TESTING FOR CONSTITUENTS
Before attempting to see how a stretch of language can be broken down into units, it is useful to be able to reinforce our intuitions as to where boundaries lie. This can be done by applying certain tests in order to identify whether a particular sequence of words is functioning as a constituent of a higher unit or not.
For instance, the following sequence, which constitutes a grammatical clause or simple sentence, is ambiguous:
Muriel saw the man in the service station
Two interpretations are possible, according to how the units that make up the clause are grouped into constituents, expressed graphically as follows:
In version 1, the prepositional phrase in the service station forms part of the constituent whose head-word is man (the man in the service station) and tells us something about the man; whereas in version 2 the same prepositional phrase functions separately as a constituent of the clause and tells us where Muriel saw the man.
Evidence for this analysis can be sought by such operations as (a) coordination, (b) wh-questions, (c) clefting, (d) passivation and (e) fronting. Tests (b) to (e) involve moving the stretch of language around and observing its syntactic behavior. Testing by coordination involves adding a coordinate that realizes the same function; only stretches of language that realize the same function can be coordinated:
(a) It can be seen that different types of conjoin are required according to the function of in the service station:
(i) Muriel saw the man in the service station and the woman in the shop.
(ii) Muriel saw the man in the service station and in the shop.
(b) The wh-question form and the appropriate response will be different for the two versions:
(i) Who did Muriel see? – The man in the service station.
(ii) Where did Muriel see the man? – In the service station.
(c) Clefting by means of it + that-clause highlights a clause constituent and thus yields two different results:
(i) It was the man in the service station that Muriel saw.
(ii) It was in the service station that Muriel saw the man.
Wh-clefting gives the same result:
(i) The one Muriel saw was the man in the service station.
(ii) Where Muriel saw the man was in the service station.
The form the one (that . . . ) is used in this construction since English does not admit who in this context (*Who Muriel saw was the man in the service station).
(d) Passivization likewise keeps together those units or bits of language that form a constituent. The passive counterpart of an active clause usually contains a form of be and a past participle:
(i) The man in the service station was seen by Muriel.
(ii) The man was seen by Muriel in the service station.
It is not always the case that a sequence responds equally well to all five types of test. Certain types of unit may resist one or more of these operations: for instance, frequency adverbs such as often and usually, and modal adverbs like probably, resist clefting (*It’s often/ usually/ probably that Muriel saw the man in the service station), resulting in a sentence that is ungrammatical. Unlike some languages, in English the finite verbal element of a clause normally resists fronting (*Saw Muriel the man in the service station). Nevertheless, if two or more of the operations can be carried out satisfactorily, we can be reasonably sure that the sequence in question is a constituent of a larger unit.
We now turn to the description of units, their classes and the relationship holding between them.
