•Linguistic category model (Semin, Fiedler, 1988),
language abstractness (Maass et al., 1989)
•Levels of abstractness in language we use to describe people
•More abstract language – perceived as more stable behavior driven primarily by the actor
characteristics
•More concrete language – perceived as more situation specific behavior driven primarily by
situation characteristics
•Language intergroup bias (e.g. Hewstone, Rubin, Willis, 2002)
•Tendency to describe negative behavior of outgroup by more abstract language and positive behavior
of outgroup by more concrete language and inversely tendency to describe negative behavior of
ingroup by more concrete language and positive behavior of outgroup by more abstract language
•Nouns – the most abstract type of language, high level of essentialism, more abstract than
adjectives
•Tendency to describe outgroup members in media by nouns instead of adjectives

The experiment
•3x2 inter-subject experiment
•1. independent variable: news article valence
(3 levels: positive, neutral, negative)
•2. independent variable: language abstractness
(2 levels: nouns, adjectives)
•There were six different news articles (3x2) about a member of Vietnamese minority.
•Negative articles wrote about an attack of a Vietnamese man on a Czech man. Positive articles
wrote about an attacked Czech man, who was saved by a Vietnamese man. Neutral articles were about
Vietnamese culture.
•3 articles used nouns for description of the actor, like Vietnamese, attacker Vietnamese, savior
Vietnamese etc.
•3 articles used nouns for description of the actor, like Vietnamese man, Vietnamese attacker,
Vietnamese savior etc.

The experiment
•Dependent variables:
•Attitudes toward Vietnamese group (attitude thermometer)
•Intergroup bias (difference between attitude towards Vietnamese and towards Czech)
•Emotion scale towards Vietnamese
•Trust towards Vietnamese
•Intervening variables:
•Level of contact with Vietnamese people in real life and in media
•National identification
•Need for congitive closure
•