LINGUIST List 9.651

Tue May 5 1998

Sum: Subject-Control Verb PROMISE in English

Editor for this issue: Martin Jacobsen <>


  1. Karen Courtenay, Subject-Control Verb PROMISE in English

Message 1: Subject-Control Verb PROMISE in English

Date: Mon, 04 May 1998 10:48:00 -0400
From: Karen Courtenay <>
Subject: Subject-Control Verb PROMISE in English

Sorry this summary is late! As a non-syntactician venturing into the
territory of others, I almost chickened out altogether. In case the
original query has been forgotten, I repeat it here:

I would like to ask other native speakers of English if they can use
the following construction in their idiolects:

(a) I promised Kris to buy the cat food. 

This is the so-called subject-control type of verb (though it seems to
be the only one of the type), where the subject "I" is the person who
is to buy the food. It is often contrasted with the object-control
type, as in:

(b) I persuaded Kris to buy the cat food.

where Kris is the one to buy the food.

I cannot, in my idiolect of English, say

 *I promised Kris to buy the cat food 

at all. Nor can several other people I know. We can only say:

(c) I promised to buy the cat food. 
(d) I promised Kris I'd buy the cat food.

Since the subject-control verb PROMISE is very often used in
argumentation in linguistics articles, I would like to know exactly
how common and widespread this construction is. How many other people
cannot use PROMISE as a subject-control verb in this construction?

- --------------------------

In all, 84 native speakers of English responded to the query. 52
speakers found (a) unacceptable, while 32 found it acceptable. The
numbers probably have little meaning; I imagine those who found the
structure unacceptable would be more likely to reply.

Responses came from speakers of a variety of English dialects from the
U.S., England, Ireland, and Australia. It does not appear that the
acceptability of PROMISE as a subject-control verb is related to
British vs. American English, or to dialect at all, since speakers
from the same place often had opposite reactions to sentence (a).

Interestingly, two of the speakers who found the construction
unacceptable said that it was acceptable if used in the negative, as
in the following:
 "I promised him never to do it again." 

(I myself find this sentence much less objectionable than the one in
my original query, but have no idea why.)

My original sentence may not have been the best one to ask about; for
example, Peter Svenonius <> wrote that he
might be more likely to say "I promised Chris to buy him some cat
food" or "I promised Chris to feed his cat," although "I promised Kris
to buy the cat food" was OK with him also. That is, as he remarked,
"this construction is slightly more natural with a coreferent pronoun
in the infinitival."

Several speakers remarked that they considered the construction
acceptable in colloquial speech, but would regard it as incorrect in
formal writing.

Some other speakers found the construction OK when they heard it, but
did not use it themselves. Nancy Hottel of Al-Akhawayn University in
Morocco ( wrote: 'This is the first time I've
ever had the following reaction to grammaticality: I can IMAGINE
myself saying "I promised my husband to buy food" and that if I did
say it it wouldn't be wrong, i.e. when I say it, it doesn't sound
wrong. But when I present myself with only: "I promised my husband..."
it always comes out as "that I would x."'

It is clear that PROMISE as a subject-control verb is not new; the OED
gives this example from the year 1467: "[The parker] hathe promessed
me to make it as wel as he kane fore me." and this one from Pope
(1737): When...we...promise our best Friends to rhyme no more."

A few speakers offered some other possible subject-control verbs of
this type. David Denison of the Univ. of Manchester
( cited a 1911 example, "The publishers
... then OFFERED the author to purchase the copyright for L100,"
adding that this was impossible in his dialect. Polly Jacobson of
Brown University ( offered THREATEN, as in
"I threatened Kim to leave," remarking "I find this a bit even
stranger than promise, but again, over the years it's been clear to me
that many people like this just fine." (I find both of these verbs
completely unacceptable as in this construction.)

Several people suggested a corpus search, but as an employee of a
software company specializing in machine translation I do not have the
free access to databases that university researchers have. Many
corpora are either not available at all to for-profit organizations or
available to them at great expense. Perhaps someone else would like to
undertake this task....

It appears that the equivalent of "I promised Kris to buy the cat
food" is just fine in several other languages, including Russian,
Spanish, Hebrew, and Norwegian (according to Peter Svenonius, the
Norwegian verb for "promise" is the only SC verb taking a bare
infinitive, as in English).

Ken Takami of Tokyo Metropolitan University
( mentioned Rosenbaum's (1967) Minimal
Distance Principle (MDP), "which states that the subject of the
infinitive clause is the nearest NP to its left. This principle
correctly accounts for sentences like "I persuaded Kris to buy the cat
food", but "I promised Kris to buy the cat food" is a problem for
it. However, there are many object-control verbs besides "persuade",
such as "order, tell, permit, force, allow, forbid, require", while
subject-control verbs seem to be restricted only to "promise." Coupled
with this, the fact that many people reject the sentence pattern (a)
would suggest that Rosenbaum's MDP is working for those people."
Polly Jacobson also mentioned this.

- --------------------

Following are just a few of the many interesting responses; I
apologize for not being able to include all of them, and would like to
thank all those who responded.

- ---------------------

I cannot thank you enough for posting this. All these years I thought
perhaps I was the only person missing some crucial syntactic
construction... While I can understand "I promised Kris to buy the
cat food," I'd rate it ungrammatical in my own speech, or at the very
least as highly, highly questionable. I, too, would far sooner use
one of your two alternatives (...promised to buy.../promised Kris I'd
buy...). And I'm not even sure I've ever heard it, except in syntax

Lance Nathan
Brown University '99 | Major: Math & Linguistics

- ------------------------

I agree in general that sentences like:

I promised Kris to buy the cat food.

are very odd, and the type of example I reluctantly accept in
linguistics textbooks (aggravated by the fact that as you say no other
verb works that way). However, to be fair, the structure DOES seem to
work for people in a case like:

I promised Kris never to buy catfood again.

That is, at least with NEVER.
I'd be interested to hear how other people react.

Paul Westney

- ----------------------------

My idiolect also does not allow for subject control. I find the
sentence "I promised Kris to buy the cat food" either ungrammatical
(under the interpretation that I am the one buying the cat food) or
semantically anomalous (under the interpretation that Kris is buying
the cat food). When I hear the sentence, I mentally imagine Kris
buying the cat food. I asked an employee of mine, and she said the
same thing (without my first having told her my intuitions about the

Interesting issue.

- Tony Wright <>
St. Philip's College
San Antonio, Texas, USA

- -------------------------- 

(The following response brings up the question of SC verbs with
passive infinitives)

Subject control with 'promise' is fine for me. Under favourable
pragmatic conditions, usually with a passive complement, I can also
get it with other verbs such as 'ask' or 'beg':

 "The boy asked/begged the teacher to be allowed to leave the room."

I am a speaker of British English (from the north of England), but my
impression is that it is not a regional thing. I know other British
speakers who, like you, do not accept subject control with 'promise'.

Mike Jones,
Department of Language and Linguistics,
University of Essex,

- ---------------------------------------

I would be astonished if you found any correlations with any
demographic variable (area, age, sex, "class")--I strongly suspect
that the variation you see is due to grammars being underdetermined by
the data they have to account for.

Georgia Green <>

- ------------------------------------------------------

Karen: I saw your posting to the LINGUIST list. I would be very
interested in seeing your results. I am one of those who find
sentences like "I promised Kris to buy the cat food" perfectly normal,
and when I first learned that there are speakers who don't accept the
construction I was surprised. My impression is that the same holds
vice versa--those who don't accept the construction are surprised that
there are prople who do. I'm not aware of any clearcut geographical,
social dialect divisions that correlate with acceptance versus
rejection of the construction, but maybe you'll get enough responses
to assess this. Radford claims, I think in his big red syntax book,
that the construction is generally unacceptable to people in Britain
(and, by implication, that it is generally acceptable to people from
America), but I'm from Britain and have no problem with the
construction, and never have. (A lot of English data judgments that I
encountered in my early years as a syntactician required some
suspension of disbelief, but not this one.)

I think that all is required by most of the arguments in the
literature that pertain to "promise" is that there should be some
variety of English that has "promise" as a subject-control
verb--varieties that allow only finite complements in the presence of
a nominal object are irrelevant, rather than counterexamples. And
quite a number of other European languages show patterns similar to
English (i.e. the variety that allows infinitive complements here,
although I don't know if there is comparable variation in
acceptability across speakers of these languages).

The variation in acceptability judgments does, however, lead me to
wonder about Carol Chomsky's (1969) results on the difficulties
children have in acquiring the "promise + nominal object + infinitive"
construction, since there seems to be a presupposition that this
construction was part of their target variety of
English. Best--Bernard.

Bernard Comrie
Dept of Linguistics GFS-301 tel +1 213 740 3674
University of Southern California fax +1 213 740 9306
Los Angeles, CA 90089-1693, USA e-mail

Address from mid-May 1998 (new telephone, fax, e-mail not yet
available): Max Planck Institute for Evolutionary Anthropology
Inselstrasse 22-26 D-04103 Leipzig, Germany

- ------------------------

When I worked at a syntax project at UCLA, we came across this
question. There were three faculty members and ten, twelve (grad)
students on the project, and we found that some of us could say

(1) I promised Fred to buy the cat food

and other could not, just like you can't. And the latter could only
use those sentences that you cite as being grammatical for you.

Later, when I also worked on dialectoloy and on sociolinguistics, I
did a little listening and questioning, but not in any formal way. I
tentatively concluded that geographically speakers of "your" dialect
were from North-East dialect areas, whereas Midland speakers (both
North and South Midland) could use sentence (1). Southern speakers
largely can too, as can Californians. (N.B. I have no data on other
areas, though in the prairy provinces of Canada, where I'm from, we
can use sentence (1). But then, the prairy provinces are much like
California, in that linguistically nearly anything goes. That's
probably due to the many recent arrivals and the resulting dialect

Socially speaking, there is some question of age; for older people in
general, and in the Southern dialect area in particular, sentence (1)
is more often unacceptable than it is for younger speakers. But there
is, as so often, also a question of linguistic conservatism: Some
younger speakers retain more of the language of their parents, while
others are more influenced by their peer-group language. This, of
course, includes the acceptability of sentence (1).

You are right in saying that "promise" is the only verb that works
like this. The only, even vaguely, similar case I know of is:

(2) I asked him what to wear

as opposed to

(3) I told him what to wear

Note that in (2) it's the subject of "ask" that's deleted, while in
(3) it's the object of "told", similar to the well known
"promise/persuade" pair.

Peter Menzel (

- ---------------------------------

There have been several psycholinguistic papers looking a subject vs.
object vs. ambiguous control verbs. The debate was about when control
info was used.

Boland, J. E., Tanenhaus, M. K., & Garnsey, S. M. (1990). Evidence for
the immediate use of verb control information in sentence processing.
Journal of Memory and Language, 29, 413-432.

is one and it should have sample materials at the end (JML is real
good about that). It should also cite the earlier work which may or
may not have sample materials, but you could write the authors' and
ask for them.

That should get you a larger set of verbs than just promise and you
can see if you and your friends like them better.

Marie Egan University Of South Carolina


Karen Courtenay

Senior Linguist <> Language Engineering Corporation
385 Concord Ave <> Belmont, MA 02178-3037 <> U.S.A.
Tel: 1-617-489-4000 x719 <> Email:
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue