Syntactic preferences and constraints

There are a number of general surface syntactic preferences and constraints in any language. Those in English include:

A. Placing a heavy (i.e. long) constituent at the end of its clause There is a preference for stating the shorter constituents first, before a heavy constituent (whatever its syntactic function). This accounts for the frequency with which a complement clause in subject function is extraposed, and replaced by impersonal it in the subject slot, e.g. It appears definite that the show has been postponed and It annoys Frances to have to say grace before meals. There is also the tendency in some colloquial styles to extract a restrictive relative clause from a pre-predicate NP and move it to the end of the main clause, e.g. The watch broke, which you gave me for Christmas. And the prepositional element of a phrasal verb is likely to be moved leftward over a heavy object NP, so that this can occur clause-finally—compare The doctor brought my father to, where to follows the object NP, and The doctor brought to [that thin man with grey hair who was brought in from that terrible smash-up ten miles down the Pacific Highway], where to precedes the bracketed object NP.

B. Omission of a preposition before complementisers that, for or to. It is a definite constraint. There are a number of verb-root-plus-preposition combinations which together function as a transitive verb; the preposition is retained before an NP or an ing complement clause, but must be omitted before a complement clause introduced by that, for or to (and may optionally be omitted before a complement clause beginning in wh-). Thus: He boasted about his victory in the tournament, He boasted about winning the tournament but He boasted (*about) that he had won the tournament.

Note that if a THAT or Modal (FOR) TO object complement clause becomes passive subject, the underlying preposition may be stated, since it is not then immediately followed by that or for.

(98) They decided on John

(99) They decided (*on) that John would be captain

(99a) That John would be captain was decided (on)

But if the THAT clause is extraposed, it will again follow the (passive) predicate, and the preposition is normally suppressed:

(99b) It was decided (*on) that John would be captain

Note also I was surprised by the fact that the plumber came, where by is followed by two constituents in apposition, the NP the fact and the complement clause that the plumber came. The fact can be omitted, and the preposition by must then be dropped from its position before that, i.e. I was surprised (*by) that the plumber came.

C. Constraint against successive verbs in -ing form within a VP

There are two circumstances in which the general grammatical rules of English would be expected to generate a VP in which successive verbs were in -ing form. In both instances an ungrammatical string results, apparently because of a proscription on successive verbs being in -ing form within a VP. Both circumstances relate to the fact that, unlike a main clause or a THAT or a WH- complement clause, an ING or TO complement cannot include a modal, or tense inflection.

(i) We describe the partial semantic equivalence between some semi-modals and some modals, e.g. be able to and can, be going to and will. The former may be used in place of the latter in functional possibilities where modals are not permitted, e.g. after another modal (if it is semantically plausible to replace modal by semi-modal in this context), as in He will be able to tell you, rather than *He will can tell you.

The semi-modal be able to may be used in a TO or ING clause corresponding to the modal can in a THAT clause:

(100a) I assume that John can climb the tree

(100b) I assume John to be able to climb the tree

(100c) I assume John’s being able to climb the tree

The same should apply to the semi-modal be going to vis-a`-vis modal will. We do get be going to in a TO complement, as in (101b), but not in the ING complement at (101c); this is because of the constraint against two successive -ing verbs within a VP:

(101a) I assume that John will climb the tree

(101b) I assume John to be going to climb the tree

(101c) *I assume John’s being going to climb the tree

(ii) We mentioned that the VP of a to complement clause can include auxiliary have (corresponding both to past tense inflection and to have in a main clause or a THAT clause) and/or auxiliary be (corresponding to be in a main or THAT clause), e.g. I noticed John to have laughed/to be laughing/to have been laughing. The VP of an ING complement can also include have, e.g. I mentioned John’s having laughed. However, it cannot include be as the sole auxiliary element; if it did so there would be successive verbs in -ing form (the initial be taking the -ing which marks this variety of complement, and the following main verb taking the -ing specified by the imperfective auxiliary be), which is not permitted, e.g. *I mentioned John’s being laughing. We can, however, have be together with have, since there is then a verb (be, plus the -en ending demanded by have) between the two -ing forms, e.g. I mentioned John’s having been laughing.

Note that it is perfectly permissible to have successive verbs in -ing form so long as they belong to the VPs of different clauses, e.g. He is enjoying painting the garage (where enjoying belongs to the main clause and painting to the object complement clause) and He’s coming rubbing his eyes (where coming belongs to the main clause and rubbing to a simultaneous appositional clause,).

There is a set of ‘in-between’ cases. Many languages code concepts like ‘begin’ and ‘try’ as affixes to a main verb. English uses lexical verbs that take an ING or TO complement clause; the complement must have the same subject as the BEGINNING or TRYING verb. There are some similarities between verbs of the beginning and trying types and modals (we group them all together as ‘Secondary-A’ verbs). Although begin to walk is best regarded syntactically as two VPs (main clause begin and complement clause walk) linked by to, it is semantically quite similar to ought to walk or be going to walk, each of which is syntactically a single VP.

Speakers vary in judgement as to whether a BEGINNING or TRYING verb preceded by the be imperfective auxiliary (which puts an -ing on the following main verb) may or may not be followed by an ING complement clause, i.e. as to whether He is beginning walking, She is trying eating less, It is continuing raining are grammatically acceptable. This reflects the Janus nature of such constructions—in some ways like two VPs and in others like a single VP.