From NBC news’s ‘Breakthrough finding’ reveals why certain Covid-19 patients die:

"Before Covid, their condition was silent," Bastard said. "Most of them hadn't gotten sick before." Bastard said he now wonders whether autoantibodies against interferon also increase the risk from other viruses, such as influenza. Among patients in his study, "some of them had gotten flu in the past, and we're looking to see if the autoantibodies could have had an effect on flu."

I have two questions.

  1. Does "could have had" in "...we're looking to see if the autoantibodies could have had an effect on flu" simply imply that they (a group of researchers) were looking to see if the autoantibodies had had an effect on flu and does "could" with the perfect infinitive" just express ability or theoretical possibility as a modal construction in the Indicative Mood? And therefore it is not a Conditional clause but an Object clause?
  2. If so, where may the grammar rules (patterns) laying down non-conditional usage of "could" with the perfect infinitive be met? (because having sought out much on these issue I got stumped to find out that it is being widely propagated throughout Grammar books that using such constructions is narrowed almost only to the Subjunctive Mood with the Type 3 conditional)
  • I would have thought autoantibodies could have had an effect is an uncertain version of the pluperfect/past perfect autoantibodies had had an effect
    – Henry
    Nov 27, 2020 at 11:16

1 Answer 1


You’ve asked about the verbs used in your quoted paragraph’s last sentence.

Among patients in his study, “some of them had gotten flu in the past, and we’re looking to see if the autoantibodies could have had an effect on flu.”

There are no conditionals here at all, just a modal perfect past in keeping with the perfect past form of the had gotten verb used earlier in that same sentence.

The rest of this post explains why this has been done and how you can predict and generate similar clauses for yourself. The Addendum at the bottom illustrates why your choices in terminology are confusing and limiting you because they do not fairly model how English grammar really works.

Complement Clauses

What you have here is a case of the verb to see X, where X is the finite clause “if the autoantibodies could have had an effect on flu”.

Whenever the object of the transitive verb see is a wh- clause like this (sometimes called an “interrogative” clause), the meaning of the matrix verb per the Oxford English Dictionary is equivalent to:

To ascertain or establish by observation, inquiry, or reflection; to investigate.

Or more fully, letting the word something stand in for your X clause:

To ascertain [something] or establish [something] by observation, inquiry, or reflection; to investigate [something].

This wh- clause here is necessarily a complement clause because it serves as the transitive verb’s complement: its object. So sure, I reckon you can call it an “object clause” if you’d like. This sentence therefore means the same thing as the original:

Among patients in his study, “some of them had gotten flu in the past, and we intend to investigate whether the autoantibodies could have had an effect on flu.”

In fact, this version, which doesn’t use a formal modal at all, also means the very same thing as the original:

Among patients in his study, “some of them had gotten flu in the past, and we are planning on investigating whether the autoantibodies had been able to affect flu.”

But exactly why those are equivalent will require a bit of explanation, given below.

Modal Verbs: Dual Modalities

This complement clause’s modal verb could (which is just the past tense of can) is here being used in the epistemic mode of prediction, not in the deontic mode of obligation.

Instead of being a plain infinitive like to affect (which is all that the phrasal verb to have an effect on means), this instance is further marked with the perfective aspect by using a “perfect infinitive” as you call it by using to have with the perfect/past participle of that verb.

Here just like anywhere else they’re used, a modal verb adds nuance by altering the “modality” of the verb following it. Instead of a simple statement of plain fact, it becomes one of either prediction or obligation. It is either a predictive statement about the possible or probable world (the so-called “epistemic mode”), or else it is a statement of how the world ought to be or become (the so-called “deontic mode”) in keeping with the speaker’s or society’s desires, demands, or norms.

It’s not just the modal verb could that does this shading of reality away from a simple indicative sense. All modal verbs do this, because that is their purpose. However, there is a problem with modal verbs. Compared with normal verbs, the nine English modals (will/would, shall/should, can/could, may/might, and must) are all defective in multiple ways.

For one, they never inflect for person or number.

For another, they lack corresponding non-finite forms such as infinitives and participles.

And to some extent, they even lack inflections for the past tense because although you can backshift some of them for tense like can to could, this is more an obligatory shift than a deliberate one, and the modal must has no past-tense version. That means that it’s hard to convey tense in a modal, and that whenever we need some non-finite version of a modal like its infinitive or a participle, it’s the verb following that modal that takes on the non-finite form.

For example, here we’ll use may/might in the epistemic mode of simple prediction:

  • He thinks that she may be hungry. (plain present)
  • He thought that she might be hungry. (plain past)
  • He has thought that she may have been hungry. (perfect present)
  • He had thought that she might have been be hungry. (perfect past)

Periphrastic Modals

Or, if we want, sometimes we can do something else altogether.

We can find some sort of periphrastic version using normal verbs to which we can then apply normal verbal inflections. Because another rule about modals is that you can only have one per verb phrase, we also have to use a periphrastic rephrasing whenever we want to apply more than one modal at a time (at least in standard English).

That’s why the can in can affect gets rewritten into be able to when we need to talk about something in the future, leaving us with constructions like will be able to affect.

Unconditional Modal Perfects

I have explained all this, and done so in this fashion, because it allows me to answer your question about how to understand the verbal patterns that allow us to generate could and perfect infinitives in places other than in conditional clauses alone.

So now let’s get back to your complement clause and see just exactly what’s happening in it. Its subject is antibodies and the verb in general is to have an effect on, which is simply a phrasal-verb version of to affect.

  • to have an effect on X = to affect X (plain infinitive)
  • to have had an effect on X = to have affected X (perfect infinitive)

This leads to four ways of writing that verb. It can be in the present (or at least, the non-past) or the past, and it can be either plain or perfect.

  • autoantibodies have an effect on = autoantibodies affect (plain present)
  • autoantibodies had an effect on = autoantibodies affected (plain past)
  • autoantibodies have had an effect on = autoantibodies have affected (perfect present)
  • autoantibodies had had an effect on = autoantibodies had affected (perfect past)

But now we need to use the modal could here. This is what leads to the perfect infinitive, because modals always take infinitives:

  • autoantibodies can have an effect on = autoantibodies can affect (plain present)
  • autoantibodies could have an effect on = autoantibodies could affect (plain past)
  • autoantibodies can have had an effect on = autoantibodies can have affected (perfect present)
  • autoantibodies could have had an effect on = autoantibodies could have affected (perfect past)

Now let’s do one more rewrite for illustration. Instead of using modal can/could X, we’ll use the periphrastic modal to be able to X.

  • autoantibodies are able to have an effect = autoantibodies are able to affect (plain present)
  • autoantibodies were able to have an effect = autoantibodies were able to affect (plain past)
  • autoantibodies have been able to have an effect = autoantibodies have been able to affect (perfect present)
  • autoantibodies had been able to have an effect = autoantibodies had been able to affect (perfect past)

So in saying that “the autoantibodies could have had an effect on flu”, this means the very same thing as “the autoantibodies had been able to affect flu”.

So there is nothing at all “conditional” involved here, despite your clause using a modal perfect. Why would there be? It is simply a backshifted epistemic modal used to make a prediction about a possibility, and the reason it’s used with a perfect infinitive is so that it reflects the same narrative time as had gotten flu from the earlier part of the sentence, and had gotten is in the perfect past. Therefore this one needs to be in the perfect past as well.

If we simplify had gotten flu from the perfect past into the plain past version got flu (or the equivalent and more common had the flu), we can also simplify the remaining verb phrase to match:

Among patients in his study, “some of them had the flu in the past, so we’re looking to investigate whether those autoantibodies had been able to affect flu.”

Or even more simply:

Among patients in his study, “some of them got the flu in the past, so we’re investigating whether those earlier autoantibodies affected later bouts with the flu.”


There are no conditionals here at all, just modal past perfects to fit with the past perfect from earlier in that same sentence.

Addendum: Unconstructive Models

Your mention of subjunctive mood and type-three conditional strongly suggest to me that you’ve been taught that these are terms that can be meaningfully applied to English. This may be where you’ve been led astray.

Although under some linguistic models, such terms can sometimes be bent into English service, it is at best an unnatural and ill-fitting servitude bitterly indentured. Those terms do not really apply to English. Instead, you have to learn how we actually use modals here.

There is nothing about modal verbs in English that restricts their use to only subordinate or complement clauses instead of main clauses. All that happens when you use a modal is that it makes a verb something other than strictly real. But please do not call it “subjunctive”. Call it modal.

Just as soon as you pull in a modal, though, your verb stops being real and becomes “modal” instead. So you should contrast real verbs with modal verbs in English, not indicative with subjunctive. Once it’s modal it’s no longer in the real mode, but whether it’s the epistemic mode or the deontic mode strictly varies by the speaker’s intent. Both modes are always possible.

And we have a whole lot of different modal verbs, not just the nine standard ones and the three or four semi-modals that come quickly to mind, but also gobs and gobs of periphrastic modals as well like used to and has to and is to and is supposed to and is able to, and many more besides.

Let’s use the verb sing, and first talk about the reals. As a finite verb you might use in a main clause, you always have four variants for real (“non-modally marked”) constructions:

  • She sings. (plain present: real)
  • She sang. (plain past: real)
  • She has sung. (perfect present: real)
  • She had sung. (perfect past: real)

Just like any real verb, any modal verb also comes in those same four basic flavors: plain-vs-perfect (sometimes called simple-vs-perfect) and present/nonpast-vs-past.

  • She will sing. (plain present: modal)
  • She would sing. (plain past: modal)
  • She will have sung. (perfect present: modal)
  • She would have sung. (perfect past: modal)

You can do that same with any of the modals.

  • She can sing. (plain present: modal)
  • She could sing. (plain past: modal)
  • She can have sung. (perfect present: modal)
  • She could have sung. (perfect past: modal)

In the same way, they can all also be used not just as the main clause but in complement clauses or subordinate clauses. First the real versions as complement clauses:

  • I think she sings opera. (plain present: real)
  • I thought she sang opera. (plain past: real)
  • I have thought she has sung opera. (perfect present: real)
  • I had thought she had sung opera. (perfect past: real)

Then the corresponding modal versions as complement clauses:

  • I do not think she can sing opera. (plain present: modal)
  • I did not think she could sing opera. (plain past: modal)
  • I have not thought she can have sung opera. (perfect present: modal)
  • I had not thought she could have sung opera. (perfect past: modal)

There’s nothing special about complement clauses; you can do all those same things as subordinate clauses as well. First as reals:

  • She always sings opera if she wants to. (plain present: real)
  • She always sang opera if she wanted to. (plain past: real)
  • She has always sung opera if she has wanted to. (perfect present: real)
  • She had always sung opera if she had wanted to. (perfect past: real)

And then as modals:

  • She will sing opera if she wants. (plain present: modal)
  • She would sing opera if she wanted. (plain past: modal)
  • She will have sung opera if she has wanted. (perfect present: modal)
  • She would have sung opera if she had wanted. (perfect past: modal)

As you see, there is nothing about the use of real-vs-modal verbs in complement clauses or in subordinate clauses that is in any way special compared with how the same real-vs-modal verbs get used in main clauses. And all those same things remain true no matter whether these are simple uses or perfective uses in every case.

Related Curiosities

The “zero” modal + infinitive

Verbs marked with a modal are always unreals, not reals. This is true even for the “zero modal” case:

  • They demand that she leave. (real present followed by bare infinitive as an untensed unreal prefaced by the zero modal)
  • They demanded that she leave. (real past followed by bare infinitive as an untensed unreal prefaced by the zero modal)

As you see, the verb leave is used there in the bare infinitive, just as though it were preceded by an actual modal verb before it. It can be useful under this model of synchronic analysis to consider it an instance of a zero-modal plus an infinitive. This isn’t much of a marker beyond signaling that it is not real, that it is an unreal case. As modals go, the zero modal is a deontic one because it is making a statement about how the world ought to be.

Under diachronic analysis this derives from what in Old English had been the present subjunctive, but that analysis doesn’t make sense synchronically. It really is not marked for tense at all, since both they demand and they demanded take the same bare infinitive unmarked for tense when used with the hypothesized zero modal.

Sometimes this is called the “mandative subjunctive” these days, but really it is a zero-modal plus a bare infinitive. It can no longer be used in the main clause, only in subordinate clauses or complement clauses.

The ancient, formulaic blessings and cursings did once upon a time take this form in the main clause to create a sort of “third-person imperative” (God bless you, God save the queen, Devil take the hindmost) but these are no longer productive in present-day English. Nowadays we instead use an explicit auxiliary verb like may or let for these types if we ever need to generate new ones: May the Force be with you!

The only non-modal unreal

Although all modals are unreals, not quite all unreals are modals. There is one non-modal unreal. The verb be retains a specially marked non-modal unreal inflection: were, which is invariant by person and number. Although under an historical diachronic analysis, it derives from what had been a past subjunctive form in Old English, under synchronic analysis it is just a special unreal form. It’s used in just a very few places, all strongly hypothetical in nature. Here is an assortment of those:

  • Imagine it were so.
  • He acted as if she were here.
  • If only she were here.
  • I wish this were easier.
  • She would have left us a note if she were ready to leave.
  • Unless she were ready, she would leave us no note.

Again, these are all unreal hypotheticals that are unreal by virtue of the unique were inflection instead of via a modal verb plus an infinitive as you must do for all other verbs except for be alone.

That’s why people sometimes rephrase certain kinds of “if I said X...” clauses into the periphrastic inflected-be + to-infinitive version: “if I were to say X...” The reason we sometimes do this is because the second feels more hypothetical than the first does due to its overt inflectional “unreality” signal.

Summary of Addendum

Don’t get wrapped up in terms like subordinate mood or type-three conditionals, because these are not useful models for describing English grammar.

Instead, think about real verb constructions versus modal verb constructions, and about simple verb constructions versus perfective verb constructions.

And definitely do not think of any of these as restricted to niche uses. They are all found in main verb clauses, in subordinate verb clauses, and in complement clauses. The real-vs-modal choices and simple-vs-perfect choices you make in all three of those clause types are freely elected depending on the speaker’s intent.

Now factor in how there isn’t just one modal the way languages with a subjunctive inflection have just one of those to go with the real indicative, but rather there are nine modals and perhaps a half-dozen semi-modals and still more periphrastic modals on top of those two sets. Double all that by understanding every one of those many modals is dual-minded in that they can all equally express both epistemic and deontic modalities.

What this provides you with is hundreds and hundreds, perhaps even thousands, of combinations available to the native speaker to produce incredibly many subtle nuances, nuances which no simplistically dualistic indicative-vs-subjunctive feature contrast alone can ever hope to approach let alone match.

See Also

  • 2
    @Eugene You can try, but it might need a new question. English tenses and aspects combine in complex ways. For example ① all these can end in present …before she leaves: I’m warning her, I’m to warn her, I’ll warn her, I’m going to warn her, I’ll be warning her, I’ve been going to warn her, I’d warn her, I can warn her, I could warn her. But ② all these can end in past …before she left: I was to warn her, I was going to warn her, I’d been going to warn her, I’d warn her, I warned her, I’ve warned her, I could’ve warned her, I’d warned her, I’d’ve warned her, I’ll’ve warned her.
    – tchrist
    Nov 28, 2020 at 15:11
  • 1
    @Eugene Sure, non-conditional modal + perfect-infinitive constructions are the primary topic here. I only mentioned conditionals because your question title referenced “type three conditionals”.
    – tchrist
    Nov 28, 2020 at 15:55
  • 1
    @Eugene There is no implied if here; you must train yourself to stop thinking there’s a hidden one that hasn't been expressed. There is not. This could uses the epistemic mode of prediction not the deontic mode of permission or obligation, so your sentence means Maybe the house blew up before she left”, just as “It might have blown up before she left” also means. That is not the apodosis (the "then" part) of a conditional, nor is there some unexpressed protasis (the "if" part) for you to imagine. This is just a way to say "maybe something happened" in the past, nothing more.
    – tchrist
    Nov 29, 2020 at 21:35
  • 1
    @Eugene Try this: Whenever you come across a modal plus a “perfect infinitive” outside of a conditional structure, don't first look for some elided or implied protasis to supply the meaning. Just take it as it is. Let ❎ = “have left early” in all these non-conditional modal + perfect infinitive examples: He must/may/might ❎. He shall/should ❎. He ought to ❎. He didn’t have to ❎ He was supposed to ❎. He needn’t ❎. I think he may/might ❎. I thought he might ❎. I am sure he will/would ❎. I was sure he would ❎. I wonder whether he can/could ❎. I wondered whether he could ❎.
    – tchrist
    Dec 1, 2020 at 0:35
  • 1
    @Eugene Native speakers learn modals by hearing each used on its own. Imagine you’re asked what’s happened to the icicle that had hung above your doorstep, so you pick the modal perfect to match how positive you are, here ordered by ascending probability: “It could have melted. It might have melted. It may have melted. It must have melted. It will have melted.” Those epistemic modal perfects are all predictions about a possible past event where your confidence varies from a light “Perhaps it’s melted” at the start to a firm “Surely it’s melted” by the end.
    – tchrist
    Dec 1, 2020 at 15:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.