Why Do Most Languages Have So Few Words for Smells? | Hacker News Reader

Why Do Most Languages Have So Few Words for Smells?

theatlantic.com

39 pointsbootload46 comments

I would hypothesize it's because smell is "Sparse". I believe we have so many receptors with such specificity that it's less useful to think in terms of "dense" parameters. In comparisson with other senses, for music we can locate a sound in a pitch spectrum since we have many receptors (millions?) with sensitivity across a single dimension (frequency). For color, we have just three receptors, lending perceived elements to be put in a nice three dimensional space. Now with smells, we have so many different and so specific receptors that there's neither a canonical order for the receptors to be put across a dimension (the case with sound) or a dense set within a few dimensions (vision, taste). So we just associate each receptor with the most common source.

Nice analysis. I'd also add that smell is a lower-brain function. We react to smells; we recognize smells; we have visceral responses to smells. But our language must label smells by the indirect response we experience. So we call them by other-sense names (sharp; hot) or by source.

Synesthetic metaphors for sensory experiences are the norm, even among sight and sound.

Colors are usually described as "loud", "quiet", "cool", "warm", "hot", "soft".

This morning I read an album review that described the music as -- “radiant”, "textured", with a "shimmering glow", "a gilded set of songs", "seemingly to soar in a striking display of absolute iridescence".

That sort of language just helps us triangulate an aesthetic experience using the language of our other senses, it doesn't mean a sense is any less direct.

But smells are almost exclusively described in that way. There is no spectrum; there is no map like touch. That's the whole point of the OP, right?

No, not at all. Smell is the least synesthetically described sense.

Generally with smell we call a smell for exactly what produces that smell: "oak-y", "fecal", "flowery", "smokey" and so on.

That is profoundly less synesthetic than our descriptors for the other senses, which are almost always described in metaphor -- bright sounds, loud colors.

The very idea we have less smell words is nonsense. We just have less abstract smell words. Consider if we used crayola color names to describe the colors, would that mean we had less color words because we're more specific with our language?

To add to that, I would say that we can't actually remember smells all that well, if at all. Its a pretty big stretch to remember the smell of hamburger, yet we can all recognize it in an instant.

Without the ability to remember something specific, its hard to effectively define it so that others can know exactly what you mean.

I'd argue that the memory of a smell is as visceral as the memory of a color or sound, which is not very.

The difference is that smell is so much more powerful in reality that we see through the memory-representation for what it is, an empty placeholder. When we smell fresh bacon in the morning, that experience can be all-encompassing. The same is not true for the optical experience of the thing.

With colors and sounds we are merely less aware of the qualitative emptiness of their mental objects.

Not sure about others, but when you said that I tried to recall the smell of a hamburger and found it immediately very easy.

I'm also hungry, and I can remember all kinds of food smells with ease right now...wonder if that has something to do with it, heh.

Well, the thought that leaps to my mind is that smell isn't like most of the rest of the senses. Eyes and ears are so easy to quantify and abstract with quantities that we've been doing it for decades now. Even if our eyes are not literally CCDs and our ears are not literally microphones, they're close enough for the quantifications to be useful. Touch and proprioception back to a geometric shape (our body) which provides obvious spatial abstractions (higher, lower, harder, softer, etc). Taste (considered separately from smell, which is normally a bad idea but bear with me here) is based on a small number of types of sensors (sweet/salt/sour/bitter/umami, if there's one or two missing it doesn't change things much).

Smell is hundreds or thousands of distinct sensor types for various chemical characteristics, with no particular relationship to each other. We don't have a systematic coverage of some vector space of X dimensions, we have quasi-random point samplings that poorly populate an n-dimensional space. If it's hard to get abstract concepts out of that, it could be because it really is mathematically demonstrably difficult to get useful abstraction out of that source of data. It sounds to me like even those cultures that happen to have them are still very, very weakly covered compared to vision or sound.

Applies to colors, beyond the basic few. That blend of red and yellow is named after a fruit. Violet is a flower. Purple comes from the Greek word for certain dye-producing mollusks. And when we get into the names of paint shades or crayons, we get names like "burnt ochre" (which could be the name of a smell too).

Except we can describe unnamed colors relatively to nearby named colors. e.g. "That's burnt ochre with a touch more blue." Every graphic designer and photographer works this way when proofing prints.

This is only possible because color is the sum of only three distinct sensations. The dimensionality of smell is likely way to high for this.

> sweet/salt/sour/bitter/umami

Better than 'umami' is the native English word: savoury. I really don't know why folks feel the need to reach for a foreign word when we already have a sufficient one, other than wanted to sound fancy.

It's like 'cilantro' instead of coriander, or 'arugula' instead of rocket.

Savory metaphorically had its chance, but whiffed it due to decades of being left out of lists of all taste buds produced by people in the west that listed sweet/sour/bitter/salty. This is the same set of lists that also showed the now-discredited "zones" of where each taste bud lives. Savory also means more than just "the flavor detected by the taste buds in question". For all I know "umami" means more things in Japanese too, but in English the new foreign import word had a chance to get nailed to the exact right concept with no previous baggage, so it won out in this narrow usage. Note it gets used nowhere else, while savory continues to lead the rich, complicated life of a "full" English word.

Perhaps GP wouldn't object if it were spelled "oomaumy"? We could even imagine a proper fake Old Norse etymology for that.

The same way we describe smells by proxy (smells like a banana), Ancient Greeks could have have used to describe objects by comparison, rather than colors (bronze sky, sheep the color of wine and so on): http://serendip.brynmawr.edu/exchange/node/61

I feel like nobody who commented here actually read the article. You have people here talking about the way smells are wired to the brain, or some inherent property of smell itself, when the entire article is about two tribal groups that have words for smell in the same way we have words for colour. This shows that having few words for smell is not universal and is likely not due to some hard limitation in the way smell is processed in the brain.

Good point. But also there are those that cannot fathom how that can be, since our own sense wouldn't/couldn't allow that ("Inconceivable!") We assume those tribesmen must be wired differently, perhaps in an inheritable way that we don't share.

I think the reason has a physiological basis.

The other senses go to the thalamus and then to the cortex

Vision - Lateral geniculate nucleus. https://en.wikipedia.org/wiki/Visual_system#Lateral_genicula...

Hearing - Medial geniculate nucleus. https://en.wikipedia.org/wiki/Auditory_system#Medial_genicul...

Taste - Ventra posteromedial nucleus. https://en.wikipedia.org/wiki/Ventral_posteromedial_nucleus

Touch - Ventral posterior nucleus. https://en.wikipedia.org/wiki/Ventral_posterior_nucleus

Smell on the other hand goes first to the olfactory cortex, and is not routed through the thalamus.

https://en.wikipedia.org/wiki/Primary_sensory_areas

One interesting consequence is how powerful an influence smell can have on us. The olfactory cortex is in the limbic system which is associated with emotion and memory. This might explain why certain smells (like the smell of something your mother used to cook) can instantly bring back memories and emotions.

In English, there are only three dedicated smell words—stinky, fragrant, and musty

I think "acrid" also applies. It's technically applicable to taste as well, but it's never used in that context. For that matter, something can taste fragrant (eg rose lemonade).

Pungent feels similar to me. I've never heard of something having a pungent taste.

Blue cheese.

Also noisome

malodorous, noxious, reeking, stenchy, noisome, pungent.

In my mind, these correspond with a sign (good/bad) and magnitude, but leave out the direction of the vector. For example, we frequently say that "my dog came home reeking OF [skunk, deer poop, ...]". So "reeking" is communicating high intensity of something unpleasant, and we leave it to the "of" to specify what that strong unpleasant smell is.

How could I have forgotten "redolent". Though I think that usually requires a complement via "of"; "redolent of <something>".

> For that matter, something can taste fragrant (eg rose lemonade).

Or any whisky younger than 12 years whose name starts with "Glen".

Guy Deutscher argues that the reason that languages in industrialized countries have developed basic color terms is due to the rise of industrial dying and color matching technologies, and that the reason that human languages typically don't have basic smell terms is because we don't have things like painter's kits and color swatches for smell that would easily allow us to compare them. http://www.nytimes.com/2010/09/05/books/review/Bickerton-t.h...

A word's meaning is a social consensus is about a semiotic relationship, and unless you have good means of comparing one thing to another, it's hard to reach social consensus about the meaning.

After all, human color vision perceives a wide range, but many languages only have basic color terms for dark, light, and red. 19th-century scientific racism even turned this into an argument for the biological inferiority of non-Europeans.

I love his question to his daughter: "What color is the sky?"... because the answer is actually almost never "blue". The sky may be the thing most varied in color in our day-to-day life. http://www.radiolab.org/story/211213-sky-isnt-blue/

"Calvin & Hobbes" had a couple of strips that touched on this topic. Hobbes, as a tiger, has an extensive vocabulary to describe smells.

http://www.gocomics.com/calvinandhobbes/1995/02/13 http://www.gocomics.com/calvinandhobbes/1995/02/14

I've always been curious if one could discover the approximate working dimensionality of the nose by computing the number of principal components of a matrix of survey results of perceived similarity between many different smells.

Even better, if one gave names to the principal components, one can invent a suitable lexicon for describing smells.

I was wondering the same. I would guess the dimension is high (maybe in the 10s or 100s?), and smells should be rarely 'similar', because I believe most smells can be traced to a single compound or family of compounds (with shared structure). For example, "fishy" smell seems to be associated with a singe molecule, Trimethylamine [1].

I got a single hit from a google search:

"On the dimensionality of odor space" http://elifesciences.org/content/4/e07865

Match the smells to the molecule:

  farts          methyl mercaptan (thiol + methyl)
  durian/onion   ethyl mercaptan (thiol + ethyl)
  skunk          butyl mercaptan (thiol + butyl)
  coffee         furfuryl mercaptan (furan + 2-methanethiol)
  grapefruit     grapefruit mercaptan (cyclohexene + 1-methyl + 4R-isopropyl-2-thiol)
  lilac          alpha-terpineol (cyclohexene + 1-methyl + 4-isopropyl-2-ol)
  almond/cherry  benzaldehyde (phenyl + formyl)
  anise          anisaldehyde (phenyl + 1-formyl + 4-methoxy)
  cinnamon       cinnamadehyde (phenyl + acrolein)
  tar            cresol (phenyl + 1-hydroxy + (2 or 3 or 4)-methyl)
  smoke          guaiacol (phenyl + 1-hydroxy + 2-methoxy)
  vanilla        vanillin (phenyl + 1-hydroxy + 2-methoxy + 4-formyl)
  clove          eugenol (phenyl + 1-hydroxy + 2-methoxy + 4-allyl)
  oregano        carvacrol (phenyl + 1-hydroxy + 2-methyl + 5-isopropyl)
  thyme          thymol (phenyl + 1-hydroxy + 2-isopropyl + 5-methyl)
  mint           menthol (cyclohexanyl + 1R-hydroxy + 2S-isopropyl + 5R-methyl)
  caraway        [S+]-carvone (1-cyclohex-2-enone + 2-methyl + 5S-isopropenyl)
  spearmint      [R-]-carvone (1-cyclohex-2-enone + 2-methyl + 5R-isopropenyl)
  pine           [S-]-limonene (cyclohexene + 1-methyl + 4S-isopropenyl)
  orange         [R+]-limonene (cyclohexene + 1-methyl + 4R-isopropenyl)
  fish           trimethylamine
  cabbage/ocean  dimethyl sulfide
  butter         diacetyl + acetoin
  vomit          butyric acid
  pineapple      methyl butyrate
  orange         ethyl butyrate
  strawberry     propyl butyrate, butyl butyrate
  pear/apricot   pentyl butyrate
  glue           methyl acetate
  pear           ethyl acetate
  apple/banana   butyl acetate, amyl acetate
  vinegar        acetic acid
  goat cheese    capric acid, caprylic acids
  popcorn        6-acetyl-1,2,3,4-tetrahydro-pyridine
  coconut        gamma-nonalactone
  peach          gamma-undecalactone
  garlic         diallyl disulfide, diallyl trisulfide, diallyl tetrasulfide

So I'd say that olfaction primitives probably number at least in the low hundreds. It seems as though just substituting a single functional group on an aromatic ring or changing the length of a carbon chain can affect human perception of the odor. Chirality usually matters. Orientation of the double bonds usually matters. Most frustratingly, concentration also matters. Indole smells like jasmine flowers at low concentrations and like poop at higher concentrations.

It may well be that we don't have dedicated words for smells because the IUPAC chemical naming conventions make them unnecessary. Also, most people don't need them. Professional aromatists would probably just say the chemical name, or use the perfumery primitives like the plant essential oils or resins, ambergris, civet, castoreum, hyrax, musk, and beeswax.

Esters are fruity. Linear terpenes are grassy or herby. Cyclic terpenes are woody or flowery. Substituted aromatics are flowery or herby. Amines are rotten or fishy. Thiols just plain stink.

Wine snobs can probably just make up whatever words they want to describe their noses and bouquets, because their words don't really transmit meaning most of the time, anyway.

And here I was thinking "wait, most programming languages don't have words for smells[1] because they are abstract concepts, not concrete features..."

Then I saw the article. Oops.

[1] https://en.wikipedia.org/wiki/Code_smell

I think the opposite is true - we have too many words. And it's because smell receptors are of such high dimensionality that we literally have to throw arbitrary and verbose vocabulary against it to try and describe it. Just listen to the wine connoisseurs describing their beverages.

Then again blind tests proved wine tasting hogwash http://www.theguardian.com/lifeandstyle/2013/jun/23/wine-tas...

My wife has synesthesia (https://en.wikipedia.org/wiki/Synesthesia) in which every word or name comes with a taste. Admittedly, it's not exactly the same as a smell, but it's funny how she describes a name. It's most often indeed not a single "crisp thing", but an elaborate description. A certain type of butter in particular state, etc. Of course I test her every year, but it's remarkably consistent. :-)

I often like to come up with new words for concepts. One of them is the equivalent for being blind/deaf, but then for smell. How would I describe myself if I can't smell things?

I'm smeff?

In french there are more words for wine than in inuit for snow (http://www.auduteau.net/oenologie/vocabulaire.shtml)

The "Eskimos have hundreds of words for snow" is actually bad linguistics--for pointers, see Mark Liberman's post on Language Log (http://itre.cis.upenn.edu/~myl/languagelog/archives/003286.h...).

I suspect that these "extra smells" are subjective, and that there's no chemical basis for any of them.

When it talks about how brothers and sisters can't sit too close together because "their smells would mingle", they could have easily substituted in the words "auras" or "mojo", and it would have the same semantic meaning.

It's more interesting to speculate on why they would develop these subjective values in the first place, than it is to wonder why they've confused those values with actual odors.

Because there are so many easy ways to describe smells as similar to known smells: like frying bacon, like unripe bananas, like ripe bananas, like overripe bannanas, etc., etc..

Very timely. Just the other day I booted an old Mac Mini Core (not 2) Duo and found Firefox to be very quick, wishing modern browsers didn't have to be resource wasteful.

um... what?

Because describing smells in detail is rarely important?

My wild guess, because we don't have easy ways to create them. Anyway, it's a bit weird considering the extreme relationship between smell, memories and emotions.

> Because describing smells in detail is rarely important?

That may be tautological. Smell describes the chemical composition of air and the substances it carries. I don't think that information is commonly trivial. One could similarly say "we rarely talk about UV radiation because it is unimportant."

Maybe I could have put that better. I'm trying to say that describing the composition of a smell is rarely important. In the typical case what matters is either that it's the smell of some specific thing (I smell fish and chips - that makes me hungry, I smell fire, I smell gas) or that it's pleasant (you smell nice) or undesirable (you stink). Composition, though, mostly just matters when describing perfume or something.

But yeah this might be a little tautological.

This stinks.

Reply on news.ycombinator.com