LINGUIST List 7.1534

Wed Oct 30 1996

Disc: _watch_

Editor for this issue: Ann Dizdar <>


  1. Knud Lambrecht, Re: 7.1525, Sum: _watch_

Message 1: Re: 7.1525, Sum: _watch_

Date: Tue, 29 Oct 1996 16:00:59 CST
From: Knud Lambrecht <>
Subject: Re: 7.1525, Sum: _watch_
I read with great interest Dick Hudson's contribution on WATCH. I'm
sorry I didn't see the previous contributions. I don't have anything
to say about the (fascinating) generation issue, which I'm totally
mystified by. But I do have some ideas about the definite/indefinite
contrast with null complements which he mentions:

>Why on earth should it matter what the other person does? Answer: it
>changes the nature of the understood object from definite to
>indefinite. (This is a well-known distinction which I'm afraid I can't
>support with references; if anyone else could offer some
>bibliographical respectability I'll share it.) In

There is very interesting work on the matter by Fillmore, in the form
of a typology of null complement types:

Charles Fillmore.1986. "Pragmatically controlled zero anaphora" BLS 12.

and more recently in 

Charles Fillmore & Paul Kay, Construction Grammar Coursebook (to appear,
 CSLI, Stanford).

On French there is (applying Fillmore's system):

Lambrecht, Knud & Kevin Lemoine. 1996. "Vers une grammaire des
complements zero en francais parle." In Absence de marque, et
representation de l'absence. Travaux du CerLiCO (Cercle linguistique
du centre ouest).

(Very respectable bibliography, especially the last entry.)

Fillmore draws distinctions between types of null instantiation
(i.e. ways in which a null argument in the valence frame of a
predicator can be interpreted): (i) "indefinite null instantiation"
(INI), (ii) "free null instantiation" (FNI), and (iii) "definite null
instantiation" (DNI). Moreover, within each type there is a
subdivision between lexically controlled and constructionally
controlled null instantiation. Furthermore, within the INI type
there's a distinction between totally indefinite (null complement =
"stuff" "things", as in "My throat hurt so much that I couldn't eat or
drink") and indefinite but representing a subset of the possible
complements of the predicator ("I haven't eaten yet" = I haven't eaten
a meal, or "She drinks" = she drinks alcohol, etc.). Finally there
is, within the DNI type, a distinction between between
"frame-controlled" null instantiation, where the understood definite
object is provided as part of a semantic frame (as in "She hung up",
i.e. the phone, or "What time do you open?" i.e. your store, or "The
phone is ringing, why don't you answer" etc.), and null complements
representing specific entities or situations evoked in recent
discourse (this latter criterion is crucial). The last type exists in
English mostly with lexical control (but constructionally governed DNI
is common in many other languages, as is well known).

So it might be useful to look at Dick's various examples in terms of
Fillmore's typology:

>(1), "I'm watching" means `I'm watching this (programme)' (definite),
>but in
>(2) it means `I'm watching tv', i.e. I'm doing some tv-watching
>(indefinite). And everyone is happy with indefinite-object
>intransitive WATCH.
>(2). It seems that indefinite understood objects are fine just as with
>verbs like SING and READ.
>(3) I sang/read/watched for an hour before going to bed. Here each
>verb has its prototypical object as its understood object - songs,
>books/papers, tv. It seems that tv is now the prototypical watch-ee!

The question here is: what kind of DNI (definite null instantiation)
is (1) and what kind of INI (indef. null inst.) is (2), and is (3)
really like (2)? Not being an oldish Brit (I'm oldish, but German), I
don't have clear intuitions on the matter, but it seems to me that
Dick's (1) is a case of frame-controlled DNI. "I'm watching" is
comparable here to "She hung up" etc. A test which I came up with to
distinguish frame- controlled from specific-entity DNI is pronoun
substitution: I assume you cannot say (1'), instead of (1) (in the
given context, of course):

(1') ?? Hey, I'm watching it.

As for (2), I think it's in fact not quite like (3): (3) seems to be
"totally indefinite", i.e. "I read" means "I read stuff, e.g. any old
book or article or whatever, and "I sang" means "I sang stuff,
i.e. songs". But (2) doesn't mean the same as (2'):

(2') Hey, I'm watching stuff! 

It is true that the understood object after 'read' or 'sing' is indeed
not just anything but obvioulsy something readable/singable,
i.e. something that is semantically compatible with the meaning of the
verbs. But the point is that it is *anything* that can be read or
sung. But in (2) the object is not "anything that can be watched", but

 >Moreover, plenty of people pointed out to me that they can omit the
>object in cases where the understood object is a definite event:
>(5) Pat danced while Jo watched. Here Jo didn't just do some
>watching, but she specifically watched Pat's dancing. (It could of
>course have the same meaning as (4).) Similarly for examples like (6).
>(6) A. Now I'm going to press this button. Are you watching (me)?
 >B. Yes, I'm watching.
>Therefore the change can't involve a prohibition on optional
>*definite* objects. The restriction to `events' is needed because I
>don't think WATCH does allow understood objects when its object is a
>person or thing (this is just my judgement):
>(7) A. Would you watch my suitcase for me while I buy a paper, please?
 >B. Sure - I'll watch *(it).

Not being a native speaker of English, I shouldn't make any judgments,
but I think this * is definitely a *. That's because of the very
general (but not absolute) prohibition gainst DNI in English
(cf. above). The reason so many people like DNI when the understood
object is an event is related, I would think, to the fact that in
general propositions can be DNI'ed much more easily in English than
entities. Evidence is the huge, though finite, number of verbs that
permit definite clausal null complements ("Let's start!" i.e. the
specific activity under consideration, "He finished before everyone
else", "I know", "I saw", etc. etc.) Why is it easier to supply
situations than entities? I don't know, but it seems to be a
cross-linguistic fact. (Specialists may have something to say about
the difference between entities and propositions with respect to the
notion of 'definiteness'.)

>However you might think that a tv programme would qualify as an event
>- after all, tv is solid wall-to-wall event, isn't it? So it's very
>odd that people accept (5) and (6) while rejecting (1).

In light of what I just said, maybe it's not so odd. (1) involves a
definite specific frame-related entity (the channel or program(me)),
while (5) and (6) involve events/situations, which generally speaking
are easier to "delete".

Maybe some of the distinctions I reported on above can be useful in
sorting out the various factors influencing people's judgments.

>change in culture. Presumably the fact that I never see tv till I was
>a teenager is relevant, but the logic would need to be spelled
>out. That's too difficult for a linguist so I give up!



Knud Lambrecht
French and Romance Linguistics
Dept. of French & Italian
UT Austin
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue