LINGUIST List 5.685

Tue 14 Jun 1994

Sum: Coordination of gerunds with to-infinitives

Editor for this issue: <>

Directory

1. R.Y.L. TANG, Coordination of gerunds with to-infinitives

Dear netters,

On May 6, 1994, I posted the following query on _Linguist_:

Consider the sentence

(1) My interests include swimming, reading and playing chess.

3 NPs are conjoined by 'and'. Consider another example which is unacceptable:

(2) * My interests include swimming, reading and to play chess.

Why is 'to play chess' not allowed in this nominal slot, considering the fact
 that it can function as a nominal in the subject position:

(3) To play chess is what I want now.

and in the complement position:

(4) His only pastime is to play chess.

I am having the notion of 'to-infinitive nominal clause' (Quirk and Greenbaum
 1973) in mind when positing this problematic case. Can anyone explain the above
 unacceptability using a non-GB framework?

After reading the replies, I hereby summarize the points into several
aspects:

A. Difference between the -ing VP and the to-VP

Various tests show that the gerund is more 'nominal' than the
to-infinitive, although they can both occur in certain slots expected to
be filled by nominal elements. Pauline Jacobson wrote: "there are
environments in which NPs can occur and in which infinitives cannot; the
reverse is also the case, and so the conclusion would seem to be that
infinitives are not NPs." She gave the examples:

(5) a. *I dislike to eat meat.
 b. I dislike the meal.

A true NP must be one which can be the verbal complement of 'dislike'.
Moreover, as noted by Pauline, infintives (ans also full CPs) cannot
occur as PP complements, while ordinary NPs and gerunds can:

(6) a. *You can rely on to win.
 b. You can rely on the victory.
 c. You can rely on winning.

Pauline also scrutinized the copular constructions of (3) and (4). She pointed
out that "the two constituents in a copular sentence do not need to be
of the same category", and therefore (3) can (4) cannot be strong
evidence which proves that the to-infinitive 'to play chess' is a
nominal element. Both Pauline and Boundary (ID I4e/r3h3umEYiW3b8wgf4A)
use the example below to illustrate that non-NP constituents can occur
around the copular:

(7) Under the bed is a good place to hide.

Although 'under the bed' can be moved to the postcopular position, as in
the case of (4), it does not mean that precopular and postcopular constituents
will be exactly identical semantico-syntactically, as Kiyoshi Ishikawa
pointed out.

Boundary used (7) to show that 'To play chess' in (3) "is not
necessarily the subject". Even if it is the subject, it is still not a
nominal, as other categories can also occur there (Yehuda N. Falk). To
reiterate, the -ing constituents are more 'nominal' than the
to-infinitive (Peter-Arno Coppen).

Nigel Love used 'verbal noun' and 'deverbal noun' to talk about the
 the examples of my own. He stated that "the
-ing form is not so much a verbal noun (i.e. gerund) as a deverbal noun,
...", and went on with the following examples to illustrate the
difference in behaviour for verbal and deverbal nouns:

(8) I like collecting stamps (verbal noun; replaceable by the
infinitive)

(9) I like stamp-collecting (deverbal noun; replaceable by any other
normal noun, e.g. "philately")

"Swimming" and "reading" in my examples belong to both verbal and
deverbal noun categories. If they are more to the deverbal side, they
may not be allowed to be conjoined with the to-VP 'to play chess' then.

B. Coordination constraints

Some respondents concentrated on the constraints for a coordinated
structure. Guy Modica wrote: "Conjunctions such as 'and' require that
the conjuncts be of the same phrasal class....When you have two
gerunds...conjoined with an infinitive, we sense it to be
ungrammatical....The crucial point is not the class of an offending
conjunct, but a restriction on what a conjunction can conjoin." Quite
opposite to Guy, Kiyoshi Ishikawa wrote: "It's wrong to require
categorical identity of conjuncts in the first place,..." He used Sag et
al. (1984)'s example to illustrate a coordination of 2 non-identical
constituents:

(10) John is a Republican and proud of it.e

Pauline Jacobson wrote: "The basic generalization though, ...is that two
constituents of different categories may conjoin just in case they are
in an environment where both are possible." Thus her examples:

(11) ?My favorite pastimes are swimming and to play chess.

(12) a. My favorite pasttime is swimming.
 b. My favorite pasttime is to play chess.

My example (2), then, is bad because 'to play chess' cannot occur in the
context of (12b) if the main verb is changed to 'include' (see C below).

Some respondents cited some awkward but acceptable examples to
illustrate that coordination of non-identical constituents is possible.
Michael Kac's example:

(13) ?I like hiking, golfing and to swim.

However, John Nerbonne thought that (14) is unacceptable:

(14) *I like swimming and to read.

Interestingly, Marit Julien stated that the kind of construction in my
example (2) is perfectly acceptable in Norwegian -- "an infinitive is
allowed in any nominal slot, in a fashion parallel to the
nominalisations in -ing". Thus, Marit thought that the restriction
governing the coordination in (2) must be "particular to English".

C. Complementation of 'include'

Some respondents stated that my problem concerns *solely* with the
requirements of its complement by the verb 'include'. In other words, it
is not a problem of coordination constraints, as the example below
(cited by many of the respondents) shows:

(15) *My interests include to play chess.

This example shows that it is 'include' which requires an NP complement
(although Boundary was skeptical about the NP status of "playing chess")
rather than a to-VP complement. Boundary wrote: "I think the answer
probably has to do with the fact that coordinate phrases have to share
certain features (in the sense of syntactic features in lexicalist
syntax) which are required by the verb (or whatever) that selects them,
or by the syntactic position in which they appear." He said that since,
for example, 'is' can take virtually any category as its complement, the
conjoined structure involving 'and' among different categories is
possible in the complement slot after 'is'.

Kiyoshi Ishikawa made a generalization concerning this: "...each
conjunct has to respect the subcategorization information imposed in the
coordinated mother....the verb imposes some requirement in its
complement which is compatible with 'playing chess' but not with 'to
play chess' (...but I think...this requirement is semantic)."

D. Rhetoric consideration

One respondent, John Cowan, said that even if my example (2) and example
(15) were acceptable, "(2) would be ruled out on rhetorical grounds by
the rule 'Match parts!'. This rule requires formal similarity between
the elements of an expressed or implied conjunction." This is what I
had in mind when I made up my query -- I was advising people to 'keep
the grammatical categories the same when writing the hobbies on your
re'sume''.

To sum up again, the very different syntactic behaviours of VP-ing and
to-VP constituents, possibility of coordination of non-identical
categories, the subcategorization requirement for the complement of
'include' and, with minor importance, rhetorical consideration, come
together to form the full picture.

Many thanks to the respondents again:

Loren Allen Billings
'Boundary'
Peter-Arno Coppen
John Cowan
Yehuda N. Falk
Dick Hudson
Kiyoshi Ishikawa
Pauline Jacobson
Marit Julien
Michael Kac
Chuah Choy Kim
John Lawler
Nigel Love
Guy Modica
John Nerbonne

Some of them have provided me with many references possibly related to
coordination. If anybody is interested please email me and I will give
him/her the titles.