• Keine Ergebnisse gefunden

1. Introduction

Adjectives are words that modify nouns. Many adjectives describe attributes of objects belonging to living and non-living semantic categories such as people, places, or things. Usually, we use adjectives for the purpose of making the denotative and affective meaning of these objects more specific and more intense (e.g., tall boy, green grass, lovely women, brutal murder) (Dixon, 1999) as well as to express our own feelings. In German language, adjectives can be used in a predicative or attributive manner (Dixon, 1999).

Emotion studies using concrete pictorial material usually select pictures from standardized picture sets such as the IAPS picture set (see pp. 9 - 10 in the General Introduction section). The IAPS provides normative subjective ratings of emotional valence and emotional arousal for a wide range of emotionally evocative and neutral picture contents that can be used as a measurement standard across studies and laboratories. These normative IAPS ratings show a characteristic distribution in affective space thought to reflect the bi-motivational organization of emotion along the major dimensions of valence and arousal (e.g., Lang et al., 1997b).

Concerning language stimuli, a similar set of standardized emotional and neutral words has been collected for English words (ANEW) (Bradley et al., 1999b). For German words standardized word sets are still lacking constraining the comparability of findings across studies and laboratories.

The primary purpose of the following project was to provide a corpus of emotionally evocative and neutral German adjectives and respective normative student ratings of valence and arousal, that can be used as a measurement standard for the current studies of emotional word processing.

To this end about 1000 adjectives were collected. Adjectives were selected from different sources including literature books, magazines, newspapers and the handbook of German word norms (Hager & Hasselhorn, 1994). From these 1000 adjectives, 486 adjectives were selected.

These 486 adjectives were then randomly assigned to four different adjectives lists, each list containing 243 adjectives. Affective ratings of valence and arousal were acquired from 45 student subjects (26 females, 19 males, mean age: 25.7 years) of the University of Konstanz for each adjective, respectively, using the Self-Assessment Manikin (SAM) affective rating system.

Both paper-pencil versions and computerized versions of the SAM were used. In both versions

subjects were asked to rate each adjective separately for valence and arousal, using 9- point valence and arousal scales. The SAM has been used for the calibration of IAPS pictures, ANEW words and emotional sounds (IADS) and has been described in detail in the General Introduction section (pp. 7 - 9). For emotional adjectives, the detailed instruction of the SAM ratings as used for the adjective ratings in the sample of German students is presented in the appendix section together with the whole corpus of adjectives and respective mean valence and arousal ratings.

2. Distribution of emotional and neutral adjectives in Affective Space

When each adjective is plotted according to its normative mean valence and arousal ratings the adjectives show a typical u-shaped distribution in the two-dimensional affective space. As apparent in Figure 1, the adjectives elicited emotional reactions including the entire range of the valence (y-axis) and arousal (x-axis) dimensions – varying continuously from extremely unpleasant to extremely pleasant and from extremely calm to extremely arousing. Adjectives rated as neutral did not attain the high levels of emotional arousal associated with either pleasant or unpleasant adjectives. These neutral words are located between the upper, pleasant and lower, unpleasant arm of the valence dimension and cluster at the calm end of the arousal dimension.

Figure 1: Affective Space of emotional adjectives

Adjectives are organized in a two-dimensional affective space, defined by normative ratings of valence and arousal. Normative ratings were acquired from 45 healthy student subjects using the SAM (see x- and y-axis of the graph, respectively). The adjectives vary across a range of semantic categories, printed in red font, with adjectives belonging to categories of sex and crime having the highest valence and arousal ratings.

The resulting affective distribution of adjectives shares several important characteristic features in common with the distribution of IAPS pictures reported for evaluative SAM ratings of student subjects (see also in the General Introduction): Akin to IAPS pictures, adjectives addressing erotic and exciting as well as threatening contents obtained the highest valence and arousal ratings. These specific emotional contents have been hypothesized to reflect strong appetitive and defensive motives of high evolutionary significance (Bradley et al., 2000; Lang et al., 1997a). Second, the covariation between arousal and valence appeared to be stronger for unpleasant adjectives than for pleasant adjectives. Indeed, as reported for IAPS picture sets, regression lines relating arousal to unpleasant or pleasant valence differed in steepness and slope, respectively (e.g., Ito et al., 1998b). As shown in Figure 2, the regression line for unpleasant adjectives is steeper than that for pleasant adjectives. Akin to the IAPS system, unpleasant adjectives cluster at higher levels of arousal whereas pleasant adjectives are distributed fairly evenly across the valence and arousal dimension. In other words, whereas highly pleasant adjectives or pictures can be either highly arousing or low arousing, highly unpleasant adjectives or pictures are always rated as highly arousing. There exist no highly unpleasant and low arousing pictures or adjectives (Lang et al., 1997b).

Figure 2: The ‘positivity offset’ and the ‘negativity bias’ as predicted by regression lines relating arousal to pleasant adjectives (valence > 5.5) and unpleasant adjectives (valence < 4.5). Neutral adjectives with valence ratings of 4.5 < x < 5.5 were omitted from regression analyses.

(Regression line for unpleasant adjectives: y = 5.3-0.41; p <.001; Regression line for pleasant adjectives: y = 6.3+0.08; p =0.5).

The comparison between pictures and adjectives evidently shows that a ‘negativity bias’ as well as a ‘positivity offset’ is inherent in the evaluative ratings of emotional adjectives, as well (see also on pp. 10 - 11 in the General Introduction). Similar characteristics concerning a ‘positivity offset’ and ‘negativity bias’ in the distribution of affective words have also been reported for English words (Ito et al., 1998a). Together, the results suggest that emotional evaluation is biased either toward approach or defense. Whereas the appetitive system appears to be more engaged at low and moderate levels of arousal (‘positivity offset’), stronger responses of the defense system can be expected at very high levels of arousal (‘negativity bias’) (e.g., Cacioppo, 2004; Cacioppo et al., 1997; Ito et al., 1998b, 2005). Some authors have explained this bias in emotional evaluation from an evolutionary point of view according to which these differences in

responsitivity of the appetitive and defensive motivational system support an organism’s survival. At moderate levels of arousal, a ‘positivity offset’ motivates individuals to approach and explore the environment for obtaining knowledge about novel environmental situations and their potentially survival protecting and survival threatening consequences. By contrast, a

‘negativity bias’ may guarantee responding rapidly and adaptively with attack, defense or withdrawal when confronted with highly arousing stimuli that threaten survival (Cacioppo, 2004;

Ito et al., 2005).

The fact that emotional ratings of pictures and adjectives show such a highly similar and characteristic distribution in affective space emphasizes the view that emotional reactions elicited by pictures or by words share a common biological basis that has been supposed to have its origin in two motivational neural brain systems, appetitive and aversive, with emotional arousal predicting the activation of either the appetitive or aversive system (e.g., Lang, 1979;

Lang et al., 1997a, b).

A full list of the adjectives as well as mean valence and arousal ratings of the 45 student subjects can be found in the appendix section.

How, when, and where in the brain emotional adjectives are processed and evaluated according to their motivational significance is the focus of the experiments described in the following chapters. Using the adjectives described here, one has the opportunity of selecting subsets of words whose distribution of valence and arousal is well known from normative ratings, thus permitting objectivity and comparability of findings across experiments, methods and subjects.

III. Chapter 2

Study 2: Emotion and Motivated Attention: Time course of