reading.
Specifically, record subjects eye movements as they read high and low frequency positive, negative and
neutral targets presented in neutral sentences.
Main effect of Frequency (p<0.001).
Main effect of Emotion (p<0.001).
Significant Interaction (p<0.05).
For LF words:
Positive = Negative < Neutral
For HF words:
Positive < Negative = Neutral
Materials & Design
2 x 3 within-subject design: Frequency (LF, HF) x Word Type (Pos, Neut, Neg).
Word type was defined by arousal and valence ratings:
Arousal Valence Frequency Length
(lo-hi, 1-9) (neg-pos, 1-9) (#per mill) (characters) Example
LF Pos valentine
Neut appliance
Neg suffocate
HF Pos exercise
Neut village
Neg accident
Note: Word frequency values: British National Corpus (BNC).
Arousal and valence ratings: Affective Norms of English Words (ANEW) (Bradley & Lang, 1999)
15 words of each type (LF-Pos, LF-Neut, LF-Neg, HF-Pos, HF-Neut, HF-Neg) → total 90 words;
matched for frequency, length, number of syllables, syntactic category.
30 sentence triples to accommodate target word triples. Each subject read each target in different sentence
frames. 3 subject groups, so that each target was viewed in all 3 sentence frames.
Sentence frame kiss news bomb
1 Phoebe discussed the _____ at great length with her friends.
2 Michelle dreamt about the _____ every night for weeks.
3 Tom delivered the _____ with great care and attention.
Discussion
Significant Frequency x Emotion interactions occurred in first fixation and gaze duration measures.
Results seem to favour a perceptual defence based theory, such as Taylor’s (1991) Mobilisation-
Minimisation hypothesis:
For LF words, high arousal words (positive & negative) are processed more easily than neutral words.
For HF positive words, high arousal levels facilitate processing without any cost incurred from their
highly activated (i.e., HF) positive valence.
For HF negative words, initial processing facilitation is offset by the disruptive effects of highly
activated (i.e., HF) negative valence.
These data suggest that an early identification of the emotional tone of words leads to differential
processing. Specifically, HF negative words seem to attract additional cognitive resources.
This is consistent with a time-line in which emotional quality either accompanies or precedes (but does
not follow) lexical access (Taylor & Fragopanogos, 2005).
Method
Participants
44 native English speakers with normal vision who were not diagnosed as dyslexic.
Apparatus
Fourward Technologies Generation 5.5 Dual-Purkinje Eyetracker.
Procedure
Participants read single-line sentences while their eye movements were monitored.
Y/N comprehension questions followed sentences on half the trials.
Results: First Fixation Duration
Main effect of Frequency (p<0.05).
Main effect of Emotion (p<0.05).
Significant Interaction (p<0.01).
For LF words:
Negative < Neutral
Positive < Neutral; Positive > Negative
For HF words:
Positive < Negative = Neutral
References & Acknowledgements
Bradley, M.M., & Lang, P.J. (1999). Affective Norms for English Words (ANEW). The NIMH Center for the Study of Emotion and Attention, University of Florida.
British National Corpus (1995).
Nakic, M., Smith, B.W., Busis, S., Vythilingam, M., & Blair, R.J.R. (2006). The impact of affect and frequency on lexical decision: The role of the amygdala and
inferior frontal cortex. NeuroImage, 31,
Scott, G.G., O’Donnell, P.J., Leuthold, H., & Sereno, S.C. (2006, August). Emotion word processing: Behavioural and electrophysiological evidence. Poster presented
at the Aruchitectures and Mechanisms for Language Processing (AMLaP) meeting, Nijmegen, The Netherlands.
Taylor, S.E. (1991). Asymmetrical effects of positive and negative events: The mobilization-minimization hypothesis. Psychological Bulletin, 110,
Taylor, J.G., & Fragopanagos, N.F. (2005). The interaction of attention and emotion. Neural Networks, 18,
This work was conducted as partial fulfilment of a PhD at the Univ of Glasgow for G.G. Scott, funded by an ESRC postgraduate fellowship. Corresponding author:
(marg)
(marg)