Subtle changes in language can have an affect on how we respond to various questions and stimuli, having consequences on how we view ourselves. In the present study, researchers wanted to examine the effects of cognitive (think) and affective (feel) verbs when reporting on thoughts about one’s self. They organized four experiments to test their hypothesis that responding to the “think” stimulus would yield a more negative self-evaluation than the “feel” stimulus.
The first two experiments looked at the effects of responding to the different verbs in open-ended responses, and to study the possible lasting effects on self-esteem using the Rosenberg (1965) self-esteem scale. The results from the experiments showed that the responses to “feel” stimuli were significantly more negative than “think” stimuli, which related to more negative self-esteem results after responding to “feel” stimuli.
In subsequent experiments, the researchers tested if manipulating the verb within the Rosenberg scale would have a similar effect on self-esteem. To do this, they changed the Rosenberg scale to incorporate either “feel” in 6 out of the 10 items or “think” in the 6 items, expecting lower self-esteem scores in the “feel” version of the test. Analysis of the results revealed that the verb effect was not significant, but separating the results by gender showed that females scored significantly lower than males in the “feel” version and higher in the “think” version.
To further investigate the verb effects in relation to gender differences, the researchers manipulated the Rosenberg scale so that every item contained either “think” or “feel” and also included an emotionality measure. This would show whether the gender difference was driven by heightened emotionality of female participants responding to the “feel” stimulus. The results showed that females scored significantly lower in the “feel” version than in the “think” version, but there was not a significant difference between the two versions for males. Furthermore, female participants’ scores on emotional orientation were higher when responding to “feel” prompts than when responding to “think” prompts, and analysis of emotionality showed that higher levels of emotionality correlated with lower levels of self-esteem.
The results from these experiments show that self-reflection does in fact rely on lexical influences as shown by the significantly more negative self-descriptions when presented with the “feel” versus the “think” manipulations. These experiments also proved that the subtle difference between cognitive and affective verbs can have a lasting impact on self-esteem. Further experimentation led to the finding that self-esteem scores vary depending on the use of an affective or cognitive verb for females but not for males. This effect might be explained by the higher level of emotionality tied to “feel” verbs for female participants which leads to lower self-esteem.
The study suggested three possible mechanisms to explain the different responses to the cognitive verb and the affective verb. One possibility is that these verbs activate different frames which make the self-relevant information retrieval processes distinct from one another. Another possibility is that the “feel” prompt is more likely to engage deeper self-processing by making people more aware of the reality as opposed to the ideal self status, which elicits a more negative response. The discrepancy between responses might also be explained by the varying confidence levels, where “think” has a higher level of confidence than “feel.” The journal article did not offer an explanation to the gender differences in self-esteem rating when prompted with “think” versus “feel” verbs. However, because women scored higher on emotional orientation when primed with “feel” prompts, this suggests that female affective processing is greater than in males.
In general, this study successfully tested how cognitive and affective verbs play a role in self-reflection and self-esteem. Because the experiment incorporated large number of participants from different backgrounds, the experiment has high external validity and would be able to be generalized to a larger population. However, it would be interesting to replicate this study within various cultures with other languages to see if the findings are still valid. In terms of internal validity, the Rosenberg scale only tests for explicit self-esteem; therefore, further research could include testing for implicit self-esteem. Furthermore, the study used multiple platforms to collect data, allowing for variance in their responses which could skew the data.
Holtgraves, T. (2015). I Think I Am Doing Great but I Feel Pretty Bad About It: Affective Versus Cognitive Verbs and Self-Reports. Personality and Social Psychology Bulletin, 41(5), 677-686.
Article Link: http://psp.sagepub.com/content/41/5/677
by Arianne Thomas and Yi-Pei Lo