What Exactly Are You Counting? — What Yucatec Maya Reveals About Objects, Substance, and Number

A number tells us how many. But before we can count, we must know what qualifies as one.

— Tymur Levitin

Imagine a table with three things on it.

A banana.

A piece of paper.

A glass of water.

You point to the banana and say:

One banana.

Then you point to the paper:

One piece of paper.

Finally, you point to the water:

One glass of water.

All three expressions contain the number one.

But what exactly have you counted?

In the first case, you counted an individual fruit.

In the second, a piece of material.

In the third, a container holding a quantity of liquid.

Now imagine that the banana is still attached to a plant.

You could count:

One banana.

One banana leaf.

One banana plant.

One bunch of bananas.

One portion of banana.

The same word, banana, can participate in several different descriptions.

Yet the units are not identical.

A fruit is not a leaf.

A leaf is not a plant.

A plant is not a bunch.

And a portion of banana is not necessarily a whole fruit.

English can distinguish these possibilities.

But it does not always require us to make the distinction explicit.

We can simply say:

Three bananas.

In Yucatec Maya, also known as Maaya T’aan, the relationship between numbers, nouns and the units being counted is organized differently.

The language has a system of numeral classifiers.

These grammatical elements occur with numerals and help identify the category or unit relevant to counting.

Some distinguish animate from inanimate entities.

Others are associated with plants, elongated objects, flat objects, portions or measures.

This raises a question that is much deeper than vocabulary:

When we count something, are we counting objects that already exist as obvious units, or are we selecting a particular way of dividing reality into countable units?

Yucatec Maya offers an extraordinary opportunity to investigate that question.

But it also requires scientific caution.

We must not assume that speakers of different languages inhabit different physical worlds.

Nor should we assume that every noun can be counted through any classifier whatsoever.

The interesting reality is more precise.

Languages differ in how they grammatically organize the relationship between number, object, substance and unit.

1. Counting is not as simple as it appears

Consider the sentence:

There are three apples on the table.

It seems straightforward.

Three individual apples.

But now consider:

There are three pieces of apple on the table.

Perhaps one apple has been cut into three pieces.

The quantity of fruit may be the same.

The number of counted units is different.

Now consider:

There are three bags of apples on the table.

We are no longer counting individual apples.

We are counting containers or packages.

The number three remains unchanged.

But the unit changes.

This gives us a fundamental distinction:

A number does not identify its own unit.

The unit must be supplied by the linguistic expression, the situation or shared knowledge.

2. What exactly is one?

Imagine a loaf of bread.

You can count:

One loaf.

One slice.

One piece.

One crumb.

One portion.

Each expression establishes a different unit.

Now imagine a tree.

One tree.

One branch.

One leaf.

One piece of wood.

One pile of firewood.

Again, the number may be one.

But the thing counted changes.

This is not merely a grammatical curiosity.

It is a fundamental operation involved in measurement, mathematics and everyday reasoning.

Before counting, we must identify what constitutes a countable unit.

In linguistics, this process is often discussed in terms of individuation.

3. What is individuation?

Individuation is the process of identifying or treating something as a distinct entity.

In everyday experience, some things appear naturally separated.

A person.

A dog.

A stone.

A chair.

Other things are less obviously divided into individual units.

Water.

Sand.

Rice.

Flour.

Smoke.

We can count individual grains of rice.

But we normally do not count rice itself in the same way that we count chairs.

Instead, we count:

Grains of rice.

Bags of rice.

Bowls of rice.

Portions of rice.

The unit is supplied through an additional expression.

Individuation is therefore closely connected to how language represents countability.

4. English distinguishes count nouns and mass nouns

English grammar provides a familiar contrast.

Count nouns:

A chair.

Two chairs.

Three chairs.

Mass nouns:

Water.

Sand.

Furniture.

Information.

We generally cannot use ordinary English count constructions such as:

Three waters.

Two furnitures.

Five informations.

without special interpretations or nonstandard usage.

Instead, we say:

Three glasses of water.

Two pieces of furniture.

Five pieces of information.

However, even this distinction is not absolute.

In a restaurant, three waters can mean three servings or containers of water.

The expression is understandable because the context supplies a unit.

This demonstrates that countability is not always an immutable property of a physical substance.

It can also depend on how the speaker conceptualizes and presents it.

5. Yucatec Maya organizes counting differently

Yucatec Maya belongs to the Mayan language family and is spoken primarily in Mexico’s Yucatán Peninsula and neighbouring areas.

It has a long linguistic history and a substantial body of descriptive research.

One important feature is its numeral-classifier system.

In a typical counting construction, a numeral combines with a classifier.

The classifier contributes information about the category or unit being counted.

The general pattern can be represented as:

NUMERAL + CLASSIFIER + NOUN

For example:

ka’a túul peek’

Meaning:

Two dogs.

Here:

ka’a = two

túul = classifier associated with animate beings

peek’ = dog

The expression does not simply place the number directly before the noun.

A classifier participates in the counting construction.

6. The difference between two dogs and two houses

Consider two documented patterns.

ka’a túul peek’

Two dogs.

And:

ka’a p’éel

Two inanimate units, when the relevant noun is understood from context.

The classifiers differ.

túul is associated with animate entities.

p’éel is associated with inanimate entities and has also developed a broader general counting function.

This means that a number may be grammatically accompanied by information about the class of entity being counted.

The difference is not that Yucatec Maya speakers possess a special mathematical ability.

It is that the grammatical structure of counting expressions differs from English.

7. The basic classifiers túul and p’éel

Two especially important classifiers are:

These categories are useful for understanding the system.

But they should not be treated as completely rigid or historically unchanging.

Modern Yucatec Maya shows variation in classifier usage.

The general classifier p’éel has expanded into contexts where more specific classifiers may traditionally have been used.

This is important because real languages change.

A grammatical description based on one community or historical period may not capture every contemporary pattern.

8. A classifier is not the same thing as a noun

In English:

Three dogs.

The noun identifies the entity.

The number identifies the quantity.

In Yucatec Maya:

ka’a túul peek’

The classifier adds another grammatical component.

But túul does not simply mean dog.

It is used with other animate referents.

It identifies a grammatical class relevant to the counting construction.

The noun still provides the lexical identity of the entity.

This distinction is essential.

A classifier does not necessarily replace the noun.

It participates in how the noun is counted.

9. A classifier can appear without an overt noun

Now imagine that two people are discussing objects already visible in the room.

One person asks:

How many?

The other answers:

Two.

English allows the noun to remain unspoken because the context identifies the relevant objects.

Yucatec Maya also permits numeral-classifier expressions without an overt following noun.

For example:

ka’a p’éel

can function as:

Two [inanimate ones].

The classifier remains part of the expression.

This reveals an important difference.

The grammatical representation of the counted category can survive even when the lexical noun is omitted.

10. The banana problem

One of the most famous discussions of Yucatec Maya numeral classifiers involves bananas.

English uses the word banana primarily for the fruit, although related expressions can refer to the plant or its parts.

Yucatec Maya has been discussed in connection with several classifier constructions involving the noun ha’s.

These constructions illustrate different ways of selecting a countable entity associated with bananas.

For example:

hun-ts’íit ha’s

One elongated banana fruit.

hun-wáal ha’s

One banana leaf.

hun-kúul ha’s

One banana plant.

hun-kúuch ha’s

One load or bunch of bananas.

hun-p’íit ha’s

A small amount or portion of banana.

These examples are important.

But they require careful interpretation.

11. Are these really five ways to count the same object?

At first glance, the banana examples seem to show that one noun can refer to several different units depending on the classifier.

That is partly true.

But we must not exaggerate the conclusion.

A banana fruit and a banana leaf are physically different objects.

A banana plant is not merely the same fruit viewed differently.

A bunch of bananas is a collection.

A portion of banana is a quantity of material.

The classifier helps identify the intended unit.

But it does not magically transform one physical object into another.

The language distinguishes different referents associated with a lexical domain.

This is an important correction to simplified accounts of linguistic relativity.

12. The same word does not guarantee the same referent

Consider English:

Chicken.

This word can refer to:

A living animal.

Meat from that animal.

A portion of chicken.

A prepared dish.

Now compare:

Three chickens.

Three pieces of chicken.

Three portions of chicken.

The same lexical root appears.

But the counted entities differ.

The first counts animals.

The second counts pieces of meat.

The third counts servings.

This illustrates why the banana examples should not be treated as proof that Yucatec Maya speakers cannot distinguish objects without classifiers.

The relevant distinction concerns the relationship between lexical meaning and the unit selected for counting.

13. The classifier ts’íit

The classifier ts’íit is associated with elongated, slender entities.

In the banana examples, it is used for an individual banana fruit.

The classifier contributes information about the kind of unit being counted.

But we should not interpret it as a universal mathematical measurement of length.

It belongs to a conventional grammatical system.

Its distribution depends on linguistic usage.

A long object is not automatically compatible with every construction involving this classifier.

The category must be understood through documented examples and grammatical conventions.

14. The classifier wáal

The classifier wáal is associated with flat or leaf-like entities.

In the banana examples, it identifies a banana leaf.

This illustrates how a classifier can select a particular part or form associated with a broader lexical domain.

The physical distinction is familiar.

A leaf differs from a fruit.

But the language can integrate that distinction into the numeral expression.

This is the important grammatical fact.

The classifier helps determine what counts as one in the construction.

15. The classifier kúul

The classifier kúul is associated with plants and trees.

Consider the documented expression:

jun kúul che’

Meaning:

One tree.

The structure contains:

jun = one

kúul = plant/tree classifier

che’ = tree or wood

This example shows that the classifier can identify a plant as the relevant countable unit.

It does not simply describe the material.

It participates in identifying the entity being counted.

16. A tree and a piece of wood are not the same unit

Imagine a tree standing in a garden.

Now imagine a branch cut from that tree.

Then imagine the branch divided into several pieces.

You could count:

One tree.

One branch.

Three pieces of wood.

The physical material may be related.

But the units are different.

Yucatec Maya classifier constructions provide grammatical resources for expressing distinctions of this kind.

The general principle is not that English cannot make these distinctions.

It can.

The difference is how the counting expression is organized.

17. The classifier kúuch

The classifier kúuch is associated with loads or burdens.

In the banana examples, it can identify a bunch or load of bananas.

This introduces a different type of unit.

We are no longer counting an individual fruit.

Nor a leaf.

Nor a plant.

We are counting a collection organized as a load.

This is particularly interesting because the unit depends on a practical configuration.

A load is not simply a geometric shape.

It is an arrangement of material relevant to carrying or transporting.

The classification therefore connects counting with ordinary human activities.

18. Counting a load is not counting its contents

Suppose a basket contains twenty oranges.

You can say:

One basket of oranges.

Or:

Twenty oranges.

Both statements may describe the same physical situation.

But they count different units.

One counts the container or load.

The other counts the individual fruits.

Now imagine that the basket is divided into two smaller baskets.

The number of oranges may remain unchanged.

But the number of containers changes.

This is why counting always depends on the selected unit.

The number alone does not determine what has been counted.

19. The classifier p’íit

The form p’íit is associated with a small amount or portion.

A documented example is:

jun p’íit taak’in

Meaning:

A little bit of money.

Another common type of example involves water:

un p’íit ha’

A little bit of water.

Here the expression does not identify a whole individual object in the same way as one tree.

It selects an amount.

This shows that numeral-classifier constructions can involve measurement and portioning, not only classification of discrete objects.

20. A portion is not necessarily an object

Consider water.

You can count:

One glass of water.

One bottle of water.

One litre of water.

One drop of water.

These expressions establish different units.

But the water itself does not come naturally divided into glasses or litres.

The unit is imposed through measurement or containment.

Yucatec Maya has grammatical resources for distinguishing such quantitative relationships.

This is where the study of numeral classifiers intersects with the distinction between objects and substances.

21. The difference between sortal and mensural classifiers

Linguists often distinguish two broad types of classifier functions.

Sortal classifiers classify entities according to properties such as animacy, shape or conventional object category.

Mensural classifiers identify quantities, measures or portions.

Consider:

One tree.

The classifier identifies a kind of entity.

Now consider:

One portion of water.

The expression identifies an amount.

These are related but different operations.

The first classifies the counted entity.

The second establishes a measurable unit.

Yucatec Maya is particularly interesting because both functions appear within its broader numeral-classifier system.

22. Why this distinction matters

Suppose someone says:

Three pieces of wood.

Are they counting three independent objects?

Or three portions of a material?

The answer depends on the interpretation.

A piece of wood can be treated as a discrete object.

But the expression also involves partitioning a larger material.

Similarly:

Three slices of bread.

The slices are countable.

Yet they may have originated from one loaf.

The linguistic unit does not necessarily correspond to an independently existing natural object.

It may be produced through cutting, dividing, packaging or measuring.

This is the deeper problem of individuation.

23. What is a natural unit?

A person is usually perceived as an individual.

A dog is usually perceived as an individual.

A chair is usually perceived as an individual.

But what about a pile of sand?

A cloud?

A forest?

A crowd?

A river?

The boundaries are not always equally obvious.

A forest contains many trees.

A crowd contains many people.

A river consists of flowing water.

Yet we can refer to each as a single entity.

This demonstrates that linguistic individuation is not always identical to physical separability.

Human categorization operates at different levels.

24. English already contains several systems of individuation

English speakers can say:

A forest.

Three trees.

Thousands of leaves.

A pile of wood.

Several pieces of timber.

These expressions organize the same broad physical domain into different units.

The language provides lexical and grammatical resources for selecting the intended level.

Yucatec Maya differs in the way numeral classifiers participate in such constructions.

The comparison is not between a language that understands units and one that does not.

It is between languages that distribute unit-selection information differently.

25. Why numeral classifiers are not simply adjectives

An adjective describes a property.

For example:

Three tall trees.

The adjective tall contributes descriptive information.

But the trees are still counted as individual trees.

A numeral classifier has a different grammatical role.

It participates in the numeral construction and helps identify the relevant class or unit.

This does not mean classifiers cannot originate historically from nouns or other lexical material.

Many do.

But their grammatical function in counting constructions differs from that of ordinary descriptive adjectives.

The distinction is structural, not merely semantic.

26. The history of classifiers

Christian Lehmann’s research on Yucatec Maya grammaticalization examines the historical development of numeral classifiers.

Some classifiers have identifiable lexical origins.

For example, forms associated with branches, sheets, plants and other physical entities can develop classificatory functions.

A word that originally names an object or material can become part of a grammatical system for counting other entities.

This is a familiar pathway in language change.

Lexical meaning can gradually become more grammatical.

However, not every classifier has a transparent origin.

And not every classifier follows exactly the same historical development.

The system contains several layers of linguistic history.

27. How a noun can become a classifier

Imagine a language in which speakers frequently say:

One sheet of paper.

Two sheets of paper.

Three sheets of paper.

Over time, a word meaning sheet may become strongly associated with counting flat entities.

Its use may extend beyond the original material.

Eventually, it may develop a more specialized grammatical function.

This is a conceptual illustration of a possible grammaticalization pathway.

It is not a claim that every Yucatec Maya classifier developed through this exact sequence.

The important principle is that words can move from naming things to organizing grammatical relationships.

28. The classifier p’éel is changing

One of the most important findings of recent research concerns the classifier p’éel.

Traditionally associated with inanimate objects, it has expanded into broader use in contemporary Yucatec Maya.

Barbara Blaha Pfeiler and Stavros Skopeteas examine this development in their 2024 study:

Numeral Classifiers in Yucatec Maya: Microvariation and Syntactic Change.

They identify variation involving the use of the general classifier instead of more specific classifiers.

They also examine constructions where the general classifier appears alongside expressions of measurement.

This means the system is not frozen.

Speakers and communities differ in how particular counting constructions are organized.

29. Why language change matters for our interpretation

Imagine a grammar textbook describing a system with many specific classifiers.

Now imagine a younger speaker who frequently uses a general classifier instead.

Does that mean the speaker is using the language incorrectly?

Not necessarily.

It may reflect an ongoing grammatical change.

Languages regularly develop more general constructions.

Specialized distinctions can weaken.

New patterns can emerge.

The relevant question is not whether contemporary usage perfectly reproduces an older description.

It is what patterns speakers actually use.

A scientifically responsible account must distinguish traditional grammatical descriptions from present-day variation.

30. The general classifier and grammatical economy

Why might a general classifier spread?

One possible explanation involves grammatical economy.

A highly specialized system requires distinctions among many categories.

A more general marker can serve across a wider range of contexts.

But we should not assume that simplification is the only cause.

Language change is influenced by frequency, analogy, contact, community practices and the reanalysis of grammatical structures.

The 2024 research identifies several partially independent developments.

The spread of p’éel is therefore not adequately explained by saying that speakers are simply abandoning a complicated system.

The actual historical process is more nuanced.

31. The influence of Spanish

Yucatec Maya has long existed in contact with Spanish.

This contact has affected vocabulary and aspects of grammatical usage.

Research on contemporary numeral expressions notes that Spanish numerals are widely used for numbers above four in some varieties and contexts.

This creates an especially interesting situation.

The numeral may come from Spanish.

But the surrounding counting construction may still reflect Yucatec Maya grammatical patterns.

Language contact does not necessarily mean replacing one complete grammatical system with another.

Speakers can combine elements from different historical sources within evolving constructions.

32. A borrowed numeral does not automatically replace the counting system

Suppose a speaker uses a Spanish-derived numeral.

That does not necessarily mean the entire noun phrase becomes Spanish.

The numeral can participate in a construction shaped by Yucatec Maya grammar.

This is important because bilingual language contact often produces mixed structures.

A borrowed word may enter an existing grammatical framework.

Or the framework itself may change.

The result depends on the language, the construction and the community.

We should therefore avoid simplistic statements such as:

Spanish numerals destroyed the Maya counting system.

The evidence points to more complex patterns of contact and change.

33. The difference between counting and mathematical ability

A language may have a particular system of numeral classifiers.

That does not determine the mathematical intelligence of its speakers.

Nor does the historical use of Spanish numerals for certain quantities imply an inability to understand larger numbers.

Mathematical reasoning and the lexical resources used to express numbers are related but distinct.

People can learn new numeral systems.

They can calculate using written notation.

They can reason about quantities beyond the expressions most common in everyday speech.

This distinction is especially important when discussing Indigenous languages.

Linguistic structure must never be confused with a measure of intellectual capacity.

34. How this differs from our study of Pirahã

In our earlier investigation, If a Language Has No Exact Word for Seven, Can You Still Think About Seven? What Pirahã Reveals About Number, we examined the relationship between exact numeral vocabulary and numerical cognition.

Yucatec Maya presents a different problem.

The issue is not whether the language has an exact word for a particular number.

The issue is how a number combines with a grammatical expression identifying the counted unit or category.

Pirahã directs attention toward exact number words and numerical tasks.

Yucatec Maya directs attention toward individuation and classification.

These are separate mechanisms.

They should not be collapsed into one general claim about languages and mathematics.

35. The famous box experiment

One of the best-known research programs on Yucatec Maya and cognition was developed by John A. Lucy.

In a classic type of task, participants are shown a target object.

For example:

A cardboard box.

They then see two comparison items.

One matches the target’s shape or object kind.

The other matches its material.

For example:

A plastic box.

And a piece of cardboard.

The participant is asked which comparison item is more similar to the target.

The question is simple.

But it reveals different possible bases for similarity judgments.

36. Shape versus material

Imagine the target:

A cardboard box.

Comparison A:

A plastic box.

Comparison B:

A piece of cardboard.

Which is more similar?

If you choose the plastic box, you prioritize shape or object kind.

If you choose the cardboard, you prioritize material.

Neither answer is automatically irrational.

The objects share different properties.

The experiment examines which relationship participants prioritize under the task conditions.

Lucy’s research found cross-linguistic differences between groups of English and Yucatec Maya speakers in such judgments.

But interpreting those differences requires care.

37. What the experiments actually found

In the original research, English-speaking adults tended to prefer matches based on shape or object kind.

Yucatec Maya-speaking adults showed a greater tendency to select material-based matches, or a more balanced distribution between shape and material choices.

Subsequent studies investigated the robustness and developmental emergence of these patterns.

The findings became influential in debates about linguistic relativity.

But the results do not mean that English speakers cannot recognize materials.

Nor do they mean that Yucatec Maya speakers cannot recognize shapes.

The task measures preferences under specific experimental conditions.

It does not reveal an absolute limit on perception.

38. Why experimental context matters

Consider the instruction:

Which one is more similar?

Similar in what respect?

Shape?

Material?

Function?

Colour?

Use?

Origin?

The question is underdetermined.

Participants must decide which dimension is relevant.

Their choices may be influenced by language, culture, task interpretation, familiarity and other factors.

Therefore, a difference in similarity judgments is not automatically proof that one language forces speakers to perceive the world in a particular way.

It is evidence of a pattern that requires explanation.

The distinction between an observed experimental result and a theoretical interpretation is essential.

39. The developmental question

Research by Lucy and Suzanne Gaskins investigated how these cross-linguistic patterns emerge in childhood.

The reported differences become especially relevant during later childhood, rather than necessarily appearing fully formed in the earliest stages of development.

This raises an important question.

Do children acquire a preference through language?

Through culturally patterned activities?

Through schooling?

Through interaction among several influences?

The research contributes evidence to this debate.

But it does not justify a simple formula such as:

Language determines perception.

Developmental findings require a more careful account.

40. The limits of linguistic relativity

Linguistic relativity concerns relationships between language and cognition.

A strong claim would be:

A language determines what its speakers can perceive or think.

That is not supported by the Yucatec Maya evidence.

A more defensible research question is:

Can habitual linguistic distinctions influence which properties speakers attend to or prioritize in particular tasks?

That question can be investigated empirically.

The answer may depend on the task, population and linguistic structure.

The important point is that grammar can be relevant to cognition without acting as an absolute prison for thought.

41. The debate over Yucatec Maya noun meanings

John Lucy’s analysis has been influential in discussions of whether Yucatec Maya nouns are less inherently individuated than English count nouns.

However, this interpretation has been challenged.

Christian Lehmann and other researchers have questioned whether the famous banana examples support such a broad theory of noun meaning.

One criticism is that the examples often involve genuinely different referents.

A fruit.

A leaf.

A plant.

A load.

A portion.

The classifier does not necessarily produce five alternative conceptualizations of one identical object.

This debate is scientifically important.

It shows why grammatical examples must be interpreted carefully.

42. Why not every classifier can combine with every noun

Imagine trying to use a classifier for a banana leaf with the noun meaning person.

That does not automatically produce a meaningful expression for a flat person.

Grammatical systems have restrictions.

Some classifier-noun combinations are conventional.

Others are unacceptable or require special contexts.

The banana examples cannot simply be generalized to every noun.

This is a crucial limitation.

The presence of several classifiers in one lexical domain does not mean every noun is semantically unrestricted.

A language can have a rich classifier system and still impose strong lexical constraints.

43. The difference between lexical polysemy and grammatical classification

A word can have several related meanings.

This is called polysemy.

English paper can mean:

The material.

A newspaper.

An academic article.

An examination.

A document.

Now consider:

Three papers.

Depending on context, we might count three documents, three academic articles or three examinations.

The noun itself has multiple related uses.

A classifier system introduces another grammatical dimension.

But we must distinguish between the noun’s lexical meanings and the classifier’s contribution.

Otherwise, we risk attributing every semantic difference to the classifier alone.

44. Why the banana examples remain valuable

The banana examples are still useful.

They demonstrate that a numeral expression can identify different countable entities associated with the same lexical domain.

They show how classifiers contribute to interpreting a counting construction.

They also provide an opportunity to distinguish grammatical classification from lexical meaning.

But they should not be treated as evidence that Yucatec Maya speakers see all physical objects as undifferentiated substance.

That would go beyond what the examples establish.

The stronger interpretation is not necessary.

The actual grammatical mechanism is already interesting.

45. The difference between object and substance is not universal in grammar

English distinguishes count and mass nouns in particular ways.

Other languages distribute related distinctions differently.

Some use numeral classifiers extensively.

Some allow numerals to combine directly with many nouns.

Some have different systems of number marking.

Some use measure expressions in contexts where English does not.

This does not mean speakers disagree about whether water is liquid or a tree is solid.

It means the grammar organizes countability and individuation differently.

The physical world provides possibilities.

The linguistic system selects particular distinctions.

46. Counting depends on a decision about boundaries

Imagine a large rock.

You count:

One rock.

Now break it into five pieces.

You count:

Five pieces of rock.

The physical material has not multiplied.

But the number of units has changed.

This reveals an important distinction.

Quantity of material and number of individuated units are not the same thing.

The relationship between them depends on how the material is divided.

This is not merely a linguistic issue.

It is also relevant to measurement, statistics and mathematical modelling.

47. Why measurement and counting are different operations

Counting asks:

How many units are there?

Measurement asks:

How much of a quantity is present?

Consider water.

We can measure:

Three litres.

Five kilograms.

Two cups.

Or count:

Three bottles.

Five containers.

Two glasses.

These expressions are related.

But they are not interchangeable.

A bottle may contain different quantities.

A litre is a standardized volume.

A glass is a container whose capacity varies.

Yucatec Maya mensural constructions make this relationship between quantity and unit especially relevant to grammatical analysis.

48. Why this matters in science

Scientific measurement depends on clearly defined units.

Imagine a researcher reporting:

We collected five samples.

What counts as one sample?

A container?

A portion of tissue?

A measurement?

A specimen?

If the unit is unclear, the number alone is insufficient.

Similarly:

Ten observations.

What qualifies as an observation?

One event?

One participant?

One measurement?

One recorded response?

The number cannot be interpreted correctly without the unit.

Yucatec Maya provides a linguistic illustration of a problem that exists throughout scientific reasoning.

49. Why this matters in statistics

Imagine a report stating:

The number of cases increased by 20%.

What is a case?

One person?

One incident?

One hospital admission?

One diagnosis?

One report?

The percentage may be mathematically correct.

But its interpretation depends on the definition of the counted unit.

A change in the definition can change the apparent result.

This is why counting is never entirely independent of classification.

Before comparing numbers, we must know whether the same kinds of units were counted.

50. Why this matters in business

Consider:

We served 100 customers.

Does that mean:

100 different people?

100 transactions?

100 orders?

100 appointments?

One person may make several purchases.

One order may contain several products.

One appointment may involve several participants.

The number is meaningful only when the unit is clear.

The same problem appears in website analytics.

Visitors.

Sessions.

Page views.

Clicks.

Conversions.

These are not interchangeable units.

Yucatec Maya offers a useful grammatical perspective on the broader problem of defining what exactly is being counted.

51. Why this matters in education

A student may know how to calculate:

3 × 4 = 12.

But understanding a word problem requires identifying the unit.

Three boxes.

Four apples in each box.

Twelve apples.

Now change the question.

Three boxes.

Four kilograms of apples in each box.

Twelve kilograms.

The numerical operation is similar.

The resulting unit is different.

A learner who memorizes arithmetic operations without understanding the units can produce mathematically plausible but conceptually incorrect answers.

This is why counting and measurement should be taught through meaningful situations.

52. The grammar of units and the teaching of English

English learners often struggle with expressions such as:

a piece of advice

a piece of information

a loaf of bread

a slice of cheese

a bottle of water

These are not arbitrary vocabulary decorations.

They help establish countable units.

The distinction between:

three breads

and:

three loaves of bread

may depend on context and intended meaning.

Similarly:

two coffees

can mean two servings of coffee.

The learner must understand how English converts substances and other mass expressions into countable units.

Yucatec Maya offers a comparative framework for understanding why such constructions exist.

53. Why translation is not just replacing the classifier

Suppose a Yucatec Maya numeral expression includes a classifier identifying a particular unit.

The English translation may require:

A noun.

A measure word.

A portion expression.

Or no separate classifier equivalent at all.

For example, an expression involving a plant classifier may translate naturally as:

One tree.

An expression involving a small-amount classifier may require:

A little water.

The grammatical structure differs.

A translator must preserve the intended unit rather than mechanically reproduce every morpheme.

This is a problem of semantic reconstruction.

The question is not merely:

What does this classifier mean?

It is:

What exactly is being counted or measured in this context?

54. The danger of translating everything as “piece”

English piece is a flexible word.

A piece of cake.

A piece of wood.

A piece of information.

A piece of music.

But these uses do not all involve the same physical relationship.

A piece of cake is a portion.

A piece of wood may be a fragment.

A piece of information is abstract.

A piece of music is a composition or work.

Therefore, translating several Yucatec Maya classifiers with piece may conceal important distinctions.

The translator must consider whether the original construction identifies a physical object, a portion, a load, a plant or another type of unit.

55. The relationship between Yucatec Maya and Navajo

In our previous article, You Gave Me What? — Why Navajo Makes the Shape of an Object Part of the Verb, we examined how the physical properties of an object can influence the choice of a verb stem.

Yucatec Maya presents a different mechanism.

In Navajo, the classification can be integrated into the verbal representation of handling or movement.

In Yucatec Maya, classifiers participate in numeral expressions and the identification of countable units.

The two systems may both involve physical properties.

But they operate in different grammatical domains.

Navajo asks how the object participates in an action.

Yucatec Maya asks how the relevant entity is classified or individuated for counting.

This distinction is central to our comparative research.

56. Why the same physical property can matter in different grammatical places

Imagine a long flexible object.

In one language, its properties may influence the verb used to describe carrying it.

In another, a classifier may be relevant when counting it.

In a third, its properties may be expressed through an adjective.

In a fourth, the distinction may remain implicit.

The physical property is the same.

But the grammatical location differs.

This is one of the most important insights in cross-linguistic comparison.

Languages do not merely differ in which concepts they can express.

They also differ in where those concepts appear in grammatical structure.

57. What does a classifier actually classify?

We can now return to the central theoretical question.

Does a classifier classify the noun?

The physical object?

The unit?

The intended interpretation?

The answer depends on the classifier and construction.

Some classifiers are associated with relatively stable categories such as animacy.

Others identify shapes or physical configurations.

Others establish portions or measures.

And some have developed more general grammatical functions.

Therefore, the system cannot be reduced to one simple statement.

A classifier may contribute to identifying the relevant entity or unit.

But the precise contribution must be established through linguistic analysis.

58. Why grammatical categories are not scientific taxonomies

Scientific classification aims to organize entities according to explicit criteria.

Linguistic classification often reflects conventional patterns of use.

A grammatical category may combine historical, semantic and structural properties.

Its boundaries may not correspond to those of biology, geometry or physics.

For example, a classifier associated with plants is not necessarily a complete botanical taxonomy.

A classifier associated with elongated objects does not necessarily define a precise length-to-width ratio.

A general classifier may spread beyond its original category.

The grammar is a system of linguistic conventions.

It is not a scientific instrument for measuring every property of reality.

59. Can the classifier system change how speakers think?

This is the question that makes Yucatec Maya especially relevant to cognitive linguistics.

A speaker who regularly uses numeral classifiers may become accustomed to selecting particular categories when counting.

That could influence attention in certain tasks.

But the effect cannot simply be assumed.

Researchers must examine actual behaviour.

They must control for alternative explanations.

They must distinguish linguistic production from nonlinguistic cognition.

And they must avoid treating every observed difference as evidence of deterministic language effects.

The research on Yucatec Maya and object similarity is valuable precisely because it attempts to investigate these questions empirically.

60. What the evidence does not prove

The evidence does not prove that Yucatec Maya speakers cannot recognize objects without classifiers.

It does not prove that English speakers cannot think about substances.

It does not prove that one language is mathematically superior.

It does not prove that grammatical categories determine all perceptual judgments.

And it does not prove that every Yucatec Maya noun is equally compatible with every classifier.

These claims would go beyond the available evidence.

The more defensible conclusion is that Yucatec Maya has a rich and historically changing system of numeral classification, and that research has found interesting cross-linguistic patterns in certain object-similarity tasks.

That is already a substantial scientific result.

61. Why language is not a prison for thought

Imagine an English speaker learning Yucatec Maya.

They can learn to use numeral classifiers.

They can understand the difference between a plant, a leaf and a portion.

Now imagine a Yucatec Maya speaker learning English.

They can learn English count and mass constructions.

They can understand when English requires a measure expression.

The languages organize counting differently.

But speakers can learn new grammatical systems.

This is why linguistic diversity should be understood as diversity in conventional organization, not as a collection of isolated mental worlds.

62. The real cognitive question

The most productive question is not:

Does Yucatec Maya force its speakers to think differently?

It is:

How does repeated use of a particular grammatical system interact with attention, categorization and communicative practice?

That question is open to empirical investigation.

It can produce nuanced results.

And it does not require exaggerated claims about the power of language.

A language can influence habitual patterns of expression without determining every thought.

It can make certain distinctions grammatically salient without making other distinctions cognitively inaccessible.

63. Counting is a relationship between number and interpretation

Consider the expression:

Three.

By itself, it identifies a numerical quantity.

But what is being counted?

Three people?

Three apples?

Three kilograms?

Three pieces?

Three groups?

The interpretation depends on the unit.

This is true in English.

It is true in Yucatec Maya.

It is true in mathematics.

It is true in everyday reasoning.

The difference is that Yucatec Maya makes particular classification and unit-selection relationships grammatically visible through numeral-classifier constructions.

The language offers a concrete demonstration of a general principle.

Numbers do not interpret themselves.

64. Why this belongs in a series about language as a way of thinking

Our research series examines how languages organize different aspects of experience.

In Navajo, we examined how object properties participate in the selection of verbal stems.

In Matsés, we examined how the source and timing of information can be integrated into grammar.

In Kalaallisut, we examined how complex grammatical meanings can be organized within a single word.

Yucatec Maya adds a new dimension.

It asks how a language identifies the units to which numbers apply.

This is not simply another article about vocabulary.

It is an investigation of the relationship between linguistic structure and one of the most fundamental operations of human reasoning:

Counting.

Conclusion: before counting, we must decide what counts as one

We began with a banana.

One fruit.

One leaf.

One plant.

One bunch.

One portion.

The number remained one.

But the unit changed.

This simple observation reveals a deeper problem.

Counting is not merely the assignment of numbers to reality.

It requires the identification of countable units.

English often expresses these distinctions through nouns, measure expressions and context.

Yucatec Maya has a numeral-classifier system that makes certain classifications and units part of the grammatical structure of counting expressions.

The system distinguishes animate and inanimate entities, plants, shapes, portions and other categories.

It also changes over time.

Contemporary research documents variation in the use of specific classifiers and the spread of the general classifier p’éel.

The language therefore does not provide a frozen list of categories.

It provides a living grammatical system.

And the cognitive research associated with Yucatec Maya raises important questions about how language may interact with attention to shape, material and object identity.

But the evidence must be interpreted carefully.

A grammatical distinction is not proof of a different physical world.

A similarity judgment is not a complete theory of perception.

And a classifier does not automatically determine what speakers can think.

The most important conclusion is more precise.

A number tells us how many units there are. Language helps determine what those units are.

Yucatec Maya makes that relationship especially visible.

And once we recognize it, we begin to notice the same problem everywhere.

In mathematics.

In science.

In statistics.

In translation.

In education.

In everyday decisions.

Before asking how many, we must ask:

How many of what?

The difficulty is not always finding the correct number. Sometimes it is deciding what deserves to be counted as one.

— Tymur Levitin

Linguistic sources and further reading

1. John A. Lucy (1992). Grammatical Categories and Cognition: A Case Study of the Linguistic Relativity Hypothesis. Cambridge University Press. A foundational study comparing grammatical number and nominal classification in English and Yucatec Maya.
2. John A. Lucy (1992). Language Diversity and Thought: A Reformulation of the Linguistic Relativity Hypothesis. Cambridge University Press. A broader theoretical account of linguistic relativity.
3. John A. Lucy and Suzanne Gaskins (2001). Grammatical Categories and the Development of Classification Preferences: A Comparative Approach. In Language Acquisition and Conceptual Development, edited by Melissa Bowerman and Stephen C. Levinson. Cambridge University Press. Research on developmental differences in classification preferences.
4. John A. Lucy and Suzanne Gaskins (2003). Interaction of Language Type and Referent Type in the Development of Nonverbal Classification Preferences. In Language in Mind: Advances in the Study of Language and Thought, edited by Dedre Gentner and Susan Goldin-Meadow. MIT Press.
5. Barbara Blaha Pfeiler and Stavros Skopeteas (2024). Numeral Classifiers in Yucatec Maya: Microvariation and Syntactic Change. Journal of Historical Syntax, 8(6). Read the research article. A recent investigation of classifier variation, generalization and grammatical change.
6. Christian Lehmann. Grammaticalization in Yucatec Maya. Read the study. A detailed analysis of the historical development of grammatical categories, including numeral classifiers.
7. Christian Lehmann. Numeral Classification in Yucatec Maya. Read the research discussion. A critical examination of numeral classification and theoretical claims about noun meaning.
8. Barbara Pfeiler. Acquisition of Number in Yucatec Maya. Read the study. Research on the acquisition of numeral classifiers and number expressions.
9. Susan J. Hespos and Elizabeth S. Spelke (2004). Conceptual Precursors to Language. Nature, 430, 453–456. Research on conceptual distinctions in infancy relevant to the relationship between language and cognition.
10. Barbara C. Malt, Steven A. Sloman and Silvia P. Gennari (2003). Universality and Language Specificity in Object Naming. Journal of Memory and Language, 49(1), 20–42. Research on the relationship between object categorization and lexical naming.
11. Digital Maya — Yucatec Maya Grammar Lessons. Numerators and Classifiers. Educational examples of numeral-classifier constructions.
12. A Historical Grammar of the Maya Language of Yucatan: 1557–2000. Chapter: Numbers and Numeral Classifiers. Historical analysis of the Yucatec Maya counting system.

About the author and individual language learning

Tymur Levitin is the founder and director of Levitin Language School, an international online school providing individual language education.

His educational approach emphasizes understanding how linguistic structures work, recognizing distinctions in meaning and learning to use language in real communication rather than memorizing isolated equivalents.

This article belongs to the Language as a Way of Thinking research series, which explores how languages organize objects, actions, number, perception, evidence, time and grammatical meaning.

The series is a comparative linguistic research project. It does not imply that every language discussed is offered as a course at the school.

For international individual online language learning, visit Language Learnings.

For professional translation, editing, proofreading, copywriting and multilingual adaptation, explore Professional Language Services.

For individual online lessons or enquiries, contact Tymur Levitin via Telegram.

Tymur Levitin
Founder & Director, Levitin Language School
Language Teacher, Translator and Author

© Tymur Levitin / Levitin Language School. All rights reserved.