Verbs: Modality

This is about modality caveats that are not discussed in the main docs about verbs and modality.

Lexical categories

Modality may be expressed in several lexical categories, not just modal verbs:

He must have seen us (modal auxiliary)

Perhaps he saw us (adverb)

It’s possible that he saw us (adjective)

There’s a possibility that he saw us (noun)

See also

Stance adverbs for modal adjuncts

Must vs. have to

Syntactic defectiveness and suppletion

Must is a central modal auxiliary. It is syntactically highly defective: it possesses only a primary form and lacks all secondary inflectional forms (infinitivals, past and present participles).

Have (in have to) is a lexical verb taking a to-infinitival complement.

Because must lacks these secondary forms and cannot co-occur with other modals, have to serves as its suppletive counterpart to fill the syntactic gaps.

Past tense:

I had to leave early.

I musted leave early.

Infinitival:

I expect to have to leave early.

I expect to must leave early.

Post-modal:

You will have to leave early.

You will must leave early.

Deontic modality

Subjective vs. objective obligation

When expressing deontic modality (obligation, permission, or duty), the choice generally hinges on the source of the authority.

Must (subjective/internal):
The obligation is imposed by the speaker (or the addressee in interrogatives). It carries strong rhetorical weight and personal investment.

You must read this book. (My strong recommendation.)

I must stop procrastinating. (I am compelling myself.)

Have to (objective/external):
The obligation originates from an external source—a rule, a law, physical circumstances, or a third party. The speaker is merely reporting the existence of the obligation.

You have to read this book for the exam. (the syllabus requires it.)

I have to wear a tie to work. (company policy, not my choice.)

Note

In contemporary English, “have to” is increasingly encroaching on the territory of must even for subjective obligation, making it the statistically dominant choice in speech.

Must retains its footing primarily in formal writing and information packaging where the author wishes to project authority.

Epistemic modality

Logical necessity

Both forms can express epistemic modality—drawing a necessary conclusion based on available evidence.

There is no answer; they must be asleep.

There is no answer; they have to be asleep.

In epistemic uses, the semantic difference is negligible.

However, must remains the overwhelmingly preferred choice for formal epistemic deduction. Have to is more common in informal contexts or when adding emphatic, spoken stress.

Scope of negation

The critical divergence

The most crucial difference between the two occurs under negation. While both forms express strong necessity in the positive, their negative counterparts diverge completely because the negation applies to different parts of the clause.

Form	Negation scope	Meaning	Example
must not	Internal (negates the situation)	Obligation not to act (prohibition)	You must not go. (It is necessary that you do not go.)
don’t have to	External (negates the modality)	Absence of obligation (exemption)	You don’t have to go. (It is not necessary that you go.)

Note

To express the absence of obligation using a modal auxiliary, English historically relied on needn’t”

You needn’t go

But don’t have to has largely supplanted it in modern usage.

Scalar strength

The modal’s scalar strength is relatively stable, while its semantic domain (epistemic, deontic, etc.) is determined by context.

Modal	General force
might	very weak
may	weak
could	weak–medium
should	medium
must	strong

A modal auxiliary retains its relative strength across modal domains, but the type of force it expresses depends on the context.

Modal domain	Scale (what varies in strength)	Weak force	Strong force
Epistemic (degree of certainty about truth)	Certainty / commitment to truth	may / might = possible	must = necessary conclusion
Deontic (obligation, permission)	Obligation / requirement	may = permitted; should = recommended	must = required
Dynamic (ability, tendency, circumstances)	Necessity / possibility of action or state	may / can = possible or able under conditions	must = necessary due to circumstances
Bouletic (desire, goals)	Strength of desire / goal requirement	might want to = weak preference or suggestion	must = required to satisfy a goal
Teleological (purpose, goal-oriented necessity)	Necessity for achieving an outcome	should = expected to achieve a goal	must = necessary to achieve the goal

Obligation scale

• should = advice / expectation (weakest)

• ought to = moral obligation

• have to = external necessity

• must = strongest modal obligation

Lexical verbs to carry modality

Modality is about the speaker’s attitude toward the truth of a proposition. It can be carried with a lexical verb.

It seems that … (semantically, seem carries epistemic modality)

Even though it doesn’t have the grammar of a modal auxiliary, “seem” does the exact same conceptual work.

When you write:

It seems that the political center has collapsed

You are injecting epistemic modality. You are stepping back from claiming it as absolute fact and framing it instead as a deduction based on appearances or evidence.

Use auxiliaries

Modal auxiliaries (must, can, should, will, might) are better whenever you need speed, raw rhetorical punch, rhythmic lightness, or direct moral conviction.

While lexical modals build academic authority and evidence-anchored precision, overusing them can make prose feel dry, clinical, and bloated.

Modal auxiliaries are your primary tool for ignition and energy.

Here is where modal auxiliaries strictly outperform lexical verbs.

The rhetorical climax

 Direct urgency, peak and verdict

When an argument reaches its peak, you do not want to hide behind abstract nouns or clinical verbs.
You want the raw, unmediated weight of the writer’s voice.

Prudence dictates that society initiate a immediate transition away from fossil fuels.” (lexical, over-lexicalized, weak, feels bureaucratic and emotionally dead)

We must stop (modal auxiliary , strong, monosyllabic, immediate, impossible to mistake.)

Use lexical verbs to build the case throughout the body of an argument; switch to modal auxiliaries at the climax to demand a verdict.

Avoid Sir Hamphrey

Sentence rhythm and compression

Lexical verbs often require extra prepositional phrases and Latinate nouns (“has the capacity to,” “creates an obligation for”).

What you are suggesting is that, within the framework of the guidelines about Open Government which you have laid down, we should adopt a more flexible posture

Modal auxiliaries are lightweight monosyllables that keep prose lean and punchy.

The mechanism can process…

“The mechanism possesses the capability to process…”

We should rethink…

Circumstances impose an obligation upon us to rethink..

It might fail It remains within the realm of possibility that it fails

Use lexical

Force intellectual clarity

Auxiliaries hide weak logic

We must reform the system (easy, but it masks an incomplete thought - who or what is forcing this requirement?)

Lexical verbs force you to complete the thought: To upgrade to a lexical verb, you are forced to name the agent or mechanics.

You cannot use a lexical modal without knowing why something is necessary or possible.

Economic decay compels reform

Moral consistency demands reform.

Fine-grained tone control

Auxiliaries are blunt instruments: You are largely stuck choosing between binary states (may vs. must, can vs. cannot).

Lexical verbs are a precision dial: You can calibrate your exact level of conviction or commitment:

The data hints… (slight suspicion)

“The data suggests… (moderate confidence)

The data indicates… (strong stance)

The data demonstrates… (absolute conviction)

Avoid skepticism trigger

Defensive Reading

Auxiliaries highlight the writer’s ego: When a reader encounters auxiliary modality hey immediately hear the writer’s opinion:

This policy will fail

We should change

This triggers natural cognitive resistance:

Says who?

That’s just your view

Lexical verbs bypass reader resistance: When a reader reads, their attention is directed away from the writer’s opinion and onto the evidence.

Historical precedent indicates failure

It feels less like being lectured and more like examining a map together.

Lower cognitive load

Auxiliaries leave holes

An unanchored auxiliary modal leaves a micro-gap in the reader’s mental map:

It must be done (who is doing it? under what authority?)

Lexical verbs complete the picture: Lexical modals establish cause and effect instantly

Regulations prohibit entry (allowing the reader to absorb the idea without stopping to reconstruct the missing context).

Auxiliary vs. Lexical vs. Adjunct

Use	Example	Effect
Modal auxiliary	The policy may fail.	Present a possibility/necessity as part of the situation. Compact and neutral.
Modal adjunct	The policy will probably fail.	Explicitly calibrate your confidence as a writer.
Lexical modal/evidential verb	The evidence suggests that the policy will fail.	Emphasise the reasoning, evidence, or source behind the conclusion.

Central modals

• can

• could

• may

• might

• shall

• should

• will

• would

• must

They take a plain form complement:

She can swim. He must leave.

They do not take to:

She can to swim.

They do not have ordinary finite/non-finite forms:

to can canning (meaning ability)

They do not take third-person singular -s:

He cans swim.

They form negatives and questions without do:

*He cannot swim.

• Can he swim?

Accepted constructions

Licenses bare infinitival

A modal licenses a bare infinitival clause.

The head of that clause can be:

Lexical verb

He may know the answer. (the lexical verb know is the head of the infinitival clause, so no be is needed.)

Copular be

He may be tired. (here the lexical predicate is not a verb but the adjective “tired”.)

English requires the copular verb “be” to relate the subject to the predicative complement.

Progressive auxiliary

He may be sleeping. (the progressive construction is be + gerund-participle, so be is obligatory.)

Passive auxiliary

He may be arrested. (the passive construction is: be + past participle, so again be is obligatory.)

Important

A modal does not select be. It selects a bare infinitival VP.

The presence of be is determined by the predicate type, not by the modal.

VP

Head / construction	After the modal	Example
Lexical verb	plain-form lexical verb	know, sleep, leave
Copular verb	be + predicative complement (AdjP / NP / PP)	be tired, be a doctor, be at home
Progressive	be + gerund-participle	be sleeping
Passive	be + past participle	be arrested
Perfect	have + past participle	have left
Perfect progressive	have been + gerund-participle	have been sleeping
Perfect passive	have been + past participle	have been arrested

Note

A preposition itself is not the head of the infinitival clause:

He may be at home. (the clause is headed by the copular verb be)

The PP (at home, in London) is its predicative complement. That’s why the row is “Copular verb”, not “Preposition”.

He may know. (the lexical verb is know)

He may be tired. (adjectives require the copular be)

He may be sleeping. (progressive requires auxiliary be.)

He may be arrested. (passive requires auxiliary be)

Accepted patterns

After a central modal auxiliary, the complement is always a non-finite VP. More precisely, a bare infinitival clause headed by a plain-form verb.

Construction	Form	Example
Plain infinitival clause	plain form verb	must leave
Progressive infinitival clause	be + gerund-participle	must be sleeping
Passive infinitival clause	be + past participle	must be arrested
Perfect infinitival clause	have + past participle	must have left
Perfect progressive infinitival clause	have + been + gerund-participle	must have been sleeping
Perfect passive infinitival clause	have + been + past participle	must have been arrested

Modal
  |
  +-- leave
  |
  +-- be sleeping
  |      |
  |      +-- progressive
  |
  +-- be arrested
  |      |
  |      +-- passive
  |
  +-- have left
         |
         +-- perfect
              |
              +-- have been sleeping
              |       (perfect + progressive)
              |
              +-- have been arrested
                      (perfect + passive)

See also

Expanded examples for each modal with these patterns (linguistic_richness)

Structure and syntactic properties

Structurally, all of these constructions share four fundamental syntactic properties.

They all operate as catenative chains where a finite head dictates the form of a non-finite complement.

Obligatory plain-form immediate complement

Regardless of how long the chain gets, the modal auxiliary (as the matrix head) always imposes one strict inflectional requirement: The verb immediately following it must be in the plain form (bare infinitive).

• In the plain infinitival, the lexical verb is in the plain form.

• In the perfect constructions, the perfect auxiliary have is in the plain form.

• In the progressive/passive constructions, the auxiliary be is in the plain form.

Rigid auxiliary hierarchy

Every multi-verb construction in the table obeys an invariant, non-negotiable structural hierarchy.

When auxiliaries are stacked, they strictly follow this order of embedding:

Modal → Perfect (have) → Progressive (be) → Passive (be) → Lexical Verb

You can omit steps in the hierarchy (e.g., jumping from Modal directly to Passive be), but you cannot invert the sequence.

The structural tree is fixed.

The modal is a the single finite head

In all of these constructions, only the first element, the modal, is finite.

It anchors the clause for primary tense and modality.

Every subsequent element in the chain (whether an auxiliary or the main lexical verb) is non-finite.

Important

Syntactically, the modal heads the main clause, and the rest of the chain forms an embedded non-finite complement clause.

Downward feature assignment

Each auxiliary in the chain dictates the inflectional form of the complement immediately below it.

• The modal dictates the plain form, “have” or “be”.

• If perfect “have” is used, it forces the next verb into a past-participle.

• If progressive be is used, it forces the next verb into a gerund-participle

• If passive “be” is used, it forces the next verb into a past-participle.

In every case, the final lexical verb rests at the bottom of the catenative chain, its morphological shape dictated entirely by the auxiliary immediately preceding it, while the modal governs the top.

General ability vs. single-occasion actualization

Positive modals fail for single past events

He could swim (fully natural, expresses a general past capacity / dispositional state over an extended time period).

He could solve the problem” (pragmatically off if actualized)

The first is natural: Could freely functions as the past-tense equivalent of dynamic “can” for ongoing, general skills.

But when referring to a bounded, single occurrence, positive declarative could conveys unactualized potential or a conditional

He could solve the problem [if he tried].

It fails to entail actual accomplishment.

To express successful actualization on a specific past occasion, English requires:

He was able to solve the problem.

He managed to solve the problem.

He solved the problem.

Negation and bounded perfective event

Negation converts an event to state

A single completed action requires bounded, perfective aspect—something core past modals cannot express on their own.

However, negation converts an event into a continuous state, the state of an action not occurring.
Because modal auxiliaries are naturally stative, negative polarity aligns with their inherent aspect without requiring periphrastic intervention.

Ability

A single attempt to unlock a jammed door:

He managed to open the door. (positive, bounded, perfective. The action reached its endpoint and was completed).

He could open the door. (fails, “could” is stative and cannot express the sudden, bounded achievement of opening the door).

Note

Could on the dynamic ability reading, a state of ability, whereas open the door describes a bounded achievement/accomplishment—an event with a natural endpoint

He couldn’t open the door. (negative, continuous state/stative, the state of “not opening” stretched continuously throughout the entire duration of his attempt. The inherently stative modal could perfectly aligns with this static condition).

Volition

A single negotiation over a contract yesterday:

The client agreed to sign the contract. (positive, bounded, perfective, a singular, completed act of consent).

The client would sign the contract yesterday. (fails. It incorrectly applies a stative modal to a discrete, completed event, forcing an unnatural habitual or conditional reading).

Note

Would does not normally report a single completed past event.

None of the meanings contributed by would is to assert the occurrence of an event. Instead, they all place the event in a non-actual or non-specific domain.

The client wouldn’t sign the contract yesterday. (negative, continuous state/stative. The refusal operates as a continuous state of unwillingness dominating that specific timeframe).

Permission

A single instance of security clearance at an event:

She was allowed to bypass the checkpoint. (positive, bounded, perfective.The specific, bounded event of being granted entry).

She could bypass the checkpoint just then. (fails. Again, the stative modal cannot encode the actualized crossing of the checkpoint).

She couldn’t bypass the checkpoint. (negative, continuous state/stative. The restriction was a continuous state spanning the entire event).

Important

Core past-tense modals fail to entail actualization for single, bounded past events in positive declaratives, but work freely under negation—is a systematic feature across the modal system.

Adding perfect

Adding the perfect auxiliary (have), bypass the aspectual clash between the stative modal and perfective actualization.

The client would sign the contract yesterday.

The client would have signed the contract

The perfect shifted the sentence from the realm of historical fact into counterfactuality.

Here is the mechanical breakdown of why the syntax suddenly works:

Absence of a bounded event

Remote + Past reality = Non event

The rule that positive modals fail for single past events only applies when you are trying to declare that an action actually happened and reached its endpoint.

With the perfect, the inherent semantic entailment is that the client did not sign it:

The client would have signed the contract (the event never materialized)

There is no completed, perfective action to report. The situation remained static: the contract stayed unsigned.

Therefore, the inherently stative modal (would) is perfectly equipped for the job.

Important

The inherent counterfactual is not because of the perfect, but the application of the modal with the perfect.

When you combine a remote modal with a perfect auxiliary, you are applying a marker of “unreality” to a fixed historical timeframe.

• The Preterite Modal (would): Expresses modal remoteness. It explicitly distances the proposition from factual reality, framing it as conditional or hypothetical.

• The Perfect (have signed): Anchors that hypothetical proposition in the past.

• The Result: A situation that was a potentiality in the past, but is marked as remote from past reality.

By definition, an event that is remote from past reality is a non-event (a counterfactual).

The Division of Labor

 Remoteness + Past Time

Just as we saw with

could have swum,

The preterite form “would” and the perfect “have” split the semantic workload:

• Would (preterite modal): It is stripped of past-time reference and devoted entirely to modal remoteness. It signals that the volition (willingness to sign) is hypothetical, usually contingent on an unstated conditional adjunct

…if we had offered a discount.

• Have signed (perfect): Operating as a secondary past tense, it anchors that hypothetical willingness in the past.

The Contrast Summary

The client agreed to sign. (preterite = past time, a factual, bounded, single-occasion actualization.

The client would sign. (preterite = past time. Aspectual clash. The stative modal cannot express the bounded completion of the signing)

The client would have signed. (preterite = remoteness + perfect = past time, no clash. A counterfactual statement regarding a non-event.)

Important

You can prove that the perfect auxiliary is not responsible for the counterfactuality by observing what happens when you pair the perfect with a present-tense modal.

Dual reading of “could”

Could itself is polysemous

The perfect does not change the modal force of could.

It contributes anteriority: the situation is located before the reference time.

The modal meaning comes from could and the context.

Without perfect:

• Dynamic ability: “He had/has the ability to solve it.”

• Conditional possibility: “He could solve it if…”

He could solve the problem.

With perfect:

• Dynamic ability/opportunity before now: “He had the ability/opportunity to solve it.”

• Epistemic possibility about the past: “It’s possible that he solved it.”

He could solve it if… (conditional possibility)

The perfect simply moves the proposition under the modal from solve the problem to have solved the problem.

That is, the modal scopes over a perfect VP:

could [solve the problem]

could [have solved the problem]

The ambiguity arises because could itself is polysemous:

• Dynamic: “was able to”

• Epistemic: “it is possible that”

With a perfect complement, the epistemic reading becomes more available because English very commonly uses:

may/might/could + have + past participle

to express present uncertainty about a past event:

He may have left.

She might have forgotten.

He could have solved it.

Note

Because “couldn’t” naturally expresses the absence of ability, and that reading applies equally well to both activities and accomplishments.

He could swim. (general ability)

He couldn’t swim. (general inability)

He could solve the problem. (less natural if you mean a particular successful event)

✓ He couldn’t solve the problem. (natural: he lacked the ability or circumstances to solve it)

Important

The perfect hasn’t converted could into an epistemic modal. Rather, it provides a past-oriented proposition that the modal can quantify over.

Perfect → anteriority.
Could → dynamic or epistemic possibility.
Context determines which reading is intended.

---

Aspectual pacing

Modals with progressive/perfect aspect.

By altering the internal infinitival structure, you manipulate the reader’s experience of time and duration without changing the overarching judgment.

Society must change. (plain, a static, holistic demand).

Society must be changing. (progressive, immerses the reader in the middle of a continuous, durative process. It slows the narrative pace, shifting the focus from an endpoint to the ongoing evolution of the event).

Society must have changed. (perfect, forces the reader to look backward from the present anchor. It builds a rhetorical distance, highly effective for historical critique or establishing logical deductions).

Passive infinitival

Like all forms of passive, it provides a thematic focus, fronting, etc, and directly impacts how the reader prioritizes elements in an argument. The passive infinitival variants are critical for managing the flow of given vs. new information.

The government should address this crisis

This crisis should be addressed (allows to front the patient, the crisis, into the subject position)

See also

Clause: Passive

Counterfactual and critical layering

Complex stacking and informational density

The deeper the stacking goes (e.g., Perfect Progressive, Perfect Passive), the more nuanced the alternative realities presented to the reader become.

The policy should have been being monitored.

This highly complex structure delivers massive informational density. It simultaneously gives the reader:

• An obligation (should)

• Places it in the unfulfilled past (have)

• Backgrounds the responsible party (been)

• Stretches the negligence over a period of time (being monitored).

Finiteness

Central modal auxiliaries are the only verbs in English that are exclusively finite.

They completely lack non-finite forms.It can feel counterintuitive to call them “finite” because they do not look like normal finite verbs—they never take the $3\text{rd}$-person singular -s (he cans is ungrammatical), and they do not use standard suffixes like -ed to form the past tense.

The anchor of the chain

The primary tense for the entire clause

Every independent declarative clause requires exactly one finite verb to anchor it in time and reality.

When you start a verb phrase with a modal, the modal immediately fulfills that requirement.

It absorbs the primary tense for the entire clause. Because English syntax dictates that only the highest verb in the chain takes primary tense, every verb that follows the modal is forced into a non-finite form (plain form, past participle, or gerund-participle).

She might (finite preterite) have (non-finite plain form) been (non-finite past participle) working (non-finite gerund-participle).

The modal sits at the absolute top of the hierarchy, dictating the tense and modality of the sentence, while the non-finite verbs below it handle the aspect and voice of the event.

Key points

Important

Language relies on an economy of information.

If a speaker knows that a single, bounded past event actually occurred, the grammar requires the most direct factual assertion available: the simple preterite:

The client signed the contract.

Important

A modal and a bare infinitive weld together into a complex single predicate.

The modal provides the tense and the epistemic/deontic lens, while the bare infinitive provides the core event, operating as a single, indivisible cognitive unit for the reader.

The modal is functionally incomplete: It carries tense and modality, but it has no independent semantic content of its own. It must govern an action.

The plain form verb on the other hand is stripped of tense, agreement, and the infinitival marker “to”. It cannot exist as a predicate on its own. It relies entirely on the modal above it to anchor it into reality.

Important

The context of a modal auxiliary environment is important. The same modal verb can be interpreted differently in different contexts.

Often the surrounding context, not the form alone, determines the interpretation.

At first glance, this may seem confusing or ambiguous, but it allows English to be more economical.

Important

Because central modals evolved from expressions of intrinsic capacity and core necessity, they present an action not as something negotiated with the outside world, but as an inherent truth.

Note

It’s interesting that epistemic and deontic meanings are not generally associated with different expressions, which suggests they are intuitively perceived as related:

You must be very tactful (epistemic or deontic)

Note

“Can” is frequently used for epistemic (possibility, things can happen…), deontic (permission, similar to may), and dynamic (can=has the ability), and has been for centuries.

May is less common in conversation, and sounds a bit more formal, and so it may sound a bit more polite

Important

Very often declarative clauses with deontic modals are used in attempts to influence what happens, rather than simply to make assertions:

They should try again

See also

Modality options (linguistic richness)

Speculation (linguistic richness)

Modality syntactic reference (linguistic richness)

Usage of “will” in contemporary English (grammar, with a note for “shall”)

Lexical verbs carrying modality (linguistic richness)

Stance adverbs for modal adjuncts (linguistic richness)

Modality (craft)

Modality and the modal system (grammar)