Although language seems to be spoken linearly, since the only way we can do language is by having one word follow another, sentences and phrases are actually formed by attaching constituents to each other in a hierarchical construct. Consider the following sentence.
(1) Harriet mistakenly went home with her cousin’s jacket.
We know intuitively who did what. Furthermore, we know that ‘Harriet’ can stand alone as a simple subject. We know that ‘mistakenly’ modifies the verb ‘went’ but not the noun ‘Harriet’. Since this is the case, then we know that ‘mistakenly went home’ is a unit since the verb ‘went’ must be accompanied by a place. We also know that ‘with her cousin’s jacket’ acts as a unit since leaving out any of its components renders it ungrammatical. So now we have a sentence parsed into its constituencies.
(2) [Harriet] [mistakenly went home] [with her cousin’s jacket.]
We also know that these constituents are arranged in a hierarchical order that cannot be altered.
(3) * [mistakenly went home] [Harriet] [with her cousin’s jacket.]
or * [with her cousin’s jacket] [Harriet] [mistakenly went home.]
Constituents can contain constituents.
(4) [Harriet, who just purchased a Jaguar,] [mistakenly went home] [with her cousin’s jacket].
So now within the constituent ‘Harriet’ we have the constituent ‘who just purchased a Jaguar’. Again, we know that ‘who just purchased a Jaguar’ is found in the constituent ‘Harriet’ due to the fact that if it is moved to another position in the sentence, we have an ungrammatical structure.
(5) *[Harriet] [mistakenly went home, who just purchased a Jaguar] [with her cousin’s jacket].
The fact that sentences are constructed of constituents that are arranged in a hierarchical order is supported by the fact that when brackets are moved showing a different grouping of words, meaning is altered. In (6) a ‘Jack and Jill’ have the same grandmother whereas in (6) b ‘Jack’ and ‘Jill’s grandmother’ are two different entities.
(6) a [Jack and Jill]’s grandmother left the party.
b [Jack] and [Jill’s grandmother] left the party.
Constituency Defined
A constituent is a word or group of words that form a unit built around a head. They can be made up of words, phrases, and even entire clauses.
Phrasal Heads
The ‘head’, the word around which the constituent is built, determines the grammatical properties of its constituent. In the example phrase ‘the Cheshire cat’, ‘cat’ is the word around which the phrase is built. It is the head of the phrase. Since cat is a noun, ‘the Cheshire cat’ is a noun phrase, or NP. A head can be one word such as ‘Harriet’.
(7)
In (7) the head is the final node ‘N’ which contains the noun ‘Harriet’. A constituent can contain other words that modify the head.
(8)
In (8), the head ‘boots’ is found in the final N node. Each of the other nodes contains words that modify ‘boots’.
Just a note on Constituency vs Dependency Grammars
Not all syntacticians are in agreement about how to understand and describe the internal structure of sentences. The two schools of thought are the phrase structure (or Constituency) grammar tradition, and the dependency grammar tradition. We will focus here on phrase structure grammars (hereafter PSGs), as most post-Chomskyan grammars fall into this category, and because phrase structure rules (see elsewhere on Linguisticsnetwork) are only relevant under PSGs. The primary difference between the two traditions is how they conceive of the initial division of a sentence. PSGs begin hierarchical analysis by dividing the subject (an NP) from the predicate (a VP). Dependency grammars see this division as artificial, and understand sentences to be built around their root verb and its valance, or requisite number of NP arguments.
Constituency Tests
Most often a speaker has an intuition as to which strings of word form a constituent. However, when this information is not clear, various ‘tests’ can be applied to identify the constituent structure of a sentence.
Below is an explanation of the constituency tests most often used. A string of words need pass only test as evidence of a constituent. Furthermore, not every test will work in every situation. The tests included here have been selected for their broad applicability.
- The Question Test – Can the candidate constituent be used to answer a question? If a group of words can answer a question, it most likely is a constituent. Let’s look at (9).
(9) [The French cooks] [love to bake] [chocolate eclairs].
The question “Who loves to bake chocolate eclairs?” can be answered by the constituent ‘The French cooks’ but not by ‘The French’, or ‘The French cooks love’.
The questions “What do the French cooks love?” can be answered by “to bake chocolate eclairs’ but not by ‘bake chocolate eclairs’.
- The Movement Test– Can the candidate constituent be moved elsewhere in the sentence while maintaining grammaticality? The ability to move, or topicalize a string of words, resulting grammatical sentence is evidence of a constituent.
(10) a No one would ride in the car with Henry [because he hadn’t showered.]
We can topicalize ‘because he hadn’t showered’ and maintain grammaticality.
(10) b [Because he hadn’t showered], no one would ride in the car with Henry.
However if only a subset of this constituent is moved, an ungrammatical sentence results.
(10) c * [Because he hadn’t] no one would ride in the car with Henry showered.
3. The Substitution/Replacement Test– Can the candidate constituent be replaced with a pro-form? The ability to replace a string of words with an NP or VP pro-form is evidence of a constituent.
(11) a [Todd and his classmates] like the Eagles.
b [They] like the Eagles.
But Not
c * [They] classmates like the Eagles.
(12) a Michelle [walked to the zoo].
b Michelle [did].
But Not
c *Michelle [did] to the zoo.
4. The Ellipsis Test– Can the candidate constituent be deleted? A constituent may be, in certain structures, elided or deleted. If the entire constituent is not included in the deletion, an ungrammatical structure results.
(13) a After the hosts taste the wine, then we will. (taste the wine)
But Not
b * After the hosts taste the wine, then we will taste. (the wine)
5. The Passivisation Test – Can the candidate constituent change places with a different argument in the sentence? If the subject and object can be reversed, or found in differing locations in a passive construction, and result in a grammatical sentence we can reliably say that each arguments is a constituents. In passivisation, the object moves to the subject position and the subject is lowered to an optional ‘by phrase’.
(14) a [The foolish five year old] teased [Susan’s remarkably short brother.] b [Susan’s remarkably short brother] was teased by [the foolish five year old.]
6. The Coordination Test– Can the sentence be modified so that the candidate constituent appears coordinated with a structurally similar constituent? If a phrase can be paired with another of the same syntactic category using a coordinating conjunction, then it forms a constituent.
(15) a Reggie ate a tuna-fish sandwich for lunch.
We can test the VP [ate a tuna-fish sandwich for lunch] by adding another VP.
b Reggie [ate a tuna sandwich for lunch] and [played with a soccer ball in the park.]
VP VP
But Not
c * Reggie [ate a tuna sandwich for lunch] and [with a soccer ball in the park.]
VP PP
(15) b is evidence that [ate a tuna-fish sandwich for lunch] is a constituent.
Final Thoughts
So you now have a better understanding of the term ‘constituency’. Drawing syntactic trees can be quite challenging. However being able to identify constituents seriously simplifies the process. Understanding the concept of constituents is also key in examining issues in speech perception and speech pathology.
For practice, go to our exercise on constituents.
Resources on Syntactic Constituency
Carnie, Andrew. Syntax: A Generative Introduction. Cornwall, UK: Blackwell Publishing, 2002.
Newmeyer, Fredrick. Language Form and Language Function. Cambridge: MIT Press, 1998.
D. Kelleher. R. Aronow