Rebecca Ha and Martha Gavidia

library(tidyverse)
library(dplyr)
library(readxl)

SensoryWords <- read_excel("SensoryWords.xls")
Goo2 <- read_excel("goo2.xlsx")

Introduction

Recognizing words goes beyond simply knowing the definition, research has shown that certain words produce a mental image much quicker than others depending on what senses are evoked when the word is read silently. We looked at a dataset containing over 5,000 mono and disyllabic words to determine whether there was a pattern in the words that evoked a stronger sensory experience. The words were reviewed by 67 native English speakers who were asked to rate each individual word on a scale of 1 to 7, the ratings are referred to as Sensory Experience Ratings or SERs. We hypothesized that words with a higher SER received such high ratings because they evoked more than one sense.

We were most interested in the words with high SERs so below are the top rated words with SERs ranging from 5.5 to 7 and a scatterplot ordering the words from highest to lowest SER.

filteredSER1 <- SensoryWords %>% 
  select(Word, AvgSER) %>% 
  filter(between(AvgSER, 5.5, 7))

filteredSER1
ggplot(filteredSER1, aes(x= reorder(Word, -AvgSER), y= AvgSER)) +
  geom_point()+
  xlab("Words")+
  ylab("Average SER")+
  theme(axis.text.x = element_text(angle = 90, hjust = 1), plot.title =
          element_text(color = "red", face = "bold", hjust = .5))

Testing Our Hypothesis

In order to test our hypothesis, Rebecca assigned the number of senses each word evoked for her. For example, the highest rated word “garlic” has an SER of 6.55 and it evoked her sense of smell, taste, sight and touch thus giving it a Sense Count of 4. If our hypothesis is correct, the graph detailing the number of senses evoked for each word should produce a similar graph as the one ordering the words from highest to lowest SER because a high SER should be equal to high Sense Count. The following is what our analysis showed:

ggplot(Goo2, aes(x= reorder(Word, -AvgSER), y= Sensecount)) +
  geom_point()+
  xlab("Words")+
  ylab("Sense Count")+
  theme(axis.text.x = element_text(angle = 90, hjust = 1), plot.title =
          element_text(color = "red", face = "bold", hjust = .5))

Results

Our plotted graph shows a lot of variation which we believe is due to the fact that, in testing our hypothesis, the data we recorded as Sense Count was that of just one person and not an overall average of numerous responses. There may have been different results if we collected data from several people but for the sake of time, data was collected from only one person.

However, the Sense Count data gathered from Rebecca did provide some interesting information. We noticed there was an outstanding number of times the sense of sight was evoked. Sight was included based on the concreteness of her own visualizations. For example, the visualization of a puppy was more readily available in her mind than an olfactory trace.

The following are charts detailing words that received SERs of 5.6, 5.7 and 5.9 and the senses Rebecca assigned to each word:

SER5pt6 <- Goo2 %>% 
  select(AvgSER,Word,Sense) %>% 
  filter(AvgSER == 5.6)
SER5pt6
SER5pt7 <- Goo2 %>% 
  select(AvgSER,Word,Sense) %>% 
  filter(AvgSER == 5.7)
SER5pt7
SER5pt9 <- Goo2 %>% 
  select(AvgSER,Word,Sense) %>% 
  filter(AvgSER == 5.9)
SER5pt9

From this data, we gathered that overall 80% of the words evoked the sense of sight, compared to 64% touch, 35% sound, 33% taste, 28% smell. More specifically, the sense of sight was evoked in 91.8% of the words with an AvgSER of 5.6, 77.8% of the words with an AvgSER of 5.7 and 80% of the words with an AvgSER of 5.9.

Interestingly, out of the words with an AvgSER of 6.0 only 33% of the words evoked the sense of sight. When we looked at the list of words with an AvgSER of 6.0, the words ocean and autumn created a more concrete visualization than the words humid and music. Humid evokes the sense of touch more than the sense of sight, and music evokes the sense of sound more than sight, which is why sight wasn’t included for either of those words. Something that also caught our attention was the word vampire. Looking at the word after analyzing the data, we realized that a mental image could easily be made from the word so it could also have the sense of sight associated with it in conjunction with Rebecca’s original assignment of touch. So, this leaves us to question under what guidelines the original participants rated these words by. If Rebecca assigned different senses to a word the second time she looked at it, it’s possible that the participants could have prematurely rated words not fully taking into account all of the possible senses that were evoked.

SER6 <- Goo2 %>% 
  select(AvgSER,Word,Sense) %>% 
  filter(AvgSER == 6.0)
SER6

Conclusion

The participants were asked to rate these words by circling a number from 1 to 7 yet, there was not a single word that had an AvgSER of 7.0. We took this to mean that there was not one word that every participant thought was worth the highest rating. This leads us to believe that even with our limited data in testing our hypothesis, it is clear that a high Sense Count is not indicative of a high SER because hypothetically a word that evokes all five senses could have been rated a 7.0.