Chapter 4: A Cross-cultural Web-based Experiment on Understanding Teacher
4.2 Method
4.2.2 Materials
4.2.2.4 Conceptual Representation: Feature Ratings
The comprehension of metaphor has long been studied through analysis of its features. According to Becker (1997), four types of features were involved in metaphor comprehension: common features (included in the representation of both tenor and vehicle), topic-shared features (only included in the representation of the tenor, but not the vehicle), vehicle-shared features (only included in the representation of the vehicle, but not the tenor) and emergent features (not part of the representation of either tenor or vehicle).41
Through analyzing the features associated with a metaphor, predictions from various metaphor theories can be tested. The traditional comparative metaphor theory argues that a common ground consisting of shared features dominates the metaphor comprehension process. The SMT proposes that interpreting metaphors
41 Nueckles and Janetzko (1997) categorized the features of metaphors into similar four types as
Becker (1997) but in different terms: redundant features (only belong to the tenor), transferred features (only belong to the vehicle), the convergent features (shared by both) and the emergent features (belong to neither).
requires shared systematic relations rather than shared features. In Gentner and Clament’s (1988) experiments, they found that more relational information rather than attributional information was generated in understanding metaphors.
A number of cognitive psychological experiments have provided evidence for the interactive theory which asserts that new features that are not associated with the topic or vehicle of the metaphor emerge in the metaphor comprehension process. For instance, Gineste, Indurkhya and Scart (2000) found that over 60% of the features produced during metaphor processing are emergent features.42 Becker (1997) suggested that emergent features are influenced mostly by one’s representation of the vehicle, as altering a metaphor’s vehicle produced greater changes in emergent content than altering the topic did. According to Coulson and Matlock (2001), emergent features should arise in the course of conceptual integration in metaphor processing and are more crucial to the meaning of the metaphor than common features. Nückles and Janestzko’s (1997) experiments showed that emergent features were encouraged when the similarity between the tenor and the vehicle of a metaphor was low or when the metaphor was more difficult to understand.
In experimental research on metaphor comprehension, features can be generated directly. For instance, Becker (1997) requested the subjects to list features not only of individually presented topic and vehicle words but also of the metaphors. Gineste, Indurkhya and Scart (2000) obtained the features through similar production task.
In the feature-generating study carried out by Coulson and Matlock (2001), subjects were asked to list the features for a series of target words that appeared either in a null context or in three sentential contexts, namely the so-called literal, metaphoric or literal mapping conditions. For example, subjects read the target word anchor in its literal sense in the literal condition, last time he went sailing, he almost forgot about the anchor. In the metaphoric condition, the anchor appeared in its metaphoric sense, i.e., Amidst all the trappings of success, his wife was his anchor. In the literal mapping condition, i.e., We were able to use a barbell for an anchor, the anchor was used for its literal sense but appeared in a way that required the subjects
42 Another interesting finding in Gineste, Indurkhya and Scart’s (2000) study is that those emergent
features took a longer time to response than topic-only features and vehicle-only features under the priming condition topic or vehicle. In contrast, under the prime condition of complete metaphors, the topic-only and vehicle-only features showed long response times, whereas the emergent features did not show any significant change in response time from one prime condition to another.
to perform integration operations similar to metaphor comprehension. The results showed that the topics in each of the three sentential conditions evoked substantially unique features. Most importantly, metaphors generated significantly more unique features than they did in the literal or literal mapping contexts.
In other cases, features of metaphors are not generated directly but rather collected in an indirect way. For instance, Tourangeau and Sternberg (1982) attained features by asking questions concerning a metaphor. In Gentner and Clament’s (1988) experiments, subjects’ objective description of individual terms of eight metaphors were further subjected to trained judges to assess the proportion of the attributional or relational information implied in subjects’ interpretation. Nückles and Janetzko (1997) requested their subjects to produce descriptions of both the tenor and the vehicle concept of metaphors in simple sentences or phrases. Their descriptions were subjected to a content analysis so as to have them well categorized into redundant features, transferred features, convergent features and emergent features.
The features used from this experiment were directly generated by the participants in pilot study II (see Appendix B). In February 2004, a total of 60 Chinese students and 60 German students at the University of Duisburg-Essen were asked to list out appropriate features to describe the concepts teacher, candle, captain and shepherd either with or without the association of a metaphor. Altogether, 187 different features were generated by all the Chinese and the German subjects. Among them, only those features that were generated by more than 75% of the participants in the same condition group were selected for the present studies. In this way, the 33 features most frequently listed were then selected and normalized to be used for this online research. The 33 features were originally presented in either German or Chinese in the online study. The English translation of those 33 features is provided here: thoughtfulness, responsibility, intelligence, leadership, watchfulness, lightheartedness, delight, patience, plainness, enthusiasm, model, diligence, love, orientation, authority, influence, romance, helpfulness, unselfishness, experience, calmness, courage, quietness, justice, strictness, self-sacrifice, optimism, friendliness, tolerance, trust, warmth, brightness and care.
In this online experiment, the participants were asked to rate the 33 selected features according to how suitably they described the concepts involved in the
metaphors. Later, the feature ratings under various conditions underwent cluster analysis and network analysis.