When you draw a tree diagram in phrase structure grammars, should you draw a complement of a verb any differently than an adjunct of a verb by using different tree structures?

(1) I slept on the floor. [adjunct]

(2) I relied on my mother. [complement]

For example, in these two sentences, "on the floor" is an adjunct of the verb "slept" whereas "on my mother" is a complement of the verb "relied".

Now, is the tree diagram of (1) any different than that of (2)?

  • No, though the function labels are of course different, i.e. 'modifier' and 'complement' respectively. – BillJ Sep 19 '17 at 9:36
  • 1
    @BillJ If you rely on a non-standard theory of phrase structure grammar, you are of course welcome to present your solution, but you should then make clear to a user that is obviously not yet deeply familiar with syntactic theory that this is not what the majority of linguists would agree on. An account of phrase structure grammar as usually understood, or as introduced in undergrad linguistics, which OP is most likely interested in, makes a structural difference between complements and adjuncts, and I'm sure that you're aware of that. – lemontree Sep 19 '17 at 10:57
  • Non-standard. Please! The adjunct and complement PPs are simply constituents within the predicate VP. But then I eschew X-Bar as quite rightly do Pullum and Huddleston – BillJ Sep 19 '17 at 11:53
  • 1
    I gave you the answer as I see things. Others may prefer X-Bar theory -- that's up to them. But the main distinction has not to do with trees, but the fact that unlike adjuncts, complements must be licensed by the head. Adjuncts are dependents in clause structure, while complements can be dependents of most heads. – BillJ Sep 19 '17 at 16:05
  • 1
    That's a question you'll have to put to the authors, though Geoff Pullum did once tell me that he dislikes X-Bar, so make what you will of that. CGEL is an attempt to do straightforward traditional descriptive grammar with a few helpful and intuitive trees thrown in where appropriate. It doesn't deal with arcane theoretical stuff, but builds on a huge amount of research to produce their award-winning grammar. CGEL's trees do have one striking advantage over X-Bar in that they distinguish between function and category, assigning two labels to each constituent. – BillJ Sep 19 '17 at 16:33

In mainstream phrase structure grammars that rely on some form of X-bar theory, yes: In fact, the differing tree structures are the crucial point where the difference comes into play (coming along with grammatical implications concerning lexical selection, government of morphological features etc.).
The distinction between complements and adjuncts behaves the same way for VPs as it does for NPs or other categories.

In a classical X-bar account as taught in basic linguistics classes:

An adjunct to a verb(al projection) is a sister to the V' it adjoins to and daughter to a new V' projection: enter image description here
I.e., the verb already forms a higher bar projection (V' instead of just V) by itself. At the level of V', the phrase is already somewhat "complete" in terms of grammaticality; what potentially misses now is a specifier, but VPs are generally not assumed to have specifiers, so at this point, we can already obtain a full verb phrase (which is visible by the V' going over directly into a VP). The adjunct acts merely as a modifier to the V' (in the sense of adding more information) and thereby produces a new V' projection; i.o.w., adjunction doesn't really change anything about the grammatical status of the verbal projection - we had a V' previously and we have a new one now; the adjunct might as well be left away and leave a healthy verb phrase.

The complement of a verb is a sister to the verbal head, and daughter to a first V':
enter image description here
I.e., a verb that requires a complmement forms the next projection level only together with its complement; without it, the sentence is ungrammatical. V does not make a V' by itself. It requires the complement as a direct sister in order to form a higher, more "complete" projection that can stand by itself. So the next level of phrasal completeness may only be added after combining the V head with its PP argument.

Other variants of phrase structure grammar will propose slightly different tree structures, e.g. by omitting redundant bar levels (redundant because there is no specifier present) and making the V' a VP right away, and all the other usual disagreement on the specifics of tree structures, but the basic functionality is the same: Adjuncts adjoin to a node that is by itself already "complete", while complements are direct sisters to the head.

Non-classical phrase structure grammars like, as mentioned by BillJ, CGEL, or tree representations of dependency analyses, ... might handle this entirely differently.


A rule of thumb that will seldom (if ever) lead you astray is that a modifier (or "adjunct") modifies a constituent of the same category as the entire modification structure. Complements do not obey this rule.

So, when we have a structure [A ... [B ... [C ...] ] ], when category A = C, then B modifies C (and B is an adjunct). But when A and C have different categories, C is a complement of B. Similarly, for post-modification, we have [A [C ...] ... B] with A = C.

A way to think about this is that adding a modifier to a constituent does not change the category of the constituent, but adding a complement does change the category.

In a pure phrase structure theory, a modification structure is described by a recursive rule, which necessarily generates an infinite number of modification structures, unlike a complement structure. (Compare adjective and adverb modifiers with subjects and objects, for example.)

So far as I know, no known syntactic theory describes this basic fact of language structure.

  • Are the terms "modifier" and "adjunct" interchangeable? In CGEL, they use the former for the structure of a noun phrase and the latter for that of a verb phrase. – JK2 Sep 20 '17 at 2:52
  • I don't quite understand your ABC structure. Could you explain the ABC structure with my sentences (1) and (2)? – JK2 Sep 20 '17 at 2:54
  • @JK2, [A=VP [C=VP slept ] [B=PP on the floor]] -- adding the PP to the VP gives something of the same category, VP, so PP is a modifier. [A=VP [B=V relied] [PP on my mother]] -- adding the PP to the V gives something of a category differing from V, VP, so in the second example, PP is not a modifier. – Greg Lee Sep 20 '17 at 4:01
  • @JK2, As you noticed, I used "adjunct" and "modifier" as synonyms. Maybe there is some distinction there that I don't understand. – Greg Lee Sep 20 '17 at 4:05
  • 1
    You may be interested to know that addition of a modifier to a constituent, thus creating a new constituent of the same category, was dubbed "Chomsky adjunction" by Haj Ross, because this is what Chomsky had proposed for the structure of the -ing participle of the progressive aspect (in Aspects): [V [V sleep] ing]. – Greg Lee Sep 21 '17 at 11:17

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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