Sometimes we name an odor’s source, saying ‘rose-like’; at other times, we sum up its qualities with a word such as ‘musty’. How closely do groupings of these words overlap with groupings of smells actually perceived through the nose?
A study published by Qi and colleagues in 2026 compared odor words used in Mandarin Chinese with the categorization of actual smells. It involved 76 participants: 44 Han and 32 Uyghur people. Beyond asking how many words they could produce, the researchers examined how they perceived similarities between words and between smells, exploring how sensory experience and language shape odor concepts.
Participants sharing a language but differing in everyday backgrounds
Both groups averaged about 21 years of age, and all participants had university-level education. Han participants were recruited in Guangdong Province. Uyghur participants came from Xinjiang and were studying and living in Guangdong when tested; they were also proficient in Mandarin.
First, participants listed as many odor-related words as possible in 60 seconds. An animal-naming task provided a comparison. Han participants responded in Mandarin, while Uyghur participants responded in both Mandarin and Uyghur.
In the task involving smelling and describing actual odors, however, everyone answered in Mandarin. Rather than analyzing the length of the entire response, the researchers extracted the first descriptor and classified it as an expression summarizing an odor quality, the name of an odor source, a reference to another sense such as taste or temperature, or an evaluation such as pleasantness.
Categorizing word cards and unnamed smells
Nine Mandarin odor terms were central to the categorization tasks. They included ‘香’, referring to pleasant fragrance; ‘臭’, referring to unpleasant odor; ‘腥’, associated with the fishy or raw smell of fish and meat; and ‘霉’, associated with mustiness. The study treated words describing these odor qualities as ‘abstract’ odor terms.
In the word task, participants looked at cards bearing the nine terms, imagined the corresponding smells, and grouped similar ones together. In the actual-odor task, they smelled nine samples selected to correspond to the words without being told their names, and grouped similar smells together. In both tasks, they could choose any number of groups.
Samples included rose-like, burnt, and fishy odors, and the contents of the containers were concealed. The comparison concerned these nine words and nine samples, not a classification encompassing every smell in the world.
Listing words and describing smells produced different results
In the Mandarin odor-word fluency task, Han participants produced an average of 13.84 words and Uyghur participants 11.59. This difference was statistically significant; differences were also found in the range of expressions for pleasant fragrances in particular.
When describing actual smells, however, Uyghur participants used more abstract expressions for some odors. For a fishy smell, for example, their first descriptor was sometimes a broad term for unpleasant odor rather than the name of a specific object.
Overall, source-based expressions were the most common in both groups. Recalling and listing words is not the same task as putting an immediately encountered smell into words, and word count alone did not determine how expressions were used.
The preliminary identification test involving 16 odors and the animal-naming task did not reveal clear statistical differences between groups. This does not, however, establish that every aspect of their olfactory or language abilities was identical.
The two groups’ ‘maps’ were very similar
The researchers tallied which words or smells tended to be grouped together and arranged them on two-dimensional maps, placing similar items closer together. This analysis is called multidimensional scaling. The maps represent similarities in categorization, not locations in the brain.
The Han and Uyghur groups’ categorizations corresponded closely: the correlation was 0.91 for word categorization and 0.93 for actual-odor categorization. Despite differences in individual placements and groupings, the groups shared substantial common ground.
The authors interpret the word maps in terms of dimensions such as pleasant versus unpleasant and plant-related versus animal-related. These dimensions interpret the categorization results; participants were not instructed from the outset to sort items according to those two criteria.
Importantly, there was also some correspondence between categorization based on words and categorization based on smelling the samples. In the Uyghur group, for example, one dimension of the word and odor maps was significantly related. This does not mean the linguistic and sensory maps matched completely. Nor do the values 0.91 and 0.93 measure agreement between these two tasks: they compare the two participant groups within each task.
Odor concepts involve both sensation and language
The authors suggest that the accessibility of words depends on expressions commonly used in a language and smells encountered in everyday life. They interpret these influences as potentially extending beyond the use of individual words to how people understand relationships among smells.
The study suggests that language does not simply copy how odors feel; rather, experience and language create groupings that partly overlap with sensory experience. However, this was not an experiment that manipulated participants’ cultural or linguistic experiences. It cannot distinguish whether culture, language, or individual life experiences produced the differences, or whether language changed perception.
The sample was limited to young, predominantly female participants with university-level education who could use Mandarin. Switching between two languages during word fluency may also have affected Uyghur participants. Moreover, a single sample representing each term cannot capture the diversity of everyday smells.
The authors call for studies involving people who are not proficient in Mandarin, a wider range of ages and backgrounds, and stimuli closer to everyday experience. These results cannot establish innate olfactory differences between ethnic groups or rank languages as superior or inferior.
Summary
There was some overlap between odor groupings made by looking at words and those made by actually smelling samples. At the same time, the participant groups differed in the words they readily used and in specific ways of categorizing smells.
The language of scent is more than a list of ingredient names. It can also reveal what we regard as similar and which qualities we group together. This study suggests that examining sensory and linguistic experience together offers a way to understand how we make sense of smells.
References
Qi, Y., Chen, L., Zhu, Z., Wu, H., Lin, K., Zhao, J., & Zou, L. (2026). From perception to concept: a cross-cultural study of olfactory abstract concepts in Mandarin Chinese. Humanities and Social Sciences Communications, 13, 1122. DOI: 10.1057/s41599-026-07501-8 / Full original article. Original article: © The Author(s) 2026, CC BY-NC-ND 4.0.
