Vanilla is sweet and soft; lemon is cool and refreshing. If we rate smells with words like these, how closely does the resulting classification match what happens when we actually smell them and group similar ones together? A 2004 study comparing 11 fragrance materials found that the broad groupings corresponded, but there were discrepancies in the distinctions between similar smells, such as rose and jasmine, and in the position of lemon.
Can a small vocabulary classify smells?
In an earlier study, the authors had selected nine adjectives for describing fragrance: soft, sweet, and mild; intense, strong, and sultry; and clear, cool, and sour. These words relate to softness, intensity, and clarity.
Being able to describe many smells with a small vocabulary could reduce the burden of assessment. However, these words carry emotional as well as sensory connotations. The present paper therefore compared classification using these nine adjectives with a method that did not ask participants to describe the smells in words.
Comparing a new group of 50 with ratings from an earlier group of 50
The new experiment involved 50 participants with a mean age of 20.7 years. They smelled 11 materials from bottles, vanilla, lemon, rose, jasmine, ylang-ylang, lavender, rosemary, chamomile, pepper, sandalwood, and hexenol, and freely grouped together those they found similar.
There was no limit on the number of groups or the time allowed, and participants could change their groupings along the way. They were instructed to classify by similarity in odor quality, rather than similarity in intensity or the alphabetical order of the bottle labels. A perfumer had diluted the materials to a moderate intensity.
The verbal ratings used for comparison came from a different group of 50 people in the earlier 2002 study. They had smelled the same 11 materials on perfumer’s blotters and rated how well each of the nine adjectives applied on a 7-point scale. The procedure was spread over 3 days to avoid fatigue. This was not a comparison in which the same 50 people tried both methods: it combined reused data from the earlier study with a new sorting experiment.
The two odor maps broadly corresponded
In the free-sorting task, smells placed in the same group more frequently were treated as more similar. In the verbal ratings, smells with similar scoring patterns across the nine adjectives were treated as closer. These relationships were converted into two-dimensional maps, with similar smells positioned near one another.
The correlation between the distances for the 55 pairs formed from the 11 materials on the two maps was 0.61. The maps were not identical, but they showed shared relationships. The authors concluded that even a small number of adjectives could be useful for distinguishing relatively different types of fragrance.
More specifically, vanilla lay toward the soft end and apart from the other materials in both methods. Chamomile, pepper, lavender, and rosemary formed a nearby grouping, while ylang-ylang and jasmine were also close together. The directions associated with softness, sultry intensity, and clarity broadly corresponded too.
Rose, jasmine, and lemon revealed differences between the methods
On the verbal map, however, rose and jasmine were quite close. A limited shared vocabulary may have made it difficult to capture subtle differences in quality, describing both as strong and sultry. The authors suggested that nine adjectives might be insufficient, and that the 7-point scale might be too coarse.
Lemon, by contrast, stood apart from the other materials in the verbal ratings, where its cool and sour characteristics became prominent. When participants sorted by overall impression, it did not occupy such a distinctly separate position from pepper, rosemary, and lavender. This illustrates how assessing features individually and comparing overall impressions can bring different distinctions to the fore.
This was not an experiment showing that using words reduced the same person’s ability to discriminate smells. The comparison concerned classifications produced by different participants and procedures. Presentation also differed between bottles and blotters, so the discrepancies between the maps cannot be attributed to language alone.
Choose words or direct comparison to suit the purpose
The authors proposed considering additional adjectives, concrete source-related terms, or finer rating methods when the aim is to capture subtle differences between similar smells. When organizing many smells, however, they also considered rating each one with a shared vocabulary easier to handle than repeatedly comparing every smell with every other one. The latter point was a discussion of methodological advantages, not a result established by testing more than 100 materials in this study.
Emotion was another issue. Because verbal ratings shared many features with free sorting despite the words’ emotional connotations, the authors discussed the possibility that emotion might also play a part in sorting without words. However, the present analysis did not use the earlier study’s ratings of emotion-related adjectives, nor did it directly measure stress or establish it as a cause of the classifications.
The study involved a limited set of 11 plant-related fragrance materials and young participants. Rather than extending its findings to claim that all malodors or complex perfumes can be classified along just three directions, it is more useful to read it as a study of choosing an expressive method to suit the differences being measured.
Summary
Classification using nine adjectives broadly corresponded to the structure obtained by grouping similar smells together. Vanilla’s separate position and the proximity of ylang-ylang and jasmine were shared features. Verbal ratings, however, represented the difference between rose and jasmine as smaller and made lemon’s cool and sour qualities more prominent.
Words help us organize smells, but a limited shared vocabulary can miss fine distinctions. Are we seeking broad categories, or trying to distinguish two similar smells? The study shows why the purpose matters when choosing between verbal description and direct comparison.
At the time of the study, one coauthor was affiliated with Shiseido’s Product Development Center and the other two with Tohoku University. No separate funding or conflict-of-interest declaration section was found in the original paper.
Reference
Higuchi T, Shoji K, Hatayama T. Multidimensional scaling of fragrances: A comparison between the verbal and non-verbal methods of classifying fragrances. Japanese Psychological Research. 2004;46(1):10–19. doi:10.1111/j.1468-5884.2004.00232.x
