When I first ran across Asifa Majid’s article with Ewelina Wnuk in Cognition, about how speakers of Maniq, a language indigenous to southern Thailand, have a vocabulary for talking about smell, I was taken aback. In anthropology, especially since the work of people like David Howes, Constance Classen, and Andrew Synott, we know very well that different cultures privilege olfaction and other senses more than Westerners do. The anthropology of the sense has made it clear that the ideological privileging of vision in the West, and relative underdevelopment of sense of smell or proprioception, is not matched elsewhere.
However, Wnuk and Majid were attacking, with empirical observations and psychometric testing, one of the pillars of Western philosophical accounts of how human senses evolved: the idea that human evolution had tipped the balance decisively away from olfaction. The alleged weakness and imprecision of olfaction was taken for granted in perceptual psychology.
Some of these theories of sensory evolution hold that our ancestors had, in a way, paid for our distinctive cognitive and perceptual development by sacrificing olfactory acuity. Vision increased precision at the expense of olfaction.
In fact, some theorists of brain evolution go so far as to suggest that there was a kind of neurological trade-off: language use could only grow as our ancestors lost a capacity for smelling. The restraint and remove from the immediate sense-world necessary for logic and abstract thought was opposed to the kind of complete immersion and sensory triggering of behaviour that other animals had because of the way aromas dominated their perception. Were the senses in a zero-sum exchange where visual acuity and a distinctly human way of life made acute olfaction impossible?
Research conducted by Asifa Majid, together with her collaborators, suggests that language and olfaction are not at odds; the right language can actually enhance the perception of aroma, as language has also enhanced, inflected and refined our other senses. Rather than a fact of human being, the neglect of olfaction in the West is a result of our own cultural presuppositions and sensory biases: smell suffers from neglect, not an inescapable evolutionary trade-off. (Majid’s research got a mention recently from Tanya Luhrmann in an op-ed in the New York Times: Can’t Place That Smell? You Must Be American: How Culture Shapes Our Senses.)
I ended up writing at length about their research in an earlier post, Giving names to aromas in Aslian languages, and Prof. Majid and I subsequently exchanged emails. She agreed to answer some questions about her research, which I’m really happy to present today as an interview (below the fold).
Asifa Masjid is Professor of Language, Communication, and Cultural Cognition at the Center for Language Studies, Radboud University in Nijmegen, but she’s also a Research Fellow, Donders Institute for Brain, Cognition and Behaviour, and affiliated with the Max Planck Institute for Psycholinguistics. You can follow her on Twitter at @asifa_majid.
Greg Downey: First off, I really want to thank you for agreeing to being interviewed, and for sharing your research with us. Although neither Daniel nor I are linguists, we’ve found repeatedly that research in linguistics and linguistic anthropology has a lot to offer in thinking about brain-culture relations.
To start us off, I wanted to ask you: How did you get interested in the language of smell and what alerted you to the fact that the Aslian languages have this remarkable vocabulary about aroma?
Asifa Majid: The interest in smell started from a cross-cultural survey on the language of perception. Across many different disciplines there is a long-standing idea that some sensory modalities are more important than others. It is said that humans are visual creatures first, followed by hearing, and then the other senses. I was interested to know whether this “hierarchy of dominance” was reflected in language(s). This led Stephen Levinson and I to ask whether some sensory modalities were more “ineffable” than others, i.e., difficult or impossible to put into language. (We lay out the theoretical framework in a paper soon to be published in Mind & Language.)
In order to test the hypothesis that some sensory modalities were easier to put into language than others, we created a standardised stimulus set of colours, sounds, tastes, smells, etc. Our colleagues used these stimuli to collect data from many different languages. Some preliminary results were published in the journal The Senses & Society, including Niclas Burenhult’s and my first paper on Jahai smell. The data Niclas first collected suggested something very different from the other languages in the cross-cultural sample. This immediately grabbed my attention, and led to the subsequent work.
GD: To get some kind of background, I wanted to ask you about the broader area of studying indigenous languages. People who study indigenous languages often point out how quickly we are losing them. Some estimate that potentially thousands of languages could be lost in the twenty-first century.
What is the current state of the Aslian languages you studied? Are there a lot of speakers? Are children learning the languages?
AM: The Aslian languages, like most languages of the world, are spoken by small communities. Speaker numbers are in the 100s for most Aslian languages; there are around 300 speakers of Maniq, and around 2,000 for Jahai. The good news is that they are still being learned by children and have wide applicability, that is, they are not restricted in their use to specific register or genre.
Fieldwork: Methods for investigating sensory experience
GD: How hard is it to get to these groups of people and to get them to tell you about their perceptions of aromas?
AM: Fieldwork is hard going but intensely rewarding. It is mentally demanding. There is a lot of information to assimilate at once: the individual personalities of the people, the different ways of sitting (yes, sitting) and behaving, the complexities of the language, etc. We work with speakers in their native language. This means that a good understanding of the linguistic structures is a prerequisite to any systematic data collection using experimental tasks. I rely heavily on my linguist colleagues for their expert knowledge here. Niclas Burenhult has been working with the Jahai for over a decade and has written the only existing grammar of the language. Ewelina Wnuk has been working on Maniq for 5 years, and has spent cumulatively over 8 months with the Maniq. Long-term fieldwork is crucial to the kind of work we do.
Aside from the intellectual demands, fieldwork can be physically grueling. Both Jahai and Maniq speakers live in tropical rainforest. This means we have to travel through jungle, wait out thunderstorms, try to avoid dysentery, malaria, dengue fever…, and be prepared for mosquitos and leeches. I think both Niclas and Ewelina are far braver than me when it comes to these challenges!
GD: Although you touch on the research method in a couple of your articles, I wanted to ask you more about your actual methods, the procedures you use for testing people on the sense of smell. How did you get the systematic data to talk about perception of smell?
A starting point was asking people to describe perceptual stimuli, such as Munsell colour chips and, for smell, the Brief Smell Identification Kit (the B-SIT), a sort of “scrach-and-sniff” way of presenting the same odors to different people. I was astounded when I first realised that Jahai speakers were making systematic distinctions between smell qualities using a smell set of verbs. This led us to probe these distinctions in additional ways.
We supplemented our knowledge of the smell terms in Jahai and Maniq by using “free-listing”: we simply asked people to list all the smell words they knew. In addition, we tried to find out more about their application by asking people to provide exemplars (What smells like X?) for each of the smell words. With the Maniq, we also conducted a “similarity judgment task”. We wanted to understand better the how the smell words were related to one another. So, we provided Maniq speakers with triads of smell words, and asked them to tell us which word was the most different to the others. It would be like asking English speakers to say which is the most different out of the following three: GREEN, BLUE, ORANGE. These judgments can be converted into similarity matrices, and then analysed using multivariate statistics to reveal the underlying structure of the smell lexicon.
These empirical methods are one prong of our smell studies. We also rely on linguistic elicitation to figure out more about the linguistic properties of the words: what word class are they, do they have special grammatical properties, etc. And, we rely on ethnographic observation to understand better what significance smell has in everyday life for the Jahai and Maniq. It is by pulling together the three stands: experimental study, linguistic documentation, and ethnographic observation that we are able to better understand smell in the Aslian langages and cultures.
GD: Sorry, but I want to go back to this method question, partially because I’m personally interested in using systematic data gathering techniques in field settings where cross-cultural issues are a big deal. How do you use the Brief Smell Identification Kit (the B-SIT)? Do you think that this way of testing your subjects’ sense of smell affects the data that you get?
AM: We use the B-SIT in an unusual way. The B-SIT is a standardised test. The way it is normally administered, is as a forced-choice test. People smell one of the scratch-and-sniff chips, and then say what smell it is from a choice of four pre-determined categories; for example, is this smell cinnamon, banana, petrol or strawberry. The four choices are given in English, so we can’t run it with the Jahai using this method.
We want to get equivalent data for the English and Jahai speakers, so instead of using a forced-choice task we use a free-naming task. Speakers simply have to say what smell is this. They can use whatever kind of description they like. We can then quantify the types of descriptors given by the two groups of speakers.
The advantage of using something like the B-SIT is that we can administer the same odorants to everyone. The odorants are standardised. However, this can also be a disadvantage. The smells are standardised for an American audience. So we cannot be sure that they are representative of the Jahai odour environment. In fact, we believe we have underestimated the Jahai abilities in our paper. We think we would learn far more if we were to adequately sample the Jahai odour space. But still, it is testimony to the potential of Jahai terminology that Jahai speakers still out-perform the English, even though the smells are biased to English smellscapes.
The philosophical implications of olfactory research
GD: You write in your article with Niclas Burenhult about Jahai language speakers that Western philosophers have long argued that ‘olfactory abstraction’ is impossible. That is, general traits or qualities of smells cannot be derived from the variety of smells we perceive, unlike abstract qualities such as color in sight or pitch in sound. Does your research with the Jahai and Maniq overturn this idea that abstraction is impossible with aromas?
AM: The Jahai and Maniq data is certainly incompatible with the view the olfactory abstraction is impossible. Have we overturned the idea? Only time can tell.
GD: Fair enough. I suppose that is the only way to really know! I hope time does tell…
When you analyze the clustering of smells and get your subjects to compare them, do they form any organized system? Do they have something that is equivalent to ‘primary colors’ or is the ‘smell space’ organized in some sort of systematic way?
AM: Yes, smell terms have a beautifully organised structure, akin to what we find for colour terms in other languages. You can find out about this internal structure by asking speakers to make judgments about the similarities between words using a similarity judgment task, as I explained before.
We gave Maniq speakers smell words, and asked them to say which were the most different. Based on these judgments, we were able to run statistical analyses that showed the Maniq smell lexicon is organised along two dimensions. The first is pleasantness: some smell terms refer to pleasant smells, others to more unpleasant smells. The second parameter seems to tap into dangerousness. For example, the terms miʔ huhuɸ, hamis and kameh sit on one end of this dimension. They refer to the smells of snakes, sun, millipedes and poison, smells that are believed by the Maniq to cause illness and pain. On the other side of the dimension are the terms miʔ bayɔɸ, lspəs and haʔıt which are not considered dangerous.
GD: Where do you think that the organization of the system of smells comes from? Is it influenced by the sense organs themselves? By the language? By the practicalities of day-to-day life, like what is good-to-eat and what is dangerous? Or is organized or structured by some other factors?
AM: Great question. In the paper we postulated that this might tell us something about the sense organs themselves, but the sense organs themselves are shaped by the practicalities of everyday life, and so forth. We’re currently expanding our research to look at other communities and environments in the hopes of teasing apart these factors.
Smell in psychological perspective
GD: You reference one of my favorite recent articles, the review by Joseph Henrich, Stephen Heine and Ara Norenzayan that explores how psychology is biased by an over-reliance on WEIRD research subjects, that is, subjects who are Western, Educated, Industrialized, Rich and Democratic. Henrich and colleagues detail a whole range of areas of cognition and psychology that are affected by this bias. Your research seems to reinforce this point so strongly for the sense of smell.
Where do you think that this bias comes from? Why do Westerners, not just the general public but also scientists who should know better, assume that our human sense of smell is so weak?
AM: Within psychology there is a lot of evidence to support the idea that the sense of smell is weak. It is a fact that English speakers struggle to name odours. There also isn’t a lexical field for smell as there is for the other sensory modalities. There have, in fact, been cross-cultural studies that confirm that this poor naming and lack of olfactory vocabulary generalises to other populations. Given that evidence it is not unreasonable to conclude that this is a pan-human trait. But, as we said in our papers, there is other evidence from within anthropology that casts doubt on the generalisation. I hope our studies will be able to make a bridge between the two fields, so we can revisit of some of these issues with a more diverse sample of people in mind.
GD: On a more personal note, do you think your research is affecting the way that you perceive aroma? Do you notice your own sense of smell changing, or do any of the people on the project report that talking about aroma is affecting the way that they perceive?
AM: I’m reflecting on smell every day in a way I wasn’t before I started this line of work, so – yes – my research is definitely affecting the way I perceive scents. It has certainly brought me a richer appreciation of everyday smells and flavours!
GD: Do you think that someone could be trained, outside of these groups, to perceive smells as accurately as they do, maybe by using their system of aroma terms as a form of training? One reason I ask is that some of the research on wine-tasters and perfume testers, people who do develop quite sophisticated olfactory senses, is inconclusive. Some research finds that these people don’t develop the kind of shared, mutually-intelligible systems for describing what they perceive that you describe in Aslian language. They develop their own idiosyncratic systems, but you don’t get classifications that generalize as well.
Do you think that this is trainable if one has the right training system, or is the context for Aslian language learners so complex that it’s hard to imagine taking the skill out of that context?
AM: I believe these categories can be learned. Your question points to the critical issue, however: what sort of learning experiences do we need exactly to learn smell categories? I think language plays a critical role. But this is something that needs to be fleshed out further with experimental studies.
As an aside, let me say that we are currently re-examining the issue of expert vocabulary for smells too. My PhD Ilja Croijmans has just finished a study investigating wine- and coffee-experts in order to establish whether they show more consensus in describing smells and flavours than novices. We are particularly interested in whether their knowledge is domain-general or domain-specific. We’ll keep you posted about our results!
GD: Thanks so much for doing this, Asifa. It’s not easy to do an interview-by-correspondence, but the research was so fascinating that I wanted to talk to you more. If readers want to learn more, check out the earlier post here, which also has links to the articles we’ve discussed.
Portrait of Asifa Majid from her Twitter account.
Photo of Majid administering the aroma identification test from the story: MPI “Language of perception” project featured on Dutch television documentary, at the website of the Max Planck Institute.
If you understand Dutch, there’s a video on YouTube of a panel interview with Prof. Asifa Masjid, as well as Profs. Mark Dingemanse and Carel ten Cate. Majid answers her questions in English, so I was able to follow her parts of it!
Final photo is public domain, generously shared by ‘Unsplash’ at Pixabay.