Introduction

In this project, I took data gathered by Lee M. Stadtlander and Latona D. Murdoch in order to compare how certain attributes ascribed to adjectives compare to each other.

Lee M. Stadtlander and Latona D. Murdoch originally gathered this data in order to test how well subjects were able to memorize adjectives based on their characteristics. In order to gather this data on each adjectives, they had participants score the words based on whether they felt it was hard or soft, hot or cold, big or small, etc. For example, participants would likely score the word ‘fluffy’ as being softer and the word ‘rocky’ as harder. They then used them, and found that ambiguous adjectives were harder to memorize. In other words, adjectives that participants weren’t able to precisely describe in terms of weight, shape, temperature, size, etc. were harder to memorize.

While the concept of using this data to test memorization is interesting and undoubtedly useful, I instead wanted to focus on whether adjective characteristics affected each other in order to get a better grasp on how we view the world in general. For example, words that were rated to be ‘heavier’ might also have a higher rating of ‘size’.

Table of Data for Each Adjective

Below is a table of raw data for each adjective. The column entitled ‘Sensory List Frequency’ shows the overall frequency for each adjective. These adjectives were gathered by having participants in the study write down as many adjectives that they could think of, and this column is the general calculation of frequency for these words.

The next column, entitled ‘Touch Focused Frequency’, shows the frequency of only those words that are related to the sense of touch in some way. I did not use this column, as it was more restricted and did not benefit my graphs in any significant way.

The third column, entitled ‘KnF Frequency’, is a calculation of how quickly participants were able to recall the words. I also did not use this column for similar reasons.

The following columns are self explanatory. The ‘M’ following some of the column titles indicate that this is the Median calculation of how these words scored for specific attributes, and the ‘SD’ stands for the Standard Deviation. For the graphs, I used only the Median as I felt that would give me a better understanding of the average score each word had for each attribute.

In understanding the scores associated with the different characteristics, it is imporant to mention that the scores are rated on scales from either 0 to 7, or -7 to 7, depending on the characteristic being described. The characteristics of hardness, roughness, size, temperature, and weight are all scored on the scale of -7 to 7. The characteristics of motion and shape are scored on the scale of 0 to 7. The higher the score, the more severe that attribute. For example, a higher score in motion means that the adjective has more movement, for shape it means the shape is more defined, and for temperature it means that the adjective has a hotter temperature. The lower the score the less severe that attribute, so a lower score in motion means that adjective is more stationary, in shape means that adjective has a less defined shape, and for temperature means that adjective has a colder temperature.

Wordcloud Depicting Frequency

Here, I created a wordcloud depicted the relative frequency for each adjective used in this dataset. The larger the word, the more frequently it appeared.

As can be seen in the wordcloud, the most common adjectives include the words ‘hard’, ‘cold’, ‘hot’, and ‘rough’. The least common adjectives include ‘cotton’, ‘glass’, ‘bendable’, and ‘nonbreakable’.

Graphs Comparing Adjectives’ and their Characteristics

The following graphs are not a complete collection of every adjective that shares some correlation between two attributes. I included only those graphs that had a significant amount of adjectives involved in order to narrow my focus. This means that those graphs that showed less than 4 adjectives were not included.

Temperature Compared to Roughness

Temperature and Roughness

In this graph, we can see that there is a coorelation between the apparent rougness and temperature of an adjective. In this case, the colder the perceived temperature of an adjective, the smoother it also seems to be. Three of the words in this graph, ‘cotton’, ‘icy’, and ‘slippery’ are shown to have a low temperature and low roughness rating. The clear between ‘icy’ and ‘slippery’ is clear, with both being connected to ideas such as ‘winter’ and ‘snow’. The word ‘cotton’ is slightly more suprising in its coorelation. The connection of ‘cotton’ being smooth makes sense as people often think of cloth when thinking of cotton, cold is more interesting because cotton is often used in clothes as a way to generate heat. It’s possible that the connection here is that people often wear cotton clothing in the Winter, thus connecting it to ‘icy’ and ‘slippery’ and therefore leading this word to have a lower perceived temperature.

The only outlier in this graph is the word ‘harsh’, which is rated as having a high temperature and high roughness rating. While it is an outlier in the sense that it has such high ratings, it still shows a strong correlation between roughness and temperature. Using this graph, we can conclude that if a word is perceived to be colder it will also be perceived to be smoother, and if it is perceived to be hotter then it will be perceived to be rougher as well.

Hardness Compared to Roughness, Shape, and Weight

In the following graphs, we can see that the most common overlaping attributes that occur within this data are the measures of hardness and roughness.

Hardness and Roughness

In this first graph showing this correlation, there is a fairly stong relationship between these two characteristics. Adjectives perceived to be smoother are also perceived to be softer, such as the words ‘velvety’, ‘suede’, and ‘satiny’. This correlation could be due that many of the smoother adjectives tend to be words related to cloth or fabric.

The correlation between hardness and roughness is present, but it is not as clear as the correlation between softness and smoothness. There are a few more outliers further up the graph, meaning that there isn’t necessarily a connection between roughness and hardness and vice-versa.

Hardness and Shape

The second graph, depicting the relationship between hardness and shape, does not have seem to have all that strong a relationship between the characteristics. However, it does seem that adjectives that contain the characteristics of both shape and hardness tend to harder in general. For this graph, there are only two adjectives, ‘crinkled’ and ‘crinkly’, that contain a negative score for the hardness rating.

Hardness and Weight

The last graph shows a mild correlation between hardness and weight. Adjectives that tend to be harder are seen to be heavier as well, and adjectives that are softer are seen as lighter. This can be seen in the adjectives ‘fluffy’ and ‘feathered’, which rate low on both weight and hardness, and ‘brick’ and ‘steel’, which rate high on both weight and hardness. If there were more data, I’m confident that there would be a clearer line of correlation in this graph.

Motion Compared to Roughness

Motion and Roughness

For this graph comparing motion and roughness, it is hard to discern a clear relationship between the two due to the lack of substantial data. However, we can see that words that contain the characteristic of motion in some way also tend to contain the characteristic of smoothness. Three out of the four words depicted in the graph are shown to have a negative roughness measurement, with those words being ‘fluid’, ‘gelatinous’, and ‘lubricated’. This shows that words connected to liquids, which are characteristically smooth, tend to be perceived as more prone to movement due to how loose and uncontained they are.

Roughness Compared to Shape

Roughness and Shape

For this graph, we can see that for adjectives that are characterized as having some sort of shape also tend to have a positive roughness rating. The only two exceptions to this are the words ‘frayed’ and ‘flat’, meaning that these words are seen as being smooth. It is interesting that there is there correspondence between roughness and shape, where there is some degree of roughness even if the adjective in question has an undefined or nebulous shape.

Conclusion and Further Studies

Overall, these graphs are interesting in that they give an insight into how people tend to connect characteristics of different adjectives. Some of these connections were predictable. For example, it makes sense that rougher adjectives will be seen as having less motion, as roughness indicates a sense of rigidity. It also makes sense that hardness and roughness would have a correlation, as harder objects (such as rocks) do tend to have a rougher texture as well. However other connections were surprising, such as how smoother adjectives having a lower temperature or objects perceived as having shape, nebulous or not, are also perceived as typically having some level of roughness.

I feel this would be an interesting topic to further explore, and it would be beneficial to collect more data on these adjectives and how they are characterized. This would help greatly in creating clearer graphs, and could even be used to study how people mentally classify words in general.

References

Stadtlander, L. M. & Murdoch, M. D., (2000). Frequency of occurence and rankings for touch-related adjectives. Behavior Research Methods, Instruments, & Computers, 32(4), 579-587.