How does metaphorical language affect individuals’ aesthetic perception in modern poetry: In the life span view

Dublin Core


How does metaphorical language affect individuals’ aesthetic perception in modern poetry: In the life span view


Qishan Liao




This study examined the relationship between the degree of metaphoricity and beauty perception as well as between cognitive load and beauty perception, by controlling for other possibly confounding variables such as familiarity and imageability. While previous research has shown that the variables of metaphoricity, familiarity and imageability influence beauty perception, no study investigate how the degree of metaphoricity and cognitive load influence beauty perception in poetic sentences reading. Therefore, this study aimed to bridge this gap. Beauty rating scale and keypress experiment were conducted, involving 22 young adults and 18 elderly adults. Because of the collinearity among metaphoricity, familiarity and iamgeability, a new variable called interpretation of metaphors was used to explain the hypotheses in the present study. Rather than cognitive load, interpretability was the predictor of beauty perception in poetry sentences reading. Young adults’ beauty perception achieved to the highest point at novel metaphors, while elderly adults considered dead metaphors as the most beautiful stimuli. This study suggests that poetic sentences are generally perceived as more beautiful when its degree of interpretability is lower in young adults rather than elderly adults. These findings provide an initial implication for future longitudinal or neuroaesthetic studies to further the understanding between metaphorical language and beauty perception.


Beauty perception


This study has been approved by the Psychology ethics committee at Lancaster University on 24/04/2018. Besides, this study were preregistered in ‘AsPredicted’ website, and the number was 11034.
The participants were 22 young adults between the ages of 18 and 30, and 20 elderly adults between the ages of 55-75. They were recruited from SONA systems, social media (e.g., Facebook advert). All young participants have not suffered from any learning disability (i.e., dyslexia) and they were native English speakers. However, two elderly participants confirmed that they had a history of dyslexia, so they were excluded. Finally, there were 22 young adults with a mean age of 21.64 years (SD=3.05) and 18 elderly adults with a mean age of 63.22 years (SD=6.07) have participated. Participants were required to give informed consent via an online consent form before completing the online survey, and they would fill in a paper version consent form before the keypress experiment. All participant would receive four pounds after finishing all experiments.
Stimuli. A bank of 92 stimuli, was generated by a previous student who was previously supervised by Dr Francesca Citron. Partially sentences are excerpted from modern poetry. The remaining sentences were created by this student, inspired by other poetic works. Creating novel sentences was to decrease the deviation caused by participants being familiar with some stimuli. All stimuli were divided into five categories, and the degree of metaphoricity of these categories was increasing. The first one is the literal expression which has concrete and pragmatic meaning and it usually equal to its literal meaning. It is not part of the metaphorical language. The following category is dead metaphors – a kind of metaphor that lose its imaginative space because of frequent use (Punter, 2007). The third one is the conventional metaphor that is commonly used in everyday life, and it is highly related to the specific culture. The fourth one is novel metaphors which is usually unusual in everyday life and challenging for the layperson to understand. The last category is extremely novel metaphors which are the most abstract and challenging. The semantic category overlap of subject and predicate in these sentences is less obvious than other categories. Considering the potential fatigue of the elderly participants, the researcher randomly selected 50 stimuli from the original stimuli bank as experimental materials (See Appendix A). There were ten sentences for each category. All stimuli were given a specific code for identification in the analysis procedure. The creator of the stimuli bank has invited 85 participants to rate the degree of metaphoricity for each stimulus via a 7-point Likert scale (1 for the minimum and 7 for the maximum). The result has shown that the degree of metaphoricity was increasing as the original design (See Figure 1).

Figure 1. Scatterplot showing the trend of metaphoricity ratings of stimuli. The categories corresponding to the stimuli number as follows: literal sentences (1-14), dead metaphors (15-28), conventional metaphors (29-53), novel metaphors (54-75), and extremely novel metaphors (76-92).
Apart from metaphoricity, these stimuli have been tested on multiple sentence-level characteristics, including familiarity and imageability in the same group of participants. Briefly, all ratings were collected by asking participants to rate "how familiar is this sentence to you?" and "how easy is it to imagine this sentence?" on two separate 7-point Likert scales. These raw data would be used for analysis in this study.
Survey. Beauty rating scale was designed as a 7-point Likert scale via the online survey software ‘Qualtrics’. The scale included a digital version of the information sheet, consent form and debrief form, and it also investigated several basic information like age, biological sex and reading frequency (Appendix B). More importantly, the scale included the questions for checking whether the participants are British native speakers and whether they have had the history of learning disability (i.e., dyslexia) since these factors can influence the beauty ratings. In the formal test, 50 stimuli would be randomly presented to the participants through Qualtrics. Participants would see the poetic sentences, as well as the question ‘How beautiful is this sentence to you?" on the page. They need to give their responses by rating from 1 to 7(1 for not at all beautiful and 7 for extremely beautiful) for each sentence.
Experiment. The researcher created a keypress experiment on ‘Presentation neurobehavioral system’ software. The material were identical to the online survey and included extra four filler sentences, five odd sentences, four questions related to the poetic stimuli. All new stimuli were generated by the researcher, but they would not be analysed eventually because of their functions (Appendix A). Filler sentences were used to let participants practice how to give their responses by the keypress. Odd sentences were unreasonable, and they were used to avoid the mechanically repeated responses. Similarly, some poetic stimuli would be followed by a question for checking whether participants have answered the question seriously. To ensure the randomness of the experimental materials, six versions of the experiment were created. Participants would be asked to read each sentence once at a time and to evaluate whether it was sensible for them by pressing a button (“F” for indicating “Yes” and “J” for indicating “No” via keyboard). Because wanting to avoid the habitual reaction caused by the participants being familiar with the traditional key press experiments, we also created six corresponding flipped version of the experiment. Overall, this experiment has 12 version, and they would be randomly allocated to the participant. Participants would take part in the experiment on the researcher’s computer, whereby the answer and the reaction time of each sentence would be collected by Presentation automatically and anonymously.

Questionnaire. When the participants decided to participate in the project, the researcher would send an anonymous questionnaire link to the participants by e-mail. The questionnaire can be completed on any electronic device, and the participants could pause the questionnaire at any time when they need a break. After clicking the link, the participants would read the information sheet and the electronic consent form orderly to ensure that they understood the necessary information of the questionnaire and gave their consents. They then need to answer ‘check questions’ to check whether they were native speakers and whether they have a previous or current learning disability. Knowing the answers to these questions was to confirm that the participants were suitable for the questionnaire. Subsequently, the demographic information would be asked, and all answers would be kept confidential.
After, a brief instruction form would be presented to explain the basic operations of the questionnaire and some important terms (e.g., beauty) involved in the questionnaire. Then, 50 poetic stimuli which were varied in the degree of metaphoricity would be presented randomly, followed by a question after each stimulus: How beautiful is this sentence to you. Participants should give their responses by rating from the 7-point Likert scale (e.g., 1 for not at all beautiful and 7 for extremely beautiful). All answers would be automatically recorded by Qualtrics.
After completing all, the participants would read the debrief sheet to understand the purpose and the design of this questionnaire. Also, the references about this questionnaire and the contact information of the experimenter would be given. When the participants completed the questionnaire, they would receive an e-mail from the experimenter to make an appointment for the keypress experiment. Time for the experiment was usually one or two days after completing the questionnaire.
Keypress experiment. All participants were required to meet the experimenter personally to complete the keypress experiment. Before the experiment began, participants were required to sign on the paper version of the consent form. After that, the experimenter would verbally explain the operation of the experiment. Then, the experimenter would randomly select one of the twelve versions of the experiments and give the participant a unique code. Participants were asked to evaluate whether the sentences presented on the screen were reasonable at the time. When they think it was sensible, they need to press the button that represents ‘Yes’, and vice versa. When they need to answer the ‘Yes/No’ questions, the operation was the same. When the participants understood the operation, they would press F or J key to start the experiment. Before the poetic sentence was presented, there would be a white fixation cross in the center on the black screen, and the duration was 1000ms. Then, the stimulus would present on the screen and last for 8700ms. Usually, the participants need to give their responses during this period, and their reaction time was automatically recorded by the software. After the presentation of a stimulus, it would be followed by a blank screen that lasts for 300ms with a white jittered fixation cross before the next sentence/question was presented. If the subject answers the question at this time, their reaction time of this stimulus will be the reaction time during this period plus 8700ms. The font for all stimuli was 12, the font colour was white, and the background screen was black.


Lancaster University






Lauren McCann













Francesca Citron

Project Level



Beauty Perception

Sample Size

22 young adults and 20 elderly adults

Statistical Analysis Type

Independent T-test
Perason's correlation
Partial correlation
Hierarchical regression
Simple regression


consent form.pdf


Qishan Liao, “How does metaphorical language affect individuals’ aesthetic perception in modern poetry: In the life span view,” LUSTRE, accessed January 25, 2021,