The Cambridge Grammar of the English Language (Page 251) has this section in Chapter 4 The clause: complements:

5 Predicatives and related elements

A predicative complement is oriented towards a predicand, normally S in intransitives, O in transitives. In both cases it may be classified as either depictive or resultative as in [1], where double underlining marks the predicand, single underlining the predicative.

enter image description here

The resultative PC typically occurs with verbs that denote a change of state. The PC denotes the state of the predicand argument at the end of the process. A depictive PC gives a property of the predicand argument at the time of the situation under consideration, without any such factor of change.

In [1'] below, the to-infinitivals in bold are all complements of the matrix verbs seemed, found, came, and helped.
Kim seemed to be uneasy. He found Kim to be intolerant.
Kim came to know the truth. He helped Kim to be happy.

In CGEL, these are called catenative complements. But the term catenative doesn't convey any syntactic function of the complements as does the term predicative.

Syntactically, the to-infinitivals in [1'] do the syntactic function of predicates. So I wonder why they are not also called predicative complements.


The book says both finite and non-finite clauses can be predicative complements:

  • The embedded interrogative clause can be a predicative complement (Page 977) as in:

(1) The main question is w̲h̲e̲t̲h̲e̲r̲ ̲w̲e̲ ̲h̲a̲v̲e̲ ̲s̲u̲f̲f̲i̲c̲i̲e̲n̲t̲ ̲e̲v̲i̲d̲e̲n̲c̲e̲ ̲t̲o̲ ̲s̲e̲c̲u̲r̲e̲ ̲a̲ ̲c̲o̲n̲v̲i̲c̲t̲i̲o̲n̲.

  • The mandative clause can be a predicative complement (page 999) as in:

(2) The main recommendation was t̲h̲a̲t̲ ̲a̲n̲ ̲o̲u̲t̲s̲i̲d̲e̲ ̲c̲o̲n̲s̲u̲l̲t̲a̲n̲t̲ ̲b̲e̲ ̲e̲n̲g̲a̲g̲e̲d̲.

  • The to-infinitival can be a predicative complement (Page 1207) as in:

(3) His goal is t̲o̲ ̲w̲i̲n̲ ̲a̲t̲ ̲a̲l̲l̲ ̲c̲o̲s̲t̲s̲.

Below (3), it does say "[H]ere we focus on catenative complements, arguing that they cannot be systematically analysed as objects or predicative complements," but note that the to-infinitival in (3) is also a catenative complement as shown in (3'):

(3') His goal is t̲o̲ ̲h̲e̲l̲p̲ ̲h̲e̲r̲ ̲(̲t̲o̲)̲ ̲w̲i̲n̲ ̲a̲t̲ ̲a̲l̲l̲ ̲c̲o̲s̲t̲s̲.

  • The gerund-participial can be a predicative complement (Pages 1251)

(4) I’d call that t̲a̲k̲i̲n̲g̲ ̲u̲n̲f̲a̲i̲r̲ ̲a̲d̲v̲a̲n̲t̲a̲g̲e̲ ̲o̲f̲ ̲a̲ ̲b̲e̲g̲i̲n̲n̲e̲r̲. [objective predicative comp]

Re (4), it says on the next page:

Finally, [(4)] is a further case of the complex-transitive construction, but this time the gerund-participial is in predicative function.

Again, however, note that the gerund-participial in (4) is also a catenative complement as shown in (4'):

(4') I’d call that r̲e̲c̲o̲m̲m̲e̲n̲d̲i̲n̲g̲ ̲t̲a̲k̲i̲n̲g̲ ̲u̲n̲f̲a̲i̲r̲ ̲a̲d̲v̲a̲n̲t̲a̲g̲e̲ ̲o̲f̲ ̲a̲ ̲b̲e̲g̲i̲n̲n̲e̲r̲.

  • Finally (Page 1255):

The construction most clearly distinct from the catenative is the reversible specifying construction, where the internal complement can be a to-infinitival, a bare infinitival, or a gerund-participial:

[20] i His intention wasf̲o̲r̲ ̲t̲h̲e̲ ̲m̲e̲e̲t̲i̲n̲g̲)̲ ̲t̲o̲ ̲b̲e̲g̲i̲n̲ ̲a̲t̲ ̲s̲i̲x̲. [to-infinitival]
ii All I did was p̲r̲i̲n̲t̲ ̲o̲u̲t̲ ̲t̲h̲e̲ ̲t̲a̲b̲l̲e̲ ̲o̲f̲ ̲c̲o̲n̲t̲e̲n̲t̲s̲. [bare infinitival]
iii The funniest thing wasK̲i̲m̲)̲ ̲t̲r̲y̲i̲n̲g̲ ̲t̲o̲ ̲h̲i̲d̲e̲ ̲i̲n̲ ̲t̲h̲e̲ ̲c̲o̲a̲l̲-̲b̲o̲x̲. [gerund-participial]

Here, the underlined gerund-participial in example [20iii] is already a catenative complement in that it has both trying and hide.


1 Answer 1


The value of distinguishing catenative from predicative complements comes from the classification of patterns that can be applied to clauses.

For example, make at the head of a clause makes possible both predicative complements and plain infinitival catenative complements, whereas become takes only predicative complements.

Predicative complements are either ascriptive or specifying - that is, they express an A=B relationship with the predicand and the predicate. This contrasts with catenative complements in which the idea of 'predication' is different, one in which there is a subject or object and a catenative complement clause which takes that as its implied subject, but is not a quality or the referent of that subject or object.

The plan was to get there early.

The plan = to get there early

To get there early is the referent of the plan.

He made a good chef.

He = a good chef

He is ascribed the quality of a good chef.

He made me get the mail.

me =/= get the mail

I am the one who got the mail, but it does not ascribe a quality or clarify a referent. Neither of the following would be possible substitutes:

He made me it.

He made me so.

What I did is get the mail, but that's not what make is expressing.

So, no, a predicative complement is not a catenative one, at least not according to the criteria presented by CGEL.

  • 1
    +1 for simplicity. I understand your answer unlike many others I've seen around here.
    – Lambie
    Commented Mar 27 at 14:48
  • By "the classification of patterns that can be applied to clauses", you seem to mean "the complementation patterns of the matrix verb", which shouldn't be the reason for distinguishing different types of complements. Even within the predicative complement (PC), become, for example, takes only ascriptive PC whereas be takes both acriptive and specifying PCs.
    – JK2
    Commented Mar 27 at 22:42
  • It's only the specifying PC that expresses an A=B relationship. She made him a good chef means not "he is the same as a good chef" but "he has the property of being a good chef." So, him =/= a good chef in the sense of the specifying PC. Therefore, even within the PC, "the idea of predication is different".
    – JK2
    Commented Mar 27 at 22:42
  • More importantly, since the terminology is "predicative complement", I don't know why only a certain idea of predication makes a complement a PC. In Kim seemed (to be) uneasy, both uneasy and to be uneasy express the "the idea of predication". If anything, the latter expresses a more complete idea of predication.
    – JK2
    Commented Mar 27 at 22:43
  • 1
    @JK2 Taking only one property of a predicative complement, as defined by CGEL, and claiming that everything that has that one property is a predicative complement isn't a productive position to take. If this question is simply a nit-picking of CGEL's terminology, ignoring the full criteria they've assigned to these terms, then we're engaging in an equivocation fallacy. My answer still stands: the difference between predicative complements and catenative complements, as defined by CGEL, is a clear and useful distinction.
    – DW256
    Commented Mar 28 at 4:42

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.