When Objects Become Too Prominent to Stay Unmarked
Grammar does not only tell us who did what to whom. Sometimes it also tells us what kind of “whom” we are dealing with.
Category: Language as a Way of Thinking
Consider two Spanish sentences:
Veo la casa.
I see the house.
Veo a María.
I see María.
At the most basic syntactic level, the structure seems identical.
Someone sees something — or someone.
The house is the direct object.
María is the direct object.
Yet Spanish treats them differently.
The house appears without a marker.
María gets a.
Why?
A learner may receive a perfectly useful classroom rule:
Use the personal a before people.
That rule helps us produce many correct sentences.
But it leaves the more interesting question unanswered:
Why should a language care whether its direct object is a person?
And Spanish is not an isolated curiosity.
Across the world’s languages, objects can be treated differently depending on properties such as animacy, humanness, definiteness, specificity and discourse prominence.
The broader phenomenon is known as:
Differential Object Marking.
And behind that technical term lies a remarkable idea:
Not all objects are equally object-like.
A stone is a very comfortable object
Imagine:
The woman lifted the stone.
There is nothing conceptually surprising about the distribution of roles.
We have:
HUMAN AGENT → ACTION → INANIMATE OBJECT
The woman can act intentionally.
The stone does not need intentions.
The woman manipulates.
The stone is manipulated.
This is close to what we might call a prototypical transitive event.
Now change only one participant:
The woman saw the man.
Grammatically, the man can still be the object.
But conceptually something has changed.
The object is now:
human,
animate,
individuated,
capable of intention,
capable of acting,
and perfectly capable of becoming the subject of another sentence:
The man saw the woman.
The grammatical object suddenly possesses many properties normally associated with highly prominent participants.
Some languages care about that.
An object can be syntactically an object without being conceptually thing-like
This distinction is crucial.
Grammar has a category called object.
Reality does not.
A person does not become less human because a verb assigns them object status.
When I say:
I saw Peter,
Peter occupies a syntactic role.
He has not become a thing.
This creates an interesting tension.
At one level:
Peter = OBJECT
At another:
Peter = HUMAN, ANIMATE, INDIVIDUATED PARTICIPANT
Languages can respond to this tension in different ways.
One solution is to mark certain objects differently.
Spanish gives us a beautifully visible example
Spanish can say:
Veo la casa.
but:
Veo a María.
The little a is traditionally called the personal a.
It is easy to learn it as an extra preposition.
It becomes much more interesting when we ask what larger grammatical problem it participates in solving.
Spanish is distinguishing between different kinds of direct objects.
The contrast is not merely:
OBJECT
versus
NOT OBJECT.
It can also involve:
WHAT KIND OF OBJECT?
This phenomenon has a name: Differential Object Marking
Linguists use the term Differential Object Marking, often shortened to DOM, for systems in which some direct objects receive special marking while others do not.
The exact conditions differ enormously between languages.
That warning matters.
There is no universal rule saying:
human object = special marker.
Instead, languages may respond to interacting factors such as:
animacy,
humanness,
definiteness,
specificity,
referentiality,
topicality,
and other grammatical or discourse conditions.
Different languages draw the boundary differently.
But the recurrence of the phenomenon raises a deeper question.
Why should these properties matter at all?
Because some objects are more expected than others
Imagine a simplified transitive prototype:
highly agentive participant → acts upon → less agentive participant
For example:
The child broke the glass.
The subject is human and animate.
The object is inanimate.
Their semantic profiles are strongly different.
Now consider:
The child pushed the teacher.
Both participants are human.
Both are animate.
Both are individually identifiable.
Both could easily be agents in another event.
The grammatical roles remain distinguishable.
But the semantic contrast between the participants is smaller.
This is where differential marking becomes particularly interesting.
A language may give extra grammatical visibility to an object whose properties make it less typical as an object.
The unexpected participant may receive more marking
This gives us a useful general model.
If an object is highly typical as an object, less marking may be necessary.
If an object has properties strongly associated with prominent agents — especially humanness and animacy — special marking becomes more likely in many DOM systems.
We should not turn this into a universal mechanical law.
But as a conceptual model it explains something that the classroom rule does not.
Spanish does not simply have a mysterious preposition before people’s names.
It participates in a broader linguistic tendency:
Languages may overtly mark participants that do not fit the most expected profile of their grammatical role.
But animacy is not simply biology
Here we need another distinction.
In linguistics, animacy is not always identical to the biological division between living and non-living things.
Languages often behave as though there were something like a hierarchy:
speaker / addressee
humans
other animate beings
inanimate entities
But the exact hierarchy and its grammatical consequences vary.
And language can conceptualize entities in ways that biology does not predict.
A dog may be treated as a highly individuated participant.
A country may become an actor.
Death may come looking for someone.
History may judge us.
The market may punish investors.
A computer may refuse to cooperate.
None of this changes biology.
It changes the linguistic scene.
Grammar does not simply photograph reality
This is one of the most useful principles in understanding language.
Suppose two speakers are looking for a doctor.
One says something equivalent to:
I am looking for a particular doctor who treated me yesterday.
Another means:
I need some doctor — I do not yet know who.
In both situations, the potential referent is human.
But the discourse status is different.
One participant is identified or highly specific.
The other may be presented mainly as someone who can satisfy a condition.
That difference can matter in languages with differential marking.
So physical reality alone cannot explain grammar.
We also need to know:
How is the speaker constructing the referent?
Specificity changes the scene
Consider the conceptual difference between:
I am looking for John.
and:
I am looking for a translator.
Both John and the future translator are human.
But they do not occupy the same position in the speaker’s mental representation.
John is already individuated.
The translator may not yet have an identity at all.
The speaker needs:
someone satisfying the category TRANSLATOR.
This is why a rule based only on biological humanness can never capture every DOM system.
The grammar may respond not only to what the entity is, but to how strongly it has been established as a particular discourse participant.
Definiteness adds another dimension
Compare:
a person
and:
that person
or:
a dog
and:
my dog
Again, the physical category may remain unchanged.
What changes is referential status.
A definite or highly identifiable participant is easier to treat as an individual already present in the discourse model.
This is one reason definiteness and specificity repeatedly interact with animacy in discussions of differential object marking.
The important insight is not that all languages calculate the same formula.
They do not.
The insight is that grammatical marking can depend on several dimensions simultaneously.
We can now build a better model
Instead of memorizing:
PERSON → SPECIAL MARKER
we can imagine several interacting questions:
Is the object human?
Is it animate?
Is it definite?
Is it specific?
Is it already identifiable?
How prominent is it in the discourse?
How typical is it for the object role in this language and construction?
The answer is not produced by adding numerical points.
Language is not a spreadsheet.
But the questions reveal the architecture behind patterns that otherwise look arbitrary.
Romanian gives us another route
Romanian also has differential object marking, prominently involving pe.
Again, it would be a mistake to declare Spanish a and Romanian pe identical.
Their grammatical distributions and histories are their own.
What matters comparatively is that both languages make certain direct objects formally different from others.
That gives us an extraordinary comparative question:
Why should languages that already possess ways of identifying grammatical objects develop additional marking for particular kinds of objects?
The answer takes us beyond vocabulary and into the organization of participants.
Other languages draw the boundary elsewhere
Differential object marking is found far beyond Romance.
Systems described under DOM occur in languages from very different families and regions.
The markers are different.
The conditions are different.
The grammatical histories are different.
But recurrent variables such as animacy, definiteness and specificity suggest that human languages repeatedly encounter a similar communicative problem:
participants differ in prominence even when syntax assigns them the same broad role.
That is a much more interesting fact than memorizing a list of prepositions.
Does special marking simply prevent ambiguity?
Sometimes disambiguation is relevant.
If both participants are human and either could plausibly be the agent, extra marking can certainly help identify their grammatical roles.
But that cannot be the whole explanation.
Languages may use differential marking even where the sentence would not realistically be ambiguous.
So:
“The marker tells us which participant is the object”
is useful but incomplete.
DOM sits at an intersection of:
syntax,
semantics,
referential structure,
and often discourse.
That is precisely why a one-line school rule cannot explain the entire phenomenon.
Languages care about probability more than textbooks sometimes admit
Grammar is often taught as a set of absolute categories:
subject,
object,
person,
thing,
singular,
plural.
Real language is also shaped by expectations.
Some combinations are highly typical.
Others are less typical.
A human agent acting upon an inanimate object is cognitively unsurprising.
An inanimate entity acting upon a human is less expected.
A highly individuated human appearing as an object is grammatically possible but semantically prominent.
Languages can exploit such asymmetries.
This does not mean speakers consciously calculate statistical probabilities before every sentence.
The patterns are already conventionalized in the language.
But their organization can still reflect recurrent properties of events and participants.
This is where grammar and cognition meet without becoming mysticism
It is tempting to make an exaggerated claim:
Speakers of languages with DOM perceive people differently.
We do not need that claim.
Spanish speakers do not possess a special cognitive ability to distinguish humans from houses.
English speakers obviously understand the same difference.
The interesting question is subtler:
Which distinctions does a language routinely ask speakers to encode when constructing an utterance?
English can distinguish people from things.
Spanish can distinguish people from things.
Romanian can distinguish people from things.
But they do not necessarily place that distinction in the same grammatical location.
That difference is enough to be linguistically fascinating.
English often leaves this distinction elsewhere
Consider:
I see the house.
I see María.
English does not insert a special object marker before María.
Does that mean English has lost the conceptual difference between a house and María?
Obviously not.
It simply does not grammaticalize this particular distinction in the same way.
English relies heavily on other resources, including word order:
María sees Peter.
is not the same as:
Peter sees María.
Participant roles are recoverable without a Spanish-style personal a.
This gives us a fundamental principle of comparative linguistics:
Two languages can represent the same event without distributing its information across the same grammatical components.
We have already seen this with motion
In You Saw Him Leave. But How Did He Move?, we examined how languages can package Manner and Path differently when describing motion.
One language may place important information in the verb.
Another may place it outside the verb.
Neither language necessarily lacks the concept.
They build the event differently.
Differential object marking reveals another version of the same larger principle.
This time the question is not:
How did the participant move?
It is:
How is the participant itself classified inside the event?
The event may remain the same.
Its grammatical architecture changes.
Translation therefore cannot be marker substitution
Suppose we translate:
Veo a María.
as:
I see María.
The Spanish a has disappeared.
Have we failed?
No.
Translation does not require every grammatical marker to acquire a visible twin.
It requires us to reconstruct the relevant relation in the target language.
The Spanish sentence identifies María as the direct object using Spanish grammatical resources.
The English sentence does the same using English resources.
A translation can preserve a relationship while changing the architecture that expresses it.
But sometimes translation loses something worth noticing
Imagine a text deliberately exploiting alternation in object marking.
Now the choice itself may carry semantic or pragmatic information.
A translator cannot simply say:
English has no personal a, so nothing can be done.
The task changes.
We ask:
What distinction is being created here?
Then:
Can English recover it through definiteness, lexical choice, word order, an additional phrase, or context?
Sometimes yes.
Sometimes partially.
Sometimes the distinction remains less visible.
That is not translation failure.
It is a structural difference between languages.
The same problem appears in language teaching
A student learns:
Put a before a person.
Then comes a real sentence that does not behave exactly as expected.
The usual response is to add another rule.
Then an exception.
Then an exception to the exception.
Soon grammar becomes a warehouse of unrelated instructions.
There is another way.
Start with the mechanism.
Ask:
Why might this language distinguish this object from a prototypical object?
Now animacy, specificity and definiteness stop being random complications.
They become dimensions of one larger problem.
Formation, mechanism, choice
This gives us three different stages of learning.
1. FORMATION
What does the construction look like?
2. MECHANISM
What distinction is the language making?
3. CHOICE
Why does a speaker use this construction in this particular context?
These stages should not be confused.
A learner can form a sentence correctly without understanding the mechanism.
A learner can understand the mechanism but still need practice choosing the natural form quickly.
Knowing a rule and using a language are related skills.
They are not identical.
Now the strange sentence becomes logical
Return to:
Veo la casa.
Veo a María.
At first, the extra a looks like an arbitrary property of Spanish.
After we understand differential object marking, it becomes evidence of a much larger system.
The language is not merely identifying an object.
It is sensitive to properties of the participant occupying the object position.
The difference between María and the house is obvious to every human being.
What varies across languages is whether, where and how grammar chooses to encode that difference.
A grammatical object is not necessarily a conceptual object
This may be the deepest lesson.
The word object can mislead us because ordinary English uses object to mean a thing.
Grammar uses object as a structural relation.
These are not the same.
A child can be an object.
A professor can be an object.
A president can be an object.
A dog can be an object.
A country can be an object.
Love can even be an object in a grammatical construction.
Syntax assigns a role.
It does not erase the conceptual properties of the participant filling that role.
Differential marking exists precisely in the space where those two layers interact.
Grammar can keep two truths at once
In:
Veo a María,
two statements are simultaneously true.
María is:
the grammatical object
and
a highly prominent human participant.
Those facts do not contradict each other.
Spanish can encode both.
This is what makes the phenomenon so revealing.
Grammar is not merely a machine for connecting words.
It is a system for organizing relationships between conceptual participants.
And this takes us beyond Spanish
Once we notice this mechanism, a much larger field opens.
We can ask:
Why do some languages distinguish human and non-human objects?
Why does specificity matter?
Why can definiteness affect marking?
Why do some grammatical systems rank speech-act participants differently from third persons?
Why do certain languages reorganize verbal morphology when the expected hierarchy between participants is reversed?
At that point, we are no longer studying a curious Spanish preposition.
We are investigating something much bigger:
how languages rank participants inside an event.
And that question can lead us into grammatical systems that look radically different from English or Spanish.
Conclusion
A house and a person can occupy the same syntactic position.
But they are not the same kind of participant.
Some languages make that difference grammatically visible.
Others do not mark it in the same place.
That does not mean one group of speakers understands animacy and another does not.
It means languages distribute information differently.
Differential object marking shows us that grammar can be sensitive not only to:
who acts,
and
who is affected,
but also to:
what kind of participant is being placed into each role.
A tiny Spanish a therefore opens a surprisingly large window onto the architecture of human language.
Because sometimes grammar does not merely tell us that something is an object.
It tells us:
this is an object — but not the kind of object you should treat like a thing.
Tymur Levitin
Founder & Director, Levitin Language School
Certified Translator and Language Teacher
© Tymur Levitin / Levitin Language School. All rights reserved.
