Why Coffee Is Difficult to Describe

What happens between sensing a coffee and finding words for it

Three experienced tasters evaluate the same coffee.

One writes peach.
Another writes apricot.
The third writes stone fruit.

Who is right?

The easiest explanation is that taste is subjective. It is also an incomplete one.

The three tasters may genuinely have perceived different qualities. But the difference may have emerged elsewhere: in what each person attended to, the memories available to them, the categories through which they organized the experience, the degree of specificity they considered justified, or simply the word they were able to retrieve at that moment.

There is a linguistic distinction here too. Stone fruit is a broader category—a hypernym—while peach and apricot are hyponyms of stone fruit and co-hyponyms of one another. They sit at roughly the same level of specificity. So the three words do not carry the same information, but neither do they necessarily contradict one another.

This is one of the central difficulties of coffee evaluation: sensory experience does not arrive already labeled.

Before a taster can describe a coffee, they must attend to a complex experience, distinguish qualities within it, recognize something familiar, relate it to previous encounters, place it within a category, and select language through which it can be shared. These are not perfectly separate stages, nor do they always occur in that order. They interact, overlap, and sometimes interfere with one another.

Description is not a direct transcription of the cup.

But that does not make it arbitrary.

The cup does not come with words

What we casually call the taste of coffee is already a much broader experience of flavor.

The gustatory system contributes sensations such as sweetness, sourness, bitterness, saltiness, and umami. Olfaction contributes much of what allows us to recognize coffee as floral, fruity, spicy, or chocolate-like, through orthonasal smelling before drinking and retronasal olfaction during consumption. Oral somatosensation contributes temperature, texture, viscosity, and astringency, alongside chemesthetic sensations mediated largely through the trigeminal system. 

These signals do not remain neatly separated in experience. They are integrated into the perceptual whole we call flavor. This is also why the same coffee can be experienced somewhat differently as temperature, context, and attention change.

Yet this integrated experience is not presented to consciousness as a completed list:

Jasmine.
Peach.
Bright acidity.
Silky mouthfeel.
Lingering aftertaste.

Those are descriptions we construct from the experience.

Anyone who tastes carefully knows the small distance between the two. Something is there. It feels familiar, perhaps even obvious, but the word does not immediately follow. 

You taste again.

Sweet. Floral. Familiar. Something fruity.

You may know two cups are different without being able to explain the difference. You may recognize an aroma as familiar but spend several seconds searching for cardamom. Or you may retrieve a word and still hesitate, unsure whether it captures what you perceived closely enough to say it aloud.

This is why “tasting ability” hides several different abilities.

Discrimination is noticing that one sensation differs from another.
Recognition is experiencing something as familiar.
Identification is associating it with a likely source or category.
Lexical access is retrieving a word.
Description is choosing language appropriate to the situation.
Evaluation is deciding what that perceived quality means for the coffee’s value or quality.

Being good at one part does not guarantee being equally good at another. Much of what we call sensory expertise takes place somewhere inside these distinctions.

Recognizing a smell is not the same as naming it

This distinction is especially important in coffee because much of its recognizable complexity is aromatic.

Speakers of English and several other extensively studied Western languages often find familiar odors surprisingly difficult to name. People may recognize that an odor is familiar, and may even distinguish it reliably from others, while struggling to retrieve its name. Olofsson and Gottfried (2015) describe odor naming as a process involving perceptual processing, lexical-semantic integration, and verbalization. The difficulty does not necessarily begin with smelling itself.

Anyone who has taught sensory classes will recognize the moment:

I know this smell.

It is something sweet.

Maybe a fruit.

Not citrus.

I can almost name it.

The experience has sometimes been described as a kind of “tip-of-the-nose” phenomenon: the sensory recognition is present while the linguistic label remains frustratingly out of reach.

It is tempting to conclude that smell is naturally beyond language. But that conclusion reflects only a narrow portion of human linguistic experience.

Asifa Majid and Niclas Burenhult (2014) compared English speakers with speakers of Jahai, a language of the Malay Peninsula with a conventional vocabulary for odor qualities. Jahai speakers named odors with a level of concision and agreement comparable to their naming of colors. English-speaking participants found the odor task considerably more difficult, despite being more familiar with many of the odor sources used. The study does not show that every Jahai speaker names every smell perfectly. It shows something more important: difficulty describing odors is not a universal and unavoidable property of olfaction. 

Languages and communities differ in the sensory distinctions they habitually encode, the vocabulary they make readily available, and the practices through which those words are learned (Winter, 2019).

English-language coffee terminology has developed an extensive professional vocabulary, but much of it remains source-based. We describe an aroma as jasmine, peach, cocoa, or black tea. We usually identify the experience by naming something associated with it rather than by using an independent term for the odor quality itself.

This strategy is useful, but it also means that description depends heavily on remembered objects and experiences. That creates another difficulty.

A flavor note is a comparison

When I write peach on a cupping form, I am not claiming that there is peach in the coffee.

I am proposing a resemblance.

Some aspect of the coffee reminds me of some aspect of what I understand as peach: perhaps its aroma, sweetness, acidity, ripeness, juiciness, skin-like astringency, or a combination of these qualities.

But peach is not one fixed sensory object.

A white peach differs from a yellow peach. The skin differs from the flesh. An underripe peach differs from a ripe one. Fresh peach differs from dried, cooked, canned, or fermented peach. Varieties available in one region may be unfamiliar in another.

The descriptor compresses this variation into a single word.

That compression is not a defect unique to sensory language. Language always selects. It allows us to draw a meaningful pattern from an experience too detailed to reproduce in full. The problem begins when we treat the selected word as though it were identical to the perception—or as though it must activate the same reference for every listener.

A flavor note is therefore relational. It connects a present sensory experience with prior knowledge.

A taster who has never encountered a particular fruit cannot use it in exactly the same way as someone who grew up eating it. A person may know the word but possess only a weak sensory reference. Another may have a detailed reference but use a broader category because they are not confident that greater precision is justified.

The words available to us and the experiences behind those words are related, but they are not the same thing.

Categories do not simply mirror the world

To describe a sensory experience, we must decide what kind of thing it resembles.

This is an act of categorization.

Traditional accounts often treat categories as containers with fixed boundaries: an object either belongs or it does not. Research in cognitive psychology and cognitive linguistics has shown that many everyday categories are structured less rigidly. Some members are more representative than others, and people can organize the same domain at different levels of abstraction (Lakoff, 1987).

William Labov (1973) demonstrated this elegantly in his classic work on the meanings of words such as cup, mug, and bowl. Participants were shown a series of container-like objects whose proportions gradually changed. As the objects became wider, people became less consistent in calling them cups and increasingly called them bowls. More importantly, changing the imagined context changed the naming boundary: an ambiguous object imagined in the context of drinking coffee or tea was more readily treated as a cup, while placing it in a food context shifted responses toward bowl.

The object itself could remain the same while the category shifted with context. 

Coffee flavor categories are not identical to Labov’s household objects, of course. But the linguistic principle is useful: word boundaries need not correspond to perfectly fixed boundaries in the world. Categories can be graded, context-sensitive, and influenced by what we are trying to do with them.

Coffee makes this even more complicated because what we are categorizing is not a stable object sitting in front of us, but a transient sensory experience being compared with remembered references.

Consider the sequence:

Fruit
Stone fruit
Peach
Yellow peach
Ripe yellow-peach flesh

Each expression narrows the field. But greater specificity is not automatically greater accuracy.

A broad description may be the most responsible choice when the perceptual evidence is unclear. A highly specific term may reflect exceptional familiarity—or unjustified confidence. The appropriate level depends on the strength of the sensation, the taster’s references, and the purpose of the evaluation.

This is why the opening example cannot be resolved by counting identical words.

The person who writes stone fruit may be recognizing the same general sensory family as the tasters who write peach and apricot, but choosing not to go further. The two more specific tasters may share substantial perceptual agreement while drawing on different sensory references. Alternatively, each may be attending to a genuinely different feature of the coffee.

The words alone do not tell us which explanation is correct.

Broad categories can be useful, but they can also become too vague to communicate much at all.

Consider tropical fruit.

It appears frequently in coffee descriptions, but what sensory information does it actually communicate?

Mango, pineapple, passion fruit, banana, papaya, guava, and lychee can differ enormously in aroma, sweetness, acidity, texture, and overall flavor character. To place them all under tropical fruit may identify a loose conceptual family, but it tells another taster relatively little about which sensory experience was actually perceived.

As an intermediate thought—I know I am somewhere in this area—the category may be useful. As a final professional descriptor, I find it much less useful.

Descriptive language should reduce uncertainty between people. A term so broad that two tasters can use it while thinking of very different sensory profiles can create the appearance of agreement without much mutual understanding.

The same problem appears in phrases such as red fruit and yellow fruit. A banana and a lemon are both yellow; a strawberry and a red apple are both red. Color alone does not explain which aromatic or flavor characteristics supposedly make those fruits a coherent sensory group. If such expressions are to function professionally, their meaning needs to be more constrained than the color of the fruit.

And then there is berry.

This example is useful for a different reason. In everyday English—and in coffee sensory language—berry comfortably brings to mind things such as strawberry, raspberry, blueberry, and blackberry. Yet botanical classification organizes the category differently: a banana qualifies botanically as a berry, while a strawberry does not.

The point is not that coffee tasters should start calling bananas berries. It is that berry does not carry one universal category structure independently of context. Botany uses the word according to how a fruit develops. Everyday language organizes it differently. Coffee sensory practice uses it according to another set of perceived similarities and conventions.

Communication works only to the extent that the people using the word understand enough of the same system.

Different descriptors do not necessarily prove different perceptions. The same descriptor does not prove identical perceptions either.

Linguistic agreement is evidence of coordination, not direct access to another person’s private experience.

Expertise organizes interpretation

Training matters.

Experienced tasters generally possess a larger set of references, greater familiarity with professional tasks, and more practice directing attention toward relevant sensory qualities. Repeated comparison can improve discrimination, recognition, and the ability to communicate within a particular field.

But expertise does not turn a person into a neutral measuring instrument.

Croijmans and Majid (2016) compared wine experts, coffee experts, and novices and found a modest, domain-specific advantage in flavor description. Coffee experts used more concrete, source-based terms and fewer general evaluative expressions, but in their experiment they were not more consistent with one another in their coffee descriptions. Their broader conclusion was that communicating smells and flavors involves not only perceptual training but linguistic practice as well.

That study should be interpreted carefully. Its category of “coffee experts” is not equivalent to a calibrated panel of certified Q Graders, and naming performance is not the same thing as overall sensory competence.

Research involving Q Graders adds an important counterpoint. Pereira et al. (2019) found high precision among seven Q Graders when evaluating excellent and outstanding coffees, while difficulties increased around a classification boundary. Fioresi et al. (2023) later examined concordance among Q Graders specifically, again showing why agreement between trained assessors has to be measured rather than assumed.

Taken together, these studies suggest something more useful than either “experts are objective” or “experts disagree, so training does not matter.”

Training can improve our ability to notice relevant differences and make professional vocabulary more familiar, but it does not ensure that two tasters will name the same experience in the same way. Expertise does not remove interpretation; it gives it more structure.

The question is not whether trained tasters are subjective or objective. The more useful question is how training makes individual judgments more attentive, repeatable, explainable, and comparable.

Description has to work between people

A sensory description is not complete simply because it feels accurate to the person who produces it.

It must do something for someone else.

At a professional table, words may be used to compare samples, explain a purchasing decision, diagnose a roasting problem, give feedback to producers, teach students, justify a score, or coordinate a judging panel. A descriptor becomes part of a shared practice.

That is why calibration cannot be reduced to asking whether everyone perceives exactly the same thing. 

The question becomes: “What do I mean by this word, and do you understand enough of what I mean for us to work with it?”

We cannot directly inspect another person’s sensory experience. What we can examine is how judgments behave: whether assessors distinguish samples consistently, use references in comparable ways, understand attributes and scales similarly, explain their decisions, and revise those decisions when presented with stronger evidence.

Calibration therefore involves perception, but it also involves categories, definitions, references, procedures, scales, and language.

This does not mean that language simply determines what is tasted. Evidence for linguistic effects in one sensory domain cannot automatically be transferred to another. For example, a cross-linguistic study by Bylund et al. (2024) found no evidence that different English and Norwegian lexical categories for edible bulbs produced corresponding differences in participants’ taste-similarity judgments. The relationship between language and perception depends on the sensory modality, the task, and the kind of linguistic distinction being tested. 

Language participates in sensory evaluation. It does not command it.

Professional calibration can instead be understood as a form of structured intersubjective agreement: individual perceptions are brought into a disciplined framework through shared methods, references, categories, discussion, and accountability.

This is not universal objective truth. But neither is it a license for every description to be treated as equally useful.

A sensory claim that can be recognized repeatedly, connected to a shared reference, explained to others, and used consistently within a defined task has more professional value than an association that cannot be reproduced or understood consistently.

The fact that a subject is involved does not make the judgment random.

The difficulty is not a failure of language

Coffee is difficult to describe because a great deal happens between sensation and words.

A changing multisensory experience must be attended to. Qualities must be separated from the whole. Familiarity must become recognition. Recognition must connect with memory. Memory must support a category. A word must be retrieved and selected. That word must then be interpreted by someone whose experiences, categories, and purposes may differ.

At each point, something is preserved and something is reduced.

Peach does not reproduce the sensation. Bright carries no acidity of its own. Clean does not arrive with one fixed meaning. Tropical fruit may conceal more sensory difference than it reveals. And a term such as flat can move between sensation, description, and evaluation so easily that two experienced tasters may use it confidently while referring to different things.

Yet without this reduction, sensory experience would remain largely private. Language makes comparison, teaching, evaluation, and collective knowledge possible.

So the aim is not simply to produce longer flavor lists or search for a perfect vocabulary. I am more interested in what happens underneath the words we already use: what a flavor note actually refers to, why some categories are more useful than others, how sensory terms move between people and languages, and what we are really aligning when we calibrate.

There is a lot hidden inside something as ordinary as saying that a coffee tastes like peach.

Coffee Language begins there.

Selected references

Bylund, E., Samuel, S., & Athanasopoulos, P. (2024). Crosslinguistic differences in food labels do not yield differences in taste perception. Language Learning, 74(S1), 20–39. https://doi.org/10.1111/lang.12641

Croijmans, I., & Majid, A. (2016). Not all flavor expertise is equal: The language of wine and coffee experts. PLOS ONE, 11(6), e0155845. https://doi.org/10.1371/journal.pone.0155845

Fioresi, D. B., Ramos, A. C., Bertolazi, A. A., & Pereira, L. L. (2023). Adherence and concordance among Q-Graders in the sensory analysis of coffees. Journal of Sensory Studies, 38(2), e12805. https://doi.org/10.1111/joss.12805

Labov, W. (1973). The boundaries of words and their meanings. In C.-J. N. Bailey & R. W. Shuy (Eds.), New ways of analyzing variation in English (pp. 340–373). Georgetown University Press. 

Lakoff, G. (1987). Women, fire, and dangerous things: What categories reveal about the mind. University of Chicago Press. 

Majid, A., & Burenhult, N. (2014). Odors are expressible in language, as long as you speak the right language. Cognition, 130(2), 266–270. https://doi.org/10.1016/j.cognition.2013.11.004

Olofsson, J. K., & Gottfried, J. A. (2015). The muted sense: Neurocognitive limitations of olfactory language. Trends in Cognitive Sciences, 19(6), 314–321. https://doi.org/10.1016/j.tics.2015.04.007

Pereira, L. L., Guarçoni, R. C., Moreira, T. R., de Sousa, L. H. B. P., Cardoso, W. S., Moreli, A. P., da Silva, S. F., & Ten Caten, C. S. (2019). Very beyond subjectivity: The limit of accuracy of Q-Graders. Journal of Texture Studies, 50(2), 172–184. https://doi.org/10.1111/jtxs.12390

Winter, B. (2019). Sensory linguistics: Language, perception and metaphor. John Benjamins. https://doi.org/10.1075/celcr.20