Logical Form and the Vernacular

Abstract

Vernacularism is the view that logical forms are fundamentally assigned to natural language expressions, and are only derivatively assigned to anything else, e.g., propositions, mental representations, expressions of symbolic logic, etc. In this paper, we argue that Vernacularism is not as plausible as it first appears because of non-sentential speech. More specifically, there are argument-premises, meant by speakers of non-sentences, for which no natural language paraphrase is readily available in the language used by the speaker and the hearer. The speaker can intend this proposition and the hearer can recover it (and its logical form). Since they cannot, by hypothesis, be doing this by using a sentence of their shared language, the proposition-meant has its logical form non-derivatively, which falsifies Vernacularism. We conclude the paper with a brief review of the debate on incomplete definite descriptions in which Vernacularism is assumed as a suppressed premise.

Full text

LogicalFormandtheVernacular
REINALDO ELUGARDO AND ROBERT J. STAINTON
Abstract: Vernacularism is the view that logical forms are fundamentally assigned to
natural language expressions, and are only derivatively assigned to anything else, e.g.,
propositions, mental representations, expressions of symbolic logic, etc. In this paper,
we argue that Vernacularism is not as plausible as it first appears because of non-
sentential speech. More specifically, there are argument-premises, meant by speakers
of non-sentences, for which no natural language paraphrase is readily available in the
language used by the speaker and the hearer. The speaker can intend this proposition
and the hearer can recover it (and its logical form). Since they cannot, by hypothesis,
be doing this by using a sentence of their shared language, the proposition-meant has
its logical form non-derivatively, which falsifies Vernacularism. We conclude the paper
with a brief review of the debate on incomplete definite descriptions in which Vernacu-
larism is assumed as a suppressed premise.
Introduction
The overall aim of this paper is to urge that (1) below is not obviously true,
and to explain why that is important.
1
The paper began life as part of the 1998 American Philosophical Association, Central Division
panel on Ernest Lepore and Kirk Ludwig’s neo-Davidsonian account of logical form. We are
grateful to them for their encouragement and questions. It was later presented, in revised form,
at Illinois Wesleyan University, in the Cognitive Science Research Seminar at Carleton Univer-
sity, as a three part ciclo de conferencias at the University of Puerto Rico at Mayaguez, and as a
panel paper at the 1999 Canadian Philosophical Association Meeting. We are indebted to the
audiences at these various spots for very useful questions and suggestions. We are especially
indebted to Andrew Botterell for his penetrating and helpful comments on multiple drafts of
this paper. We would like to thank the anonymous referee for Mind & Language for some very
helpful comments. Many thanks as well to Kent Bach, Lenny Clapp, Monte Cook, Steven Davis,
Bernard Linsky, Randal Marlin, Robert May, Paul Pietroski, and Daniel Stoljar for discussion
and comments. Since writing this paper, we came across a paper by Dan Sperber and Deirdre
Wilson that dovetails with ours in certain respects, cf. Sperber and Wilson (1998).
Address for correspondence: Reinaldo Elugardo, Department of Philosophy, University of
Oklahoma, Norman, Oklahoma 73019–2006, USA.
Email: relugardo얀ou.edu
1
Other philosophers have defended the general view that propositions (thoughts, conceptual
representations) have their logical forms fundamentally rather than derivatively, cf. Grice (1989),
Fodor (1978), Sperber and Wilson (1981). Although we argue for the same thesis in this paper,
our main argument differs from theirs in that we argue that the phenomenon of non-sentential
speech pose a serious problem for Vernacularists. Julius Moravcsik has argued that natural languages
cannot be formally represented in any (possible) system of logic, cf. Moravcsik (1998). To mention
just one of his arguments, English generic sentences, e.g. ‘The beaver builds dams’, cannot be
assigned a single logical form that captures all of their entailment relations (Moravcsik (1998,
pp. 73–84)). If Moravcsik’s general thesis is correct, then no natural language sentence has a logical
form (in the sense of having a logical formula that represents its meaning.) But if no natural
language sentence has a logical form, then Vernacularism is false since it entails that some
do have logical form. In this paper, we grant in our main argument the Vernacularist premise
that natural language sentences have a logical form. Hence, even if Moravcsik were wrong
about natural languages—and we take no stand on that issue here—we would still have a
strong argument against Vernacularism if our argument is cogent.
Mind & Language, Vol. 16 No. 4 September 2001, pp. 393–424.
Blackwell Publishers Ltd. 2001, 108 Cowley Road, Oxford, OX4 1JF, UK and 350 Main Street, Malden, MA 02148, USA.

394 R. Elugardo and R.J. Stainton
1. Vernacularism: The view that logical forms are fundamentally assigned
to expressions of natural language, and are only derivatively assigned
to anything else: e.g. propositions, mental states, etc.
We will begin, however, by saying a little about what Vernacularism amounts
to—by specifying what, by our lights, it is to have a logical form; and what
it is to have one ‘fundamentally’.
2
1. Preliminaries
1.1. Having a Logical Form Fundamentally
We won’t even try to canvass everything philosophers and linguists have
intended by ‘logical form’.
3
But to clarify what we mean by it, let us say this:
2. Logical Form: The logical form of ␣is that in virtue of which agents
recognize ␣’s structural entailment relations.
To have a logical form, then, ␣must be capable of being true or false (possibly
2
We recognize that, in focusing on what it is to have a logical form, we are dodging the
issue of what sort of things logical forms are. Are they concrete, or abstract? Or maybe
mental? Are they objects, or are they properties? More specifically, are they expressions (e.g.
of some artificial language), or are they patterns—exhibited by natural language sentences,
propositions, beliefs, etc.? For present purposes it doesn’t much matter what logical forms
are. What we care about is when it’s correct to say that a thing has a logical form. And,
according to our stipulation, as long as something has structural entailment relations, and
those are recognizable by agents, then it has a logical form. In which case—and this will be
essential in what follows—as long as something can be used as a premise in an argument, it
must have a logical form. For, if it is to be used as a premise, the thing must have entailment
relations; and those entailments relations must be recognizable to the interlocutors. Some
philosophers will deny our claim that if xcan be used as a premise in an argument or have
entailment relations, then xhas a logical form, cf. Moravcsik (1998). We suspect that their
arguments against our claim are based on the assumption that the logical form of a thing is
some logical formula used to represent its meaning or content. We are not committed to
that assumption, cf. 20ff.
3
By ‘logical form’, we do not mean the linguist’s notion of a level of representation (LF) of
certain grammatical properties, such as scope and anaphoric relations. Thus, we are not
claiming that certain non-linguistic items have their LFs fundamentally. For a discussion on
the differences between the logician’s notion of logical form and the linguist’s notion of LF,
see May (1991).
Blackwell Publishers Ltd. 2001

Logical Form and the Vernacular 395
in context);
4
and it must entail things (and must itself be entailed) in virtue of
its structure.
5
These are necessary conditions. One further necessary condition,
for having a logical form, is that agents recognize these latter features (i.e. entail-
ment relations) in virtue of the item’s logico-structure. As we use this term,
the logical form just is that in virtue of which the item exhibits structural
entailment relations, and in virtue of which agents are able to recognize said
entailment relations. In brief, these necessary conditions for having a logical
form are jointly sufficient.
That said, (2) still may not be clear enough. So let us make a few remarks
about two of the key concepts that appear therein: namely, recognition and
structural entailment relations. By ‘recognize’, we mean a certain kind of
mental competence. Putting aside memory restrictions, life span, etc., our view
is that an agent’s mental competence can assign entailment relations to
sentences, propositions, etc. Indeed, a certain recognition-competence on the
part of agents is required for something to have a logical form. By contrast,
recognition-performance is unnecessary for a thing’s having a logical form.
That said, we do rely on an episodic sense of ‘recognize’when we treat recog-
nition as a sufficient condition for something’s having a logical form. The idea
is this: one cannot recognize—on this occasion, as a matter of performance—
␸’s logical form unless one’s competence (non-episodic sense) recognizes the log-
ico-structural entailment relations of ␸. Thus, one cannot occurrently assign a
logical form to ␸unless one has the competence to do so; hence occurrently
recognizing an item’s logical form is ipso facto asufficient condition for having
a logical form.
6
4
It may be objected that this condition rules out certain things that have logical form but which
are incapable of being either true or false, e.g. imperatives, interrogatives, etc. Although we
are only discussing the declarative mood in this paper, we can broaden the discussion thus:
␸has a logical form only if the sentence-radical part of ␸has satisfaction-conditions, in
roughly Tarski’s sense. That will allow for imperatives and yes-no interrogatives to have
logical forms, since their sentence-radicals stand for propositions, which have truth-conditions
and thus have satisfaction-conditions. It will also allow wh-interrogatives to have logical
forms since their sentence-radicals, though not capable of being true or false, will be satisfied
by sequences since they stand for a propositional function rather than for a proposition. For
instance, the sentence-radical of ‘Who bought the book?’is something like ‘[␭x:person x]
(xbought the book)’. For a defense of this view of wh-interrogatives, see Stainton (1999).
5
The emphasis on an expression’sstructural formal properties is important. Otherwise, we would
have to say, e.g. that it is part of the logical form of every truth-evaluable English sentence
that it entails every necessarily true English sentence. In particular, part of the logical form
of ‘Clinton had an affair’would be that it entail ‘Water is H
2
O’if the latter is a necessary
truth. Which is unacceptable. What’s more, on our view, it is no part of the logical form of
‘Water is H
2
O’, whatever logical form is, that the sentence is a necessary truth: if it is a
necessary truth, then it is so in virtue of its semantic content, semantic meaning, or modal
facts, but not in virtue of its logical form.
6
Readers who are skeptical of the competence/performance distinction may read our
recognition-condition thus: If an agent recognizes (in the performance sense) some of ␸’s
structural entailment relations, then he or she does so (in part) because of ␸’s logical form.
We take that conditional to express a minimally necessary condition for ␸’shaving a logical
Blackwell Publishers Ltd. 2001

396 R. Elugardo and R.J. Stainton
So much for recognition. Now, a not-so-brief note on structural entailment
relations. We use ‘structural entailments’to include both of what Gareth Evans
(1976) calls ‘structural semantic entailments’and ‘logical entailments’. For
instance, ‘John is a fat dentist’seems to entail, in virtue of its structure, ‘John
is fat’and ‘John is a dentist’. We are unsure whether this shows that ‘John is
a tall midget’and ‘John is a fat dentist’have a different structure. They may
or they may not. Still, both cases differ from cases of ‘logical entailments’, e.g.,
‘John is rich and 5 is odd’logically entails ‘John is rich’and ‘5isodd’.(We
suspect, in fact, that Evans’distinction between ‘structural semantic entail-
ments’and ‘logical entailments’is a matter of degree.) But what is common
in all three examples is that the entailments hold in virtue of certain structural
properties of the sentences. In contrast, ‘Shenae knows that fish swim’entails
‘Shenae believes that fish swim’but not in virtue of their structural properties.
We exclude cases of informal meaning-based entailments from our characteriz-
ation of logical form. Whenever we speak of ‘entailment relations’, this is
shorthand for relations of structural logical entailment.
If you supposed that only natural language expressions had structure, then
it would follow immediately that only they have logical forms, because only
they could have their structural entailment relations recognized. In this paper,
we don’t suppose that natural language expressions are the only things that
have structure of the desired kind. We take for granted, for instance, that
propositions and mental representations have constituents, and that they are
combined together in the right sort of ways.
Which leads to our next assumption, namely that there are logical forms.
This isn’t a very big assumption. After all, in general, it is not a brute fact
about an item that it has the entailment relations that it has. Nor is it a brute
fact that items possessing entailment relations are recognized as such. Which
is just to say that there is something—maybe some cluster of properties, maybe
some relations—in virtue of which the thing has its ‘entailment properties’,
and in virtue of which they are recognized.
7
Given this, we are pretty sure
form. It doesn’t follow, then, that for ␸to have a logical form, there must be, at some time
or other, someone or other, who actually recognizes (in the episodic, performance sense)
some of ␸’s structural entailments. Hence, our account of having a logical form is compatible
with the claim that some things have logical form even though it may never be the case
that someone or other recognizes (performance sense) its entailment relations. It’s also com-
patible with the claim that ␸may have a logical form and yet be so structurally complex
that no one could ever cognitively grasp ␸and, thereby, recognize (performance sense)
its entailments.
7
One caveat should be made. One might reasonably ask: ‘What if there is some kind of thing
for which it is a brute fact, about a thing of that kind, that it has the entailment relations it
does?’To accommodate such cases, if they exist, we will stipulate that if ␣’s having the
entailment relations it does is a brute fact, then the logical form of ␣just is being ␣. In such
a case, ‘that in virtue of which’␣has its entailment relations is . . . that it is ␣. Thus, to take
an example, suppose that the proposition THAT-CLINTON-IS-PRESIDENT is constituted
by its entailment relations. In that case, this proposition has its logical form not so much ‘in
virtue of’some cluster of properties/relations, but in virtue of being the proposition THAT-
Blackwell Publishers Ltd. 2001

Logical Form and the Vernacular 397
that logical forms in our sense exist. Still, pretty sure isn’t the same as mortally
certain. And, truth be told, we aren’t mortally certain that logical forms exist.
(Such are the dangers of broadly empirical inquiry.) The thing is, it’s possible
to doubt that the items which have semantic content—and hence stand in
entailment relations—are the very same things which have structure.
To note a couple of examples, if you thought that the only things which
could genuinely be attributed content are global states of persons, then, since
whole persons clearly don’t have compositional syntactic structure—what
would that mean?—you might deny that logical forms, as we’ve defined them,
really exist. (Taking this holistic line needn’t entail the radical claim that nothing
has syntactic structure. It could be that there are linguistic types which have
all the structure one could wish. But, one might still suppose, nothing which
strictu dictu has content has structure as well.) Or, to consider a rather more
plausible view, if you thought that only actions—or uses of sentences, or
maybe ‘what the speaker said’—really stood in entailment relations, you might
say: sentences have structure, but they don’t stand in entailment relations; and
actions have entailment relations, but they don’t have syntactic structure. Once
again, this would lead to the conclusion that there aren’t really any logical
forms in our sense. The reason, to repeat, is that it is a necessary condition
for having a logical form that the entity entail things in virtue of its structure.
So, if the only things that do the entailing don’t have structural properties, they
don’t have logical forms; and since, ex hypothesis, nothing else has entailment
properties, nothing else can have a logical form either. Logical forms, farewell.
At this point we could undertake a defense of the idea that there are things
which simultaneously have structure and stand in entailment relations. We
could do that. But we won’t. What we’ll do instead is just state that, in this
paper, we are assuming that this picture can be made to work. We note, as
well, that the Vernacularist shares this assumption with us. Any reader who,
unlike us, and unlike the Vernacularist, is convinced that nothing has both
structure and entailment relations might as well stop reading.
The next step, now that we’ve explained (2), is to distinguish items that
have logical forms ‘fundamentally’from those that have them only
‘derivatively’.
8
The notion is slippery: generally speaking, we think we know
CLINTON-IS-PRESIDENT. Also, we are willing to suppose that it’s conceivable that the
properties that underlie the having of entailment relations are different from those that under-
lie their recognition. But we put this possibility aside for present purposes.
8
We are actually somewhat suspicious of the whole ‘fundamental/derivative’distinction. So,
we are of course hard pressed to give an analysis of it. Nonetheless, there are cases where
it is more plausible that something has a logical form only derivatively. In his ‘The Thought’,
Frege gives an example of a painting that may be interpreted as expressing a Thought. The
idea is this: the observer must have a sentence provoked by the painting. She then grasps
the sense of this sentence and treats the painting as derivatively expressing this sense. It is
plausible, then, that paintings don’t have logical forms fundamentally, even though someone
could use a painting to ‘argue’for some claim. The reason is that the observer, who is
‘receiving’the argument, must first grasp something with a fundamental logical form. She
Blackwell Publishers Ltd. 2001

398 R. Elugardo and R.J. Stainton
it when we see it, but it isn’t easy to analyze. We can say this, however: if,
in order to recognize the entailment relations of ␣, an agent must find some
other item ␤that has logical form, then ␣does not have its logical form
fundamentally. Rather, ␣has its logical form derivatively from ␤. In contrast,
if a hearer can recognize the entailment relations of ␣without finding any
other logical form-bearing item ␤,then␣has its logical form non-derivatively.
Again, this isn’t intended as an analysis of ‘␣gets its logical form from ␤’. But,
analysis or no, the psychological test given is, we think, a sure sign of one
thing getting, or not getting, its logical form derivatively from another.
9
Given how we have characterized ‘having a logical form fundamentally’,
our burden is to establish two things. First, there are non-linguistic entities
that stand in entailment relations, such that agents recognize those entailment
relations in virtue of something about the non-linguistic entities. If we establish
this, we will thereby show that logical forms are assigned to things other than
expressions of natural language. Second, those non-linguistic things are not
understood via the recovery of some linguistic expression that has a logical
form. That is, we must make it plausible that a hearer can recognize the entail-
ment relations of such an item without having to find any natural language
expression that has that logical form. That would yield the conclusion that
things other than expressions of natural language have logical forms non-
derivatively.
10
1.2. Why Vernacularism is Attractive
Vernacularism undoubtedly has its adherents.
11
Some philosophers are attracted
to it because they consider propositions (and other abstracta) to be ‘creatures
is prompted to do this by the painting. But the painting doesn’t really have the logical form,
except in this watered-down, causal, sense. (The same may be said for other non-linguistic
tokens, e.g. mimed performances, sophisticated gestures.) Now the question is, do things
other than sentences have logical forms fundamentally? Or, is it the case that every instance
of a non-sentential logical form is basically like Frege’s painting example? In this paper, we
will defend an affirmative answer to the first question and a negative answer to the second.
9
It follows from the way we defined ‘having a logical form’that even things which have them
‘fundamentally’, have them in virtue of something. So having a logical form ‘fundamentally’
emphatically does not entail having it as a basic feature of the universe. That is not the
sense of ‘fundamental’at play.
10
Although we will later argue that some propositions have logical form non-derivatively, we
do not take ourselves to be arguing that propositions are the only things that have logical
form non-derivatively. We are ecumenical about the kinds of things that can have a logical
form fundamentally. As far as we argue in this paper, natural language sentences (types and
tokens), artificial formulae, propositions, and mental representations can all have non-deriva-
tive logical forms. Indeed, there is no reason why two different things, e.g. a proposition
and a sentence, can’t have the same logical form non-derivatively from each other and from
anything else.
11
We are hard pressed to find anyone who explicitly holds and defends Vernacularism in the
literature. Peter Geach and Wilfrid Sellars come close, cf. Geach (1957) and Sellars (1963).
According to them, the functional/conceptual role of a thought, which constitutes its struc-
tural formal properties, is modeled on the functional/conceptual role of its natural language
Blackwell Publishers Ltd. 2001

Logical Form and the Vernacular 399
of darkness’: propositions, it is said, have no clear identity-conditions, they
don’tfit nicely with a naturalistic view of the world, etc., cf. Quine (1979).
Sentences, on the other hand, are empirically accessible (at least their tokens
are), and they have somewhat clear identity-conditions. Besides, even philos-
ophers who are comfortable positing abstract propositions may hold that
human beings can only grasp propositions via sentences. Either way, goes this
line of thought, if anything has a logical form par excellence and fundamen-
tally, it’s not abstract propositions—it’s sentences. Of course this alone doesn’t
yield Vernacularism: one could feel an aversion to propositions, while taking
mental representations to be perfectly respectable, even as fundamental bearers
of logical form—as does Fodor (1975) and Fodor (1978). But some philo-
sophers are hostile both to abstract propositions and to inner mental states,
except in so far as the latter are understood in terms of, for instance, dispo-
sitions to utter public language sentences (Carnap 1967)—or possibly in terms
of ‘inner speech’(Geach 1957). If that is one’s view, it simply won’tdoto
countenance mental representations and/or propositions, which have logical
forms, such that no natural language sentence endows that logical form.
Hence Vernacularism.
Or again, suppose you find deflationary accounts of content plausible, cf.
Schiffer (1994). You might maintain, for instance, that contents are simply
linguistic posits, which enter into language by ‘something-from-nothing’
transformations of the following sort:
3. ‘Anita lives in Ottawa’⇒‘The proposition that Anita lives is Ottawa
is true’.
sentence-analogue. Since the functional/conceptual role of a thought/sentence will include
its inferential/entailment connections to other thoughts/sentences, an object-of-belief that
has the same functional/conceptual role as its counterpart-sentence (in a natural language
L) will also have the same logical form as its linguistic counterpart. On the other hand, it
is unclear whether Geach and Sellars also hold that, for each (possible) thought-type, there
must also be a unique natural language sentence-type from which it derives its logical form.
If they don’t hold this, then they are not committed to Vernacularism as we have defined
it. Some philosophers hold the strong conceptual thesis that having a natural language is
both logically necessary and sufficient for having any thoughts, e.g. Davidson (1975) and
McDowell (1994). Others hold the weaker, empirical, thesis that some thoughts are the
outputs of a biologically-determined language module whose inputs are natural language
lexical items, cf. Carruthers (1996) and Carruthers (1998). Neither thesis directly entails
Vernacularism, although Carruthers’view comes close. A closely related position is Jerrold
J. Katz’s‘Principle of Effability’:‘Each proposition (thought) is expressible by some sentence
in every natural language’(Katz 1981, p. 226). According to Katz, the principle implies that
‘the semantic structure of natural languages is complete with respect to the full range of
objects to which laws of logic apply and that the expressive structure is complete with
respect to the stock of senses semantic structure supplies’(Katz 1981, p. 239ff). That isn’t
quite the same as Vernacularism but it comes very close. In any case, we will argue below
that some philosophers tacitly hold Vernacularism, which is enough to take it seriously even
if no one ever explicitly endorsed it.
Blackwell Publishers Ltd. 2001

400 R. Elugardo and R.J. Stainton
4. ‘Anita believes that Chre
´tien smokes’⇒‘Anita believes the prop-
osition Chre
´tien smokes’.
Taking this line, you might well insist that propositions/belief-contents are
real enough, but maintain all the while that they pack no ontological punch.
Of course, if propositions/belief-contents are just ‘deflationary shadows of
sentences’, then surely there won’t be propositions (or contents) without a
corresponding sentence. And, once again, Vernacularism will strike one as
congenial.
12
In sum, certain philosophers will be tempted by the idea that, if abstract
propositions or inner mental states have logical form at all, we should under-
stand this in terms of the sentences that express them having that logical form.
Indeed, the story would go, we are only able to grasp ‘the logical form’of
the non-linguistic entities by finding an appropriate linguistic entity. So that
such non-linguistic things, if they have logical forms at all, do so only deriva-
tively. That just is Vernacularism.
Vernacularism may have its attractions, but it doesn’t come for free: it has
broadly empirical consequences that can lead one to doubt its ultimate plausi-
bility. Of crucial interest to us, Vernacularism rules out the possibility of
objects-of-belief (e.g., propositions meant by speakers), for which no precise
natural language encoding is found by the hearer. It is this aspect of Vernacular-
ism, i.e. the conflict between it and the existence of objects-of-belief that have
logical forms non-derivatively, that will be our primary concern.
1.3. The Game Plan (and some Underlying Issues)
We conclude this introductory section by sketching our ‘game plan’—and by
noting some issues that will lie just below the surface in the discussion ahead.
We hope this will clarify our motivation in discussing Vernacularism.
The most fundamental background concern will be the relationship
between thought and talk. We of course do not take what follows to establish
that, for example, thought is something other than inner speech. But we hope
that our discussion of Vernacularism contributes to on-going debates in this
area. Also closely tied to what follows are debates between proponents of the
indeterminacy of content and staunch realists about the mental and/or about
propositions: specifically, much of what follows is plausible only given the
assumption of content-realism. Some will take this as an argument, from the
best explanation, for content-realism: ‘Look what we can explain by positing
language-independent mental representations/propositions!’Others will com-
plain that a central question has been begged. Be that as it may, we want
explicitly to flag that such issues are at play. Equally at play, especially in the
12
Deflationary accounts of belief contents may be found in Barber (1997,1998), and Schiffer
(1994, 1996, 1997).
Blackwell Publishers Ltd. 2001

Logical Form and the Vernacular 401
final section, will be the distinction between perceiving an object or a property,
partly in virtue of its presence in the environment, and thinking about the
object/property via tokening some linguistic description of it. In part, we
maintain, Vernacularism fails because objects/properties can be thought about
in virtue of being in the environment, not solely in virtue of satisfying some
natural language description running through the mind of the thinker.
Curiously, this last point has a lot to do with the nature of ellipsis: specifically,
the contrast between drawing attention to things, versus implicitly/elliptically
drawing attention to words-for-things; and it thereby highlights issues about
the boundaries between syntax and semantics and between semantics and
pragmatics.
With those cryptic remarks, we now introduce the game plan. Our dis-
cussion of Vernacularism has two parts. First, we argue that Vernacularism is
not obviously true. That is, it isn’t at all clear that everything that has a logical
form either is a natural language sentence, or has its logical form derivatively
from some natural language sentence. We will argue that speakers can mean
things which have logical forms, but do not have those logical forms ‘endowed
upon’them by the recognition of a natural language expression. Second, we
support our claim that this is important by giving one example in which
Vernacularism is assumed, namely debates about Bertrand Russell’s Theory
of Descriptions.
2. Logical Form and Speaker’s Meaning
Our question can be divided into two sub-questions:
5. a. Are there things, other than natural language expressions, which
have logical forms?
b. Are there things of this kind which have their logical forms
fundamentally?
We are now going to mount an argument—not, admittedly, an airtight
demonstration—that one should answer ‘yes’to both questions. We will intro-
duce a linguistic phenomenon that shows that speakers mean, and hearers
recover, things that have logical forms, which are not themselves natural langu-
age expressions. That yields a ‘yes’to (5a). Our example is also one in which
such logical forms are not shared by any natural language sentence employed
by the hearer.
13
That yields a ‘yes’to (5b) as well. Given an affirmative answer
to both, Vernacularism is ipso facto false.
13
That word ‘available’will be important. To anticipate a little, we don’t mean to claim that
there is no possible human language in which a sentence, sharing the logical form of the
meant-proposition, exists. Who knows what riches possible languages might yield. All we
claim, and all we need to claim, is that there are cases in which propositions are meant,
and understood, without the psychological processing of a sentence that allows the agent
to assign to the proposition a logical form.
Blackwell Publishers Ltd. 2001

402 R. Elugardo and R.J. Stainton
The phenomenon we have in mind is one in which a speaker produces an
ordinary word or phrase, in an argument, and thereby communicates a full
proposition. Because the thing-meant is not itself a sentence, but is instead a
proposition, one recognizably used in an argument, this thing-meant forces a
positive answer to (5a), or so we will shortly argue. We then argue that the
hearer need not assign the logical form to the proposition-meant by recovering
some item of language which encodes the proposition. He can do so ‘directly’,
as it were.
There are multiple ways in which one could argue for this latter conclusion.
An experimental psychologist might do studies of reaction times, priming
effects, and so on, in cases of non-sentential speech; a (presumably future)
neurolinguist might bring to bear PET-scan studies of brain activity, clinical
case studies of bizarre aphasias, or who knows what. Either route might imag-
inably uncover evidence that logical forms are assigned, in cases of non-senten-
tial communication, without the aid of mediating full sentences. In a more
familiar vein, a corpus linguist might provide multiple examples of attested
speech, and argue that in many actually observed cases the hearer did not
recover any sentence. (Some initial steps within the latter approach were taken
in Stainton (1989/92) and Stainton (1990/91.) We would be keen to see the
results of such work. But our methodology will be more akin to that of gener-
ative grammar and philosophy of language. That is, we will imagine a dialogue
situation, and we will then mine intuitions about what, plausibly, is actually
going on in the imagined situation.
14
Our conclusion will be that the hearer
in the imagined situation does not recover a sentence when he understands the
non-sentential speaker. In so far as the imagined situation really is plausible,
it supports the conclusion that an agent can recover a thing-meant which has
a logical form, and which is not a sentence, such that the thing-meant has its
logical form fundamentally. (The transition is: what does happen in a possible
situation is something that can happen in the actual world.) It is this possibility
which makes (1) questionable. Since the argument hinges on the plausibility
of the situation described, we then consider many arguments designed to show
that our description is actually implausible.
2.1. The Non-Sentential Example
Here, then, is the imaginary case. A discussion has taken place between Alice
and Bruce in which Bruce takes the position that there are not really any
colored objects. A day or so later, Alice meets Bruce. Having just read G.E.
Moore, she offers the following counter-argument. She picks up a red pen,
and says ‘Red. Right?’. Bruce, guileless fellow that he is, happily agrees. Alice
14
We recognize that there is something odd about this methodology of imagining a case, and
then asking what is ‘really’happening in the case described. But it would take us too far
afield to justify the overall methodology.
Blackwell Publishers Ltd. 2001

Logical Form and the Vernacular 403
continues, ‘Red things are colored things. Right?’Bruce nods. At which point,
Alice springs her trap: ‘So Bruce, there is at least one colored thing. This thing.’
Here is what we wish to stress about the example. First, Alice is making
an argument: She communicates propositions, which are premises in the argu-
ment; and these premises do indeed have implications with respect to the
existence of colored things. And, of course, Bruce recognizes those impli-
cations. But premises that fit together into an argument, and are understood
to do so, cannot help but have logical forms. So, what Alice communicated
at each step had a logical form. (Whether she asserts the proposition, or merely
implicates it, is not important for present purposes. What is crucial is that
she communicates something that is correctly recognized to have structural
entailments.) Second, given the lapse of time, it will not be clear to Bruce or
to Alice what words were employed during their first exchange, although they
both recall the general topic of discussion. The particular words or phrases
that were produced will not be immediately available to them during their
second exchange. In which case, Bruce will not, during the second meeting,
employ those specific expressions in interpreting what Alice meant. Notice,
also, that all the evidence that Bruce actually uses is what we described. He
does not need to ask Alice any questions, access further evidence that would
be available to a third-party, etc. Nor does he need to wait until her argument
is done to know what she has communicated with ‘Red’. Finally, let us add
that Bruce cannot say ‘which natural language sentence he used’to understand
Alice. Plausibly, then, Bruce understands her just on the basis of her utterance
of the word ‘red’and the fact that she held up a red pen.
Our contention is that, in this imagined case, Bruce did not recover any
sentence in understanding Alice. What he did, instead, was to understand the
predicate ‘red’, and apply its meaning to the salient object, the pen in Alice’s
hand. Doing this, he came to grasp a proposition. And that proposition had
a logical form. If this description of what Bruce did is plausible, then it’s also
plausible that a real agent can in fact recover something that isn’t a natural
language expression, yet which has a logical form non-derivatively. Precisely
because, in the described possible situation, this is just what Bruce does.
2.2. Resisting the Non-Sentential Example
The ‘Red’example given above is highly controversial. The Vernacularist can
make many moves to diffuse its force. In this section we consider a host of
these, each of which ultimately amounts to a claim that our description of the
case is mistaken. Not to hold you in too much suspense: we find that none
of these means of resistance is very effective. To keep things as straight as
possible, we divide the objections into four main varieties: those which deny
that the thing-meant has a logical form; those which deny that the thing-
meant is something other than a linguistic expression; those which deny that
it has a logical form non-derivatively; and those which appeal to the primacy
of sentences. That is, we divide objectors into: one group which insists that
Blackwell Publishers Ltd. 2001

404 R. Elugardo and R.J. Stainton
(5b) can still be answered ‘no’; two sub-groups which insist that (5a) can still be
answered ‘no’; and those who, these questions aside, think there is something
fundamentally wrong with the example itself. The first three of these objec-
tions can be further sub-divided according to the following taxonomy:
OBJECTIONS TO THE IMAGINED CASE
(i)
The thing-meant
lacks a logical form
it just is a natural
language sentence
The thing-meant has
a logical form, but...
(ii)
an ordinary
sentence
an elliptical
sentence
(iIi)
from something
produced by Alice
from something not
produced by Alice
it has its logical
form derivatively
(iv)
syntactically
elliptical
(v)
semantically
elliptical
(vi)
a single
sentence
(vii)
a collection
of sentences
Objection (1): The Thing Recovered Just Is A Sentence.
One’sfirst reaction to the example might be that Bruce clearly did recover
an ordinary sentence when he understood what Alice meant. Indeed, it might
seem perfectly obvious what the sentence was: ‘This thing is red’. But why
include the word ‘thing’as opposed to ‘object’or ‘doohickey’? Is it so clear
that Bruce definitely recovered the former and did not recover either of the
latter? Fair enough, one might say. It’s not (6) but (7) that Bruce obviously
must have recovered:
6. This thing is red.
7. This is red.
But is (7) any better? As far as we can tell, nothing adequately justifies the
claim that Bruce obviously mentally tokened a bare English demonstrative
rather than a complex demonstrative or a definite description. For instance,
nothing rules out ‘The pen in my hand is red’as an equally good (or better,
equally bad) candidate. Moreover, why must the verb be ‘be’? Why not
‘instantiates’or ‘exemplifies’or ‘embodies’?
15
The basic difficulty is this: It
15
Robert May suggested to us that there could be minimal completions of non-sentential
expressions, and that this would justify choosing ‘This is red’. Applied to the example at
hand, the idea is that the minimal way of getting a sentence from [
AP
red] is (a) to embed
this phrase in a VP headed by a light verb, thereby creating, for example, [
VP
is [
AP
red]];
and (b) to concatenate the result with a minimal demonstrative, for instance [
DP
this]. On
this view, then, while the speaker produces a non-sentence, what the hearer recovers is
nevertheless a sentence. And it is this sentence which fundamentally has a logical form. We
Blackwell Publishers Ltd. 2001

Logical Form and the Vernacular 405
seems to us, as it would likely seem to anyone who wasn’t antecedently
assuming something like Vernacularism, that the proposition which Alice com-
municated is essentially less precise than any fully sentential paraphrase of it. Just
as none of ‘vermilion’,‘crimson’,‘maroon’or ‘scarlet’precisely captures the
meaning of ‘red’, so no available sentence precisely and uniquely expresses
what Alice meant. Hence it really is plausible that Bruce understood Alice
without recovering an ordinary sentence—because otherwise it should be easy
to say what sentence it was that Bruce recovered. Indeed, if Bruce had re-
covered a sentence, it would be easy for him to say which sentence it was.
And yet it surely cannot be denied that Bruce may be perfectly unable to pick
out ‘the’sentence he recovered.
Putting (ii) aside, then, the next natural thought is that Alice uttered, and
Bruce recovered, an elliptical sentence: not ‘This thing is red’or ‘This is red’
or ‘The pen in my hand is red’, but rather a non-ordinary sentence which is
pronounced red. This proposal can be understand in one of two ways:
8. Two Ellipsis Hypotheses.
(a) ‘Red’is syntactically a single word, but one which expresses
a proposition in a context. (That is, ‘Red’is a ‘one word
sentence’, semantically distinct from the ordinary word ‘red’:
whereas the latter is of type ⬍e,t⬎, the former is of type ⬍t⬎.)
(b) ‘Red’is not really a word at all. Rather, it is syntactically an
elliptical sentence—one which, like syntactic sentences
generally, expresses a proposition in context.
Option (8a), which Stainton (1995) has called ‘the Semantic Ellipsis
Hypothesis’, violates the semanticist’sversionofOccam’sRazor:Donot
multiply ambiguities beyond necessity. For, in order for the sound red to some-
times express a proposition in context, the sound type must have at least two
meanings, namely, the property RED (which it contributes, for example, in
‘That pen is red’) and the proposition that the contextually salient object is
red (which it purportedly contributes in the Alice-Bruce dispute above).
16
But
grant that this proposal is initially plausible for this case, taken in isolation. But we do not
believe that minimal completions are generally available. For instance, what would be the
minimal completion of ‘A pint of English bitter’, understood as uttered to a bartender? It
clearly isn’t‘This is a pint of English bitter’. But is it ‘I want a pint of English bitter’,‘Give
me a pint of English bitter’,‘I’d like a pint of English bitter’, or something else again?
Which of these is minimal?
16
We of course distinguish the case in which a speaker means that Pby using a sentence
which, even in the context, does not mean P, from the case in which a speaker means that
Pby using a sentence that does mean Pin the context. On our view, the ‘Red’example
is like the first case but is unlike the second. For, Alice did mean a proposition—but her
word, even given a context, does not semantically mean a proposition. In support of that
claim, we distinguish between type meaning, literal token meaning, and speaker meaning,
where literal token meaning is determined by type meaning plus some contextual elements
explicitly anticipated in the rules of use for the expression type. Given this three-fold distinc-
Blackwell Publishers Ltd. 2001

406 R. Elugardo and R.J. Stainton
the multiplication of meanings doesn’t stop there. For ‘Red’can also be used,
on its own, to communicate the proposition that red is a ‘warm’color. This
proposition has a different form from the one Alice communicated. After all,
Alice communicated a first-order proposition, predicating redness of an object,
while this is a second-order predication, applying WARM COLOR to red-
ness. ‘Red’can also be used to communicate an identity statement between
properties (e.g. Red =John’s favorite color), a universally quantified statement
(e.g. every pen hereabouts is red), etc. But no univocal expression can express
an identity statement, a quantified statement, a first and a second order predi-
cation, etc. (That is, no univocal expression can be type-synonymous with all
of: ‘red’,‘This is red’,‘Red has this’,‘Red is identical to this’, etc.) Hence,
if it were the one-word sentence ‘Red’which Bruce recovered, the sound red
would have to be multiply ambiguous.
Now, ambiguities do happen. But one shouldn’t posit them unnecessarily
especially when what’s at stake is the multiple ambiguity of essentially every
phrase in the language. And that is, indeed, at stake because, as Barton (1991)
argues, any phrase whatever can be used in isolation, to perform a speech act.
Consider ‘Another glass of that delicious German beer’,‘Nice car’,‘Three
letters from Spain’, and so on. However, the ambiguity posited here is
unnecessary, because pragmatics can easily fill the gap between the content of
the word uttered (i.e. a property) and the content of the thing communicated
(i.e. a proposition). After all, the speaker couldn’t possibly be communicating
a property—what would that amount to? So the hearer, to preserve the
assumption (in familiar Gricean fashion) that the speaker is cooperating, must
look for a proposition communicated. She could find such a proposition
directly, by applying redness to the salient pen.
So much, then, for the first ellipsis hypothesis. The second option can be
shown false on empirical grounds: ‘Red’is not overtly headed by an inflectional
element; nor is it the least bit plausible, given what linguists know about syn-
tactic ellipsis, that ‘Red’has a covert subject or a covert INFL node. In support
of this second claim, we note two other salient facts. First, syntactically elliptical
sentences cannot be used to initiate discourse, except under very special cir-
cumstances. For instance, one cannot, without awkwardness, walk into a room
and say ‘Alex does too’. In contrast, ‘Red’and other bare phrases can occur
as freely in discourse initial position as non-elliptical sentences: the discourse
initial use of ‘Red’is no more restricted than the use of ‘This is red’;in
contrast, its discourse initial use is much less restricted than the use of ‘This
does too’. Also, syntactic sentences (including syntactically elliptical sentences)
can license sluicing and other genuinely elliptical constructions in ways that a
tion, the question is whether the type meaning of ‘Red’, is the same as the type meaning
of ‘This is red’. If it is not, then the literal meaning of utterances of the two, holding context
fixed, cannot be the same.
Blackwell Publishers Ltd. 2001

Logical Form and the Vernacular 407
non-sentence (like ‘Red’) cannot. Consider in this regard the contrast between
the (A) and (B) discourses:
Sluicing post-sentence and post-non-sentence
(A) Peter: Who is at the door?
Ernie: The man from Paris.
Rob: I wonder why.
(B) Peter: The man from Paris!
Rob: *I wonder why.
In (A), the sound the man from Paris is plausibly paired with a syntactic sentence,
occurring as it does in reply to a wh-interrogative. That is, though Ernie
produces only the sound the man from Paris, the sentence he produced could
well be (something like) [
IP
[
DP
The man from Paris][
I’
is at the door]]. And
here, ‘I wonder why’is fine. By contrast, in (B), the sound the man from Paris
is plausibly taken to correspond simply to the phrase [
DP
The man from Paris].
The sluicing construction is not licensed here precisely because, in the (B)
case, what was produced was not a sentence—not even an elliptical sentence—
and sluicing demands a sentence.
In sum, ‘Red’is not an elliptical sentence, at least not in any sense that
lends comfort to Vernacularism.
17
Any feeling that ‘Red’must be elliptical
derives, we suspect, from the indisputable fact that what the speaker means
by it is a proposition. But given the distinction between semantics and prag-
matics, this by no means establishes that ‘Red’has the syntax or the semantics
of, for example, ‘This object is red’. In which case, ‘the elliptical sentence’
cannot serve as the linguistic expression that Bruce recovered.
Objection (2): There is a Logical Form, but it is Derivative
So far we have responded to objections (ii), (iv) and (v) in the previously
noted taxonomy. We turn now to a series objections according to which the
thing-meant, which Bruce recovers, is not itself a sentence, but it nevertheless
‘gets its logical form’from a sentence. Assuming the logical form of what Alice
meant is derivative on a natural language expression, there are only two options
for the ‘source’expression, recovered by Bruce. Either it is some expression
that Alice did not produce, or it is an expression that she did produce.
We can rapidly deal with the proposal that Bruce recovers a proposition,
but assigns it a logical form by means of (also) recovering the linguistic
expression that Alice produced. This could only work if Alice produced some-
thing which encoded a proposition. That is, if she produced something which
17
This isn’t the first paper to make hay of non-sentential speech. For more on the pragmatics
of non-sentential speech, see Stainton (1994). On semantic ellipsis and syntactic ellipsis, see
Stainton (1995) and Stainton (1997) respectively. A general overview, and an application
to quantifier phrases, can be found in Stainton (1998).
Blackwell Publishers Ltd. 2001

408 R. Elugardo and R.J. Stainton
was not an ordinary word or phrase, but was instead an elliptical sentence.
But that hypothesis immediately runs into the just noted objections.
Well, but, isn’t there one more sense in which the logical form could derive
from ‘the thing used by Alice’? Here is what we have in mind. It’s quite clear
that there is such a thing as Alice’sspeech act. What’s more, at least in a broad
sense, that speech act is something linguistic. And, it has exactly the same
content as what Alice meant. So why can’t the logical form of the thing-meant
come from it? True enough, Vernacularism as stated would strictly speaking
be false, since there would still be no natural language expression from which
the proposition-meant gets its logical form. Nevertheless, use of language
would still be fundamental in accounts of logical form. And that is surely the
deep issue here.
The point is well taken. But we think the following reply is compelling.
In order for an entity ␤to be that from which some item ␣derives its logical
form, ␤must itself have a logical form. After all, the agent assigns a logical
form to ␣by recognizing the logical form of ␤. However, to have a logical
form, an entity must have constituent structure. That was part of the definition.
And—here comes the punch line—speech acts do not have constituent
structure. Of course the thing uttered, in the speech act, clearly does have syntac-
tic structure. But the act itself does not. Unless, of course, what is meant by
‘the syntax of the speech act’is just, ‘the syntax of the thing uttered’. Then
one could speak of the syntactic structure of the speech act. The problem is,
we have already established that the expression uttered cannot be what gives
the proposition-meant its logical form, because the thing uttered is a ‘plain-
old word’. So, on this understanding, ‘the structure of the speech act’can-
not do the job either. In sum: in one sense, speech acts don’thavestructural
entailments at all; in another they do, but they are inherited directly from
the structure of the expression used. Either way, the speech act cannot be
that from which the thing-meant derives its logical form. We move, then,
to the hypothesis that the logical form is derivative on something not used
by Alice.
We have already noticed one reason why we need not take the logical form
to be derivative on any one specific sentence of English: each such sentence is
too precise to be the one which Bruce ‘obviously used’. Alice did not mean
‘This pen is red’as opposed to ‘This thing is red’, any more than she ‘really’
meant vermillion by ‘red’. One might say, in reply: There is a sentence all
right—it’s just hard to figure out which one it is. But, once again, if Vernacu-
larism were true, why should it be hard? For what Alice meant must ‘get its
logical form’from this unique sentence in the minimal sense: Bruce is pur-
portedly only able to recognize the entailment relations etc. of the thing meant,
by finding a sentence which expresses the thing meant, and then recognizing
the structural entailment relations of that sentence. He does not assign a logical
form directly to the proposition-meant. Taken this way, propositions-meant
do not have logical forms fundamentally. But how can one recognize the
Blackwell Publishers Ltd. 2001

Logical Form and the Vernacular 409
proposition-meant in this manner, and yet find it hard to state the specific
sentence which one supposedly had to recover in order to find the logical
form?
18
Why, when questioned, doesn’t Bruce know that he emphatically did
not recover the sentence ‘This pen is red’, but instead recovered ‘This is red’?
Why is he any less certain than if Alice had simply produced a full sentence?
After all, in each case, Bruce supposedly recovers the sentence. The right thing
to say, in answer to all of these rhetorical questions, is that the proposition
has a logical form that we recognize, and that it does not get this logical
form derivatively.
Nor, turning to (vii), will it do to maintain that several sentences taken
collectively determine the logical form of the thing-meant. For this gets the
explanatory order wrong. Surely it’s because of the nature of the proposition
communicated that all of (9) count as close paraphrases, and not vice versa.
9. a. This pen is red.
b. This thing in my hand is red.
c. This is red.
d. The pen here is red.
To give an analogy, there are two ways of understanding the location of the
bull’s-eye vis a
`vis an arrow. One might say, ‘The bull’s-eye is located just
here because this arrow came very close, that arrow missed by a few inches,
and that arrow was nowhere near’. One might also say: ‘This arrow came very
close because the bull’s-eye is just here, and the arrow struck just there’.In
the case of the bull’s-eye, it’s clear that the former gets the order of explanation
wrong: the bull’s-eye does not come to have its location because such-and-
such an arrow missed. Now consider: in virtue of what is ‘This thing is red’
a quite good paraphrase of what Alice meant, whereas ‘This doohickey exhibits
red’isn’t very good, and ‘My plane is late’is nowhere near? Is ‘the target’
determined by which things come close and which are far away? Or is the
closeness of the paraphrase determined by where the target is? We take the
latter line. (And, we believe, only someone antecedently in the grip of Ver-
nacularism would insist otherwise.) In which case, the paraphrases are not prior
to—do not determine—the thing-meant.
Objection (3): There is No Logical Form Bearing Thing-Meant
We have argued that the thing-meant is not itself a sentence, and that it does
18
It could be that the explanation is that the recognition/recovery process is a subconcious
process. Maybe so. But Vernacularists cannot appeal to such an explanation in response to
our argument. For, they restrict the relevant mental states involved in communication to
public-language things, e.g. dispositions to utter, affirm, deny, etc., public natural language
sentences. On their view, positing tacit mental states, which cannot be analyzed as public
linguistic dispositions, is out of the question.
Blackwell Publishers Ltd. 2001

410 R. Elugardo and R.J. Stainton
not have its logical form derivatively. If it has one at all, it has it fundamentally.
At this point, one might think that it was a mistake to have agreed that there
is a proposition-meant that has a logical form. Maybe the thing-meant is not
even propositional. Or, it is a proposition but it is too vague to have a logical
form. The phenomenon, so construed, would not be: ‘There is definite prop-
osition for which we cannot find a sentence available to the hearer’.Onthe
contrary, there is no proposition-meant to be ‘captured’at all. We have arrived
at objection (i).
Of course there certainly is the illusion of the thing-meant being a
proposition with a logical form. Without question there is an argument here,
whose premises fit together in a way that supports the conclusion. The expla-
nation of this illusion might run as follows. There are many things that corre-
spond to the thing-meant, each of which has a logical form. For instance, all
of the sentences in the cluster (9), noted above, have logical forms; and all of
them are rough paraphrases of what was meant; moreover, any number of
them could combine with the sentence ‘Red things are colored things’, ther-
eby supporting something like Alice’s conclusion. (Whether the resulting
reconstructed argument was valid solely in virtue of form would depend upon
which of ‘This thing is red’,‘This instantiates red’, etc. is selected.) In sum,
there is something meant by Alice, and there is at least one thing that has a
logical form, but they aren’t the same thing.
How to respond? Well, there are things which quite precisely capture the
logical form of the thing-meant. None, however, is an English sentence plausi-
bly recovered by Bruce. Here is one example, which is an expression in an
artificial logical language:
10. Red(x
3
).
(10) is an open sentence. It is the sort of thing that combines with either a
bare quantifier (e.g. ‘[∀x]’) or a quantifier phrase (e.g. ‘[∃x: pen(x)]’)toform
a closed sentence. When the variable is free, however, we stipulate that it
refers to the third element of the given variable assignment. Spelling this out
a little, we might stipulate the following sequence as a variable assignment:
11. ⬍The moon, Noam Chomsky, the demonstrated pen, Jerry Fodor,
the sun, . . .⬎
(To simplify, let’s agree that ‘how the object is presented’, i.e. its mode of
presentation, is unimportant here.) Given this assignment, (10) expresses pre-
cisely what Alice meant: the open sentence ‘Red(x
3
)’is true relative to this
sequence if and only if the third element of that sequence is red. Given
sequence (11), the whole is indeed true, because the demonstrated pen is red;
(10), given, (11), pretty clearly has a logical form. So, if the logical form of
(10)-given-(11) is the same as that of the proposition which Alice meant, then
Blackwell Publishers Ltd. 2001

Logical Form and the Vernacular 411
what Alice meant has a logical form. The generalization here is: if xhas the
same logical form as ydoes, and yhas a logical form, then patently xhas a
logical form. (It’s also worth noting that, taken in abstraction from any
sequence, ‘Red(x
3
)’doesn’t have the same logical form as the proposition-
meant, since the latter has a truth-value (which is, of course, context-
independent), while the former does not. But given a sequence, they share a
logical form. That said, (10)-given-(11) has a logical form; and it has the same
logical form as the proposition-meant. So the proposition-meant has a logi-
cal form.
19
)
The objection, recall, went like this. It was a mistake to have agreed that
there is a proposition-meant that has a logical form. To which we replied:
There must be a proposition-meant with a logical form because (10)-given-
(11) has the same logical form as what Alice meant. Two points about our
response should be emphasized. First, treating (10)-under-(11) as sharing the
logical form of the thing-meant does not immediately grant victory to
Vernacularism because (10) is not a natural language sentence. Second, quite
obviously, if a thing’slogical form can be accurately described, then it must have
a logical form. What we’ve just done, by introducing ‘red(x
3
)’, is describe
what Alice meant, including describing its logical form. Thus, there is some-
thing which Alice meant and it has a logical form.
One might worry that the existence of (10) makes our rejection of
Vernacularism less interesting. After all, the really interesting claim is that
objects-of-belief have their logical forms fundamentally. But, one might say,
the proposition-meant by Alice—which, notice, is the object of various atti-
tudes by both speaker and hearer—can have its logical form ‘endowed’by
(10). Hence, the object-of-belief still ends up having its logical form deriva-
tively—though admittedly, not derivatively on a natural language sentence.
We believe, however, that this worry can be easily diffused. The key move
goes like this: Bruce, we can safely agree, does not know the artificial logical
language that we just introduced. He doesn’t know about free variables,
sequences, etc. Hence, his grasp of the proposition-meant, and his apprehen-
sion of its logical form, cannot be via his recognition that, for example, the
proposition-meant shares the same logical form as (10)-under-assignment-(11).
In which case, it cannot be that (10)-given-(11) endows the proposition-meant
with its logical form. In brief, for the proposition-meant to ‘get its logical
form’from something else, it must be something available to speaker and hearer
19
We would like to prevent a possible misunderstanding of our argument. We have argued
that if there is such a thing as a logical formula that has the same logical form as x, then
surely xhas a logical form. Our conclusion is not based on the assumption that the logical
formula is the logical form of x. If it were based on that, then we might be accused of
conflating the logical form of xand the logical formula used to represent x’s meaning. Our
argument does not, however, rest on any such confusion. Rather, it proceeds from the valid
general principle that if yhas the same ␾as z, and if yhas a ␾, then zhas a ␾.
Blackwell Publishers Ltd. 2001

412 R. Elugardo and R.J. Stainton
which can do the job. Our reply to the worry, then, is that (10) simply is not
available to Bruce.
20
A rather different worry about our response in this section is the following.
It might be suggested that appealing to (10), in describing the logical form of
the proposition that Alice meant by ‘Red’, leads to a pragmatic paradox of
some sort. It better be possible to explain, in natural language, what (10) means.
For instance, one needs to describe the artificial logical language, and how
variable assignment works for that language, etc. Yet if one can do that, runs
the objection, then there are English sentences that express what Alice meant.
And maybe they are the ones used by Bruce. In a nutshell, either one can say,
in English, what Alice meant or one can’t. If one can, then Vernacularism is
true. If one cannot, then there is no reason to believe that a proposition really
was meant. Our rejoinder is that there is an equivocation on ‘there are English
sentences that express what Alice meant’. We are not claiming that it’s imposs-
ible to describe what Alice meant, using English. Indeed, we have been at pains
to characterize what she did (and did not) mean. Nor do we maintain that
Bruce would be at a loss to describe, in complete sentences, what he under-
stood. Our contention, rather, is that there is no unique sentence of English
which Bruce used to understand Alice. That is, there is no available sentence
of English that is a ‘direct translation’of her thought—though what she meant
can be described by us, and by Bruce.
21
In contrast, the content of (10) does
capture, given the sequence, just what was communicated: i.e. given the
sequence, (10) encodes the proposition that Alice communicated. In which case,
what Alice meant has a logical form, and, as we argued above, it is one that
cannot be assigned derivatively in terms of the logical form of an available
natural language expression.
20
Another closely related clarification should be made. We don’t wish to commit ourselves
to the (very strong) claim that no sentence of any natural language could encode what Alice
meant (cf. Sperber and Wilson, 1981). Maybe there are sentences of Russian, or of Urdu,
or of who-knows-what, which do express the very proposition-meant. But even if that
were the case, it wouldn’t establish that the thing-meant gets its logical form, derivatively,
from a natural language sentence. For it cannot be the case that monolingual speakers of
English, who perfectly well understand the proposition communicated, do so by recovering
anon-English natural language sentence, finding its logical form, and then assigning that
logical form to the proposition meant. Rather, monolingual Anglophones must grasp the
logical form of the proposition-meant, without grasping any natural language sentence shar-
ing that logical form. That is, here again, there is no available natural language sentence to
endow the proposition-meant with its logical form. Which surely means that the proposition
must have a logical form, independently of any sentence whatever. (The first worry, about
whether an artificial language sentence might endow the logical form, was raised by David
Anderson. The second worry, about whether a sentence in some other natural language
might share the logical form, was noted by Ernest Lepore and Kirk Ludwig. We would
like to thank all three for raising these points.)
21
See Bar-On, 1994, for a related distinction applied to ‘There is a translation’. She notes
that ‘There is no English translation for sentence S’does not entail ‘We cannot describe
S’s meaning in English’.
Blackwell Publishers Ltd. 2001

Logical Form and the Vernacular 413
Objection (4): The ‘Context Principle’Objection
We have sometimes heard the following sort of complaint about the ‘Red’
example:
As Frege stressed long ago, only sentences have meaning in isolation.
Later, Dummett and Wittgenstein rightly added that only sentences can
be used to make a move in the language game.
22
Given this, there must
be something amiss with the Red example.
What this complaint highlights is that there seems to be a tension between
our example and both Frege’s (1884) notorious ‘Context Principle’and the
Wittgenstein-Dummett dictum about use. Given that both are very widely
assumed, we had better say something about how these claims intersect with
our views.
The first point we need to make is that there are at least three ways of
understanding ‘sentence’. These are:
12. Three Senses of ‘Sentence’
a. Sentence
syntactic
: An expression with a certain structure/form,
b. Sentence
semantic
: An expression with a certain content/meaning,
c. Sentence
pragmatic
: An expression with a certain use.
An example of a syntactic characterization of sentence can be found in Chomsky’s
Government and Binding Theory. There, a sentence
syntactic
just is a (maximal
projection) phrase structure whose grammatical head is an INFL node—where
INFL nodes contain tense (e.g., present, past, future) and agreement (i.e., per-
son, gender, number) markers. A more traditional syntactic characterization
might be: ‘A sentence
syntactic
has a subject, verb and (optional) object’. What
both of these have in common should be clear enough: they categorize
expressions by the kinds of syntactic elements that they are built from.
What about (12b)? Here, one might say: ‘A sentence
semantic
is an element
which expresses a proposition’. Of course this will need to be refined, to cover
sentences which are context sensitive. Better would be: ‘A sentence
semantic
is
an expression, tokens of which, once reference has been assigned to all context
sensitive elements, express propositions’. For present purposes it is enough to
say, in the notation of Montague grammar, that sentences, in this semantic
sense, are of type ⬍t⬎. (Notice that this divide, between syntactic-sentences
and semantic-sentences, closely parallels that between the two kinds of ellipsis
discussed above. This is very far from a coincidence.)
Finally, one might divide up expressions into those which can be, and those
22
See Wittgenstein (1953) and Dummett (1981 and 1973, p. 194).
Blackwell Publishers Ltd. 2001

414 R. Elugardo and R.J. Stainton
which cannot be, used on their own (i.e., unembedded in any larger structure)
to perform a speech act. The former would then be called sentences
pragmatic
,
while the latter would be non-sentences
pragmatic
.
Given this three-fold division, let us first address the Wittgenstein-
Dummett dictum. It’s perfectly true that only sentences
pragmatic
can be used in
isolation. Indeed, it’s tautological. So, read as a claim about sentences
pragmatic
,
the dictum is on solid ground. We are not therefore insisting otherwise.
Instead, what we’re committed to is that non-sentences
syntactic
and non-sen-
tences
semantic
can be used, unembedded, to perform speech acts. And, of course,
it would be a fallacy of equivocation to infer from ‘Sentences
pragmatic
are the
only things which can be used and understood in isolation’to the conclusion
that only sentences in either of the other two senses can be so used.
Our view, put another way, is that ordinary words and phrases, with both
the form and content of words and phrases, are sentences
pragmatic
. In particular,
words and phrases can be used on their own to communicate thoughts, includ-
ing premises in arguments. Since words and phrases just are sentences
pragmatic
,
nothing we say conflicts with the truism that only sentences
pragmatic
can be
used. But, crucially, we deny that this shows that the things communicated
are ‘really’sentences
syntactic
. To imagine otherwise would be to mix up issues
about the content that the speaker conveyed with issues about linguistic form.
(It would, in fact, be on a par with a use-mention confusion: Alice meant p,
therefore Alice meant ‘p’.) Less obviously, that words and phrases can be used
to communicate thoughts does not show that their tokens encode the thoughts
thereby communicated. Especially when pragmatic mechanisms are indepen-
dently available to bridge the gap between what the word/phrase means, in
context, and what its speaker meant. (Compare: one can mean
NN
that Mr. X
is an awful student, by tokening ‘Mr. X has neat handwriting and usually
arrives on time for class’; but this clearly doesn’t show that one—tacitly, ellipti-
cally, under one’s breath as it were—produced an expression whose meaning
in the language is MR. X IS AN AWFUL STUDENT.) Once the distinction
in (12) is carefully made, then, we see no particular danger from the truism
that ‘only sentences may be used to make a move in the language game’.We
accept this claim, on its truistic reading; nevertheless, we reject it as false, on
empirical grounds, when read in terms of sentences
syntactic
or sentences
semantic
.
But, the objector will insist, isn’t there something right about Frege’s Con-
text Principle? We concede that there is: To find what a word or phrase means,
one must ask what the word’s content ‘joins with’in order to yield a Gedanke
(thought). For example, to find the meaning of ‘seven dogs’, it is a mistake
to look either for an associated mental image or for an object. Instead, the
meaning of ‘seven dogs’must be understood in terms of how its content com-
bines with a predicate-meaning to yield a Gedanke. Talking this way does make
Gedanke ‘primary’, in some sense. (Which, we suspect, is all Frege himself
really cared about.) So be it. We are perfectly happy with that. What it doesn’t
do is guarantee that only certain syntactic forms (e.g. Inflectional Phrases) or
Blackwell Publishers Ltd. 2001

Logical Form and the Vernacular 415
linguistic items bearing a certain kind of content (e.g. items of semantic type
⬍t⬎), can be used in conversation. Nor does it guarantee—far from it!—
that only these linguistic items have logical form fundamentally. In which case,
properly construed—i.e. construed as a claim about the role of Gedanke in
determining word-meaning, rather than as a claim about sentences—there
likely is no conflict between our view and Frege’s Context Principle. But, so
read, the Context Principle does not support Vernacularism.
We can sum up Part 2 as follows: If propositions-meant have logical forms,
and if those logical forms are not derivative on the logical forms of natural
language expressions, then Vernacularism is false. What we have tried to do
to render this conditional of non-trivial interest. The conditional claim would
be of only trivial interest if its antecedent were obviously false. But, we have
suggested, given the use of ordinary words to communicate propositions, it is
quite plausible that there are things-meant, which have logical forms, but
which are not linguistic items. And it’s also plausible that those things-meant
have their logical forms fundamentally.
3. Vernacularism Assumed: Incompleteness and Russell’s
Theory of Descriptions
Recall the central aim of this paper: to argue that (1) below isn’t obviously
true, and to suggest why this is important.
1. Vernacularism: The view that logical forms are fundamentally assigned
to expressions of natural language, and are only derivatively assigned
to anything else: e.g. propositions, mental states, etc.
Part 1 of the paper was dedicated to explaining what Vernacularism amounts
to. Part 2 argued that it isn’t obviously true. In this final part, we will give one
example—namely, recent discussions of Russell’s Theory of Descriptions—of
why this result is important. Our aim is to give an example in which Vernacu-
larism plays an important role in certain discussions in the philosophy of langu-
age. The example will therefore help support our claim that Vernacularism is
a significant philosophical view.
The example serves other purposes as well, however. First off, one might
reasonably complain that it remains obscure just how a speaker manages to
communicate a full-fledged proposition, with a logical form, on the basis of
uttering a mere word or phrase. Specifically, one might wonder how Alice
could have communicated a proposition, about the pen, using the word ‘red’.
Our response, in a nutshell, is this: a speaker may draw the hearer’s attention
to an object, ‘get the hearer to think about that object’, and then produce a
word whose content can be predicated of the object. That, we say, is how
Alice achieved her impressive result. A further role of the following example,
then, is to clarify the nature of this kind of conversational move, and to drive
Blackwell Publishers Ltd. 2001

416 R. Elugardo and R.J. Stainton
home its ubiquity. In particular, we will come to see a clear difference between
recovering a word and noticing a thing. But this time it will be in the context
of much more familiar philosophical terrain. Here, then, is the example.
A certain criticism of Russell’s Theory of Description assumes Vernacular-
ism. It goes like this. Suppose that Jones points to a photograph of Frank
Sinatra and assertively utters (13), to Smith:
13. The man is a great singer.
(We ignore, for present purposes, the fact that Sinatra is dead.) According to
Russell’s (1905) Theory of Descriptions, (13) is logically equivalent to (14):
14. There is exactly one man, and every man is a great singer.
This latter sentence entails, by conjunction simplification, that there is exactly
one man. Since this is patently false, it seems that Jones’utterance of (13) must
itself be false—completely independently of Sinatra’s vocal prowess. Hence
Russell’s Theory of Descriptions is reduced to absurdity. Call this criticism,
‘the Original Reductio’(Strawson (1950)).
A natural defense of Russell’s theory is to contend that Jones’utterance is
semantically equivalent to a completion of (15):
15. There is exactly one man who is H, and every man who is His a
great singer.
The predicate, represented by the schema who is Hin (15), is said to be
fixed by the context in which the incomplete description, ‘the man’, is uttered.
In our example, the instance of who is H, is some salient, contextually rel-
evant predicate that is uniquely true of Sinatra—a predicate that Jones
implicitly meant when he uttered (13). (For example, the instance of who is
Hmight be ‘who is pictured in that photograph’, where Jones points to
Sinatra’s photograph.) Taking ‘who is pictured in that photograph’as the
intended completing description, (13) is said to be true exactly if:
16. There is exactly one man who is pictured in that photograph and every
man who is pictured in that photograph is a great singer.
By hypothesis, Frank Sinatra is the only man in the only indicated photograph.
So, (13) is predicted to be true if Sinatra is a great singer. In brief, if (13) is
properly translated as (15), and if (15) is completed as in (16), then Jones’s
utterance can come out true. The Original Reductio is diffused.
Some philosophers remain unconvinced and have presented what we’ll call
‘The Many Descriptions Objection’to show why this defense of Russell’s
theory fails, cf. Wettstein (1981). Relative to the context in which Jones
Blackwell Publishers Ltd. 2001

Logical Form and the Vernacular 417
uttered ‘The man is a great singer’, there are many equally salient descriptions
which are uniquely true of Sinatra and which are mutually known to be so
by Jones and Smith. Here is a partial list:
17. Salient Descriptions of Sinatra
a. The skinny crooner who was the idol of many bobby-soxers dur-
ing the early 1940s.
b. The guy who headed Hollywood’s notorious ‘Rat Pack’during
the early 1960s.
c. The actor who won an Oscar for his supporting role in From Here
to Eternity.
None of the listed descriptions of Sinatra is sufficiently salient for Smith to
identify it as the one that Jones tacitly meant by his utterance of (13). Nothing
about the context fixes for the hearer, among these many equally good candi-
date-descriptions, ‘the one’that Jones had in mind at the time of the utterance.
In fact, it isn’t really necessary that there be a particular one that Jones meant
(tacitly or otherwise). Thus even Jones himself might be at loss to specify ‘the’
description of Sinatra he meant. (For more on this, see Bach (1987), Elugardo
(1997), Neale (1990), Reimer (1992), and Schiffer (1995).)
Suppose, then—to mount The Revised Reductio—that Vernacularism and
the neo-Russellian view of incomplete definite descriptions are correct. Then,
Jones will have meant a proposition that has a logical form only if he meant
a proposition whose logical form is derivative on the logical form of a specific
natural language sentence, a sentence whose (potential) utterance is a com-
pletion of (15):
15. There is exactly one man who is H, and every man who is Hwas
a great singer.
However, that general condition will not be met because, by hypothesis, there
is no unique, contextually salient, completing description of Sinatra that Jones
had in mind when he uttered (13). So runs the Many Descriptions Objection.
Put otherwise, without a specific instance to take the place of who is H
there is no sentence available from which to derive a logical form for the
specific proposition asserted. Without that, there can be no proposition which
is meant and which also has a logical form. (This last step is the one that
employs Vernacularism.) Ergo, on this neo-Russellian line, if Vernacularism
is true, then no logical-form-bearing proposition is meant. But that is absurd:
patently, Jones did mean a proposition, and it did have a logical form. In
particular, he communicated a proposition whose truth depends upon the
vocal powers of Frank Sinatra. Furthermore, he meant something which, in
this context, entitles Smith to use as a premise in an inference that Frank
Sinatra was a vocalist; in which case, Jones meant something which has a logical
Blackwell Publishers Ltd. 2001

418 R. Elugardo and R.J. Stainton
form. So Russell’s view is reduced to absurdity once again if Vernacularism
is true.
The neo-Russellian might respond to the Many Descriptions Objection by
saying that Jones didn’t really mean any particular proposition. But this really
isn’t plausible. What gives the objection its force is the hard-to-deny intuition
that Jones meant just one thing by his utterance of ‘The man was a great
singer’. The idea that he vaguely meant indefinitely many, extensionally equiv-
alent, descriptive propositions by his utterance is a bit too much to accept. If
anything, that would give one some reason to think that maybe Jones really
didn’t mean anything at all by his utterance, but certainly that is unacceptable.
Given that, one might try to save Russell, while holding onto Vernacular-
ism, by saying that Jones didn’t assert any proposition made-true by Sinatra,
though Jones did mean something by his utterance. What he asserted, runs
this defense, was a metalinguistic existential proposition–something like:
18. There exists some description dsuch that when dis substituted for
the variable in ‘There is exactly one man who is H, and every man
who is Hwas a great singer’the result is a true sentence.
But this won’t do either. For if that was what Jones really said, it is hard to
see how he managed to convey a proposition whose truth-maker is Sinatra and
his singing abilities—something he unquestionably managed to do. (18) is far
too weak if dis in no way circumscribed. The reason being that, after all,
every adult human male is such that there is a description of him which, when
substituted into (15), yields a truth! In which case, (18) clearly will not allow
the hearer to draw any reasonable inference about what sort of adult male is
being discussed. In particular, it won’t allow the hearer to figure out that Jones
is referring to Sinatra. On the other hand, once one tries to circumscribe d,
the previously noted problem arises in a slightly different guise: nothing about
the context fixes, among the many candidate circumscriptions-of-d, the one that
Jones had in mind at the time of the utterance.
The upshot so far is this. The neo-Russellian view apparently requires that
Jones and his interlocutor have a description in mind. Without that, either there
is no sentence from which the proposition-meant can get its logical form—
which would entail, given Vernacularism, that there is no logical-form-bearing
proposition at all—or the sentence is merely metalinguistically existential, as
in (18)—which leaves the hearer at a loss to figure out what Jones is on about.
But there’s a nagging problem: though the Russellian approach demands one,
there actually is no such description.
Our real point, in the end, is that The Revised Reductio of Russell’s
Theory presupposes Vernacularism. The reductio implicitly contains the premise
that the lack of a predicate, to be substituted for who is H, entails that there
cannot be an appropriate proposition meant. But if propositions can have
logical forms, without there being a sentence that ‘provides’or ‘endows’that
Blackwell Publishers Ltd. 2001

Logical Form and the Vernacular 419
logical form, then Jones could mean such a proposition, even though no sen-
tence is available to him or to his hearer, which expresses the proposition.
Roughly put, Jones could intend a restricted set of adult males, and Smith
could guess what set Jones intended, without Smith or Jones fixing on an
English predicate that encodes the restriction.
We still have not explained precisely how this might go, however. Nor is
this an idle issue. For, it might reasonably be said, if arguments against Russell’s
Theory of Descriptions can be straightforwardly reconstructed, without appeal
to a many descriptions objection, then it won’t be obvious why it matters so
much whether Vernacularism is true. Our aim at this point, then, to put it
another way, is to show that Vernacularism isn’t just a suppressed premise in
the argument discussed, but an important (i.e. not easily dispensable) sup-
pressed premise.
We will therefore attempt to say rather more about how neo-Russellians
might defend themselves. We want to stress, however, that in discussing this,
we don’t wish to commit ourselves to either view.
The first step, in answering the proposed reductio, would be to reconstruct
the defense of Russell as follows. What (13) amounts to is not (15), but
simply (14):
13. The man is a great singer.
14. There is exactly one man, and every man is a great singer.
15. There is exactly one man who is H, and every man who is His a
great singer.
However, despite initial appearances, literal utterances of (15) can easily be
true. The reason is, whenever (14) is uttered, there is an intended domain of
quantification. And, sometimes, that domain of quantification contains just one
man.
23
The same will therefore hold, given Russell’s Theory, for (13).
Crucially, however, the intended domain of quantification can be determinate
even if there is no salient predicate that has the domain as its extension.
Specifically, the domain of quantification can itself be salient—because it is
(or its elements are) easily perceptible in the environment. (Compare a lesson
we take from Gareth Evans: someone can see an object in the distance, without
tokening a natural language description of the object—not even ‘object which
I see in the distance’. What makes the person’s thought be about that object?
Not, says Evans, some purported linguistic description of it; instead, it is
because that is the object she is perceiving.) Looked at this way, a set itself—
rather than a linguistic description of it—is salient in the environment because,
for example, it (or its elements) is perceptible. Taking this set to be the domain
23
We stipulate, for the purposes of simplifying discussion, that it is Sinatra himself, and not
just the photograph of Sinatra, who is salient.
Blackwell Publishers Ltd. 2001

420 R. Elugardo and R.J. Stainton
of quantification, (14) can turn out true. So, one cannot then use the ‘obvious
falsehood of (14)’to produce a reductio of Russell’s Theory.
Here is another way of making the same point—a way which helps to
further clarify the difference between recovering a word and noticing a thing.
Stephen Neale distinguishes an ‘explicit’and an ‘implicit’approach to in-
complete quantifiers.
24
He writes:
There are two main approaches to incompleteness in the literature, what
we might call the explicit and the implicit approaches. According to the
explicit approach, incomplete quantifiers are elliptical for proper quantifi-
ers. As Sellars puts it, the descriptive content is ‘completed’by context.
According to the implicit approach, the context of utterance delimits the
domain of quantification and leaves the descriptive content untouched.
Consider [the sentence ‘Everybody was sick’] again. On the explicit
approach, the quantifier ‘everybody’(as it is used on this occasion) is
elliptical for ‘everybody at the dinner party I had last night’, or some such
‘narrower’quantifier. On the implicit approach, the domain of quantifi-
cation is understood as restricted to some favored class of individuals (or
to some favored part of the world) (Neale 1990, p. 95).
The explicit approach really does demand, in order to ‘narrow down’adefinite
description like ‘the man’, that some completing linguistic description be found:
e.g. the phrase ‘in that photograph’. That is, the explicit approach is as much
committed to Vernacularism as the critic who puts forward the Many Descrip-
tions Objection. But the implicit approach is crucially different.
25
(And, in
fact, it’s a close cousin of our story about how ordinary words and phrases are
understood in conversation.) Assuming Vernacularism isn’t true—so it is
possible to contextually restrict the domain of quantification by means of
something other than a description—the implicit approach remains on solid
footing, even if no unique description can be found.
We can draw two lessons from this entire discussion. First, once Vernacular-
ism is explicitly noticed, and once its falsehood becomes a real possibility, extra
elbow room opens up in the philosophy of language. Options not previously
explored suggest themselves. In the case at hand, the idea that domain of
quantification can be grasped without the hearer settling on a description
makes the Russellian committed to less than what one might suppose. Specifi-
24
This topic is revisited in Neale (2000), which responds to Stanley and Szabo (2000). See
also Bach (2000).
25
This highlights the fact, stressed by Reimer (1998, p. 106ff ), that the two approaches are
not ‘notational variants’of one another. To put it in terms introduced above: whereas the
explicit approach involves ellipsis, or anyway the recovery of the part of the hearer of some
unpronounced linguistic item, the implicit approach requires instead that the hearer notice
a salient non-linguistic something.
Blackwell Publishers Ltd. 2001

Logical Form and the Vernacular 421
cally, she needn’t be committed to the implausible claim that there is always
a single salient linguistic description, to complete the description used—as
some critics have supposed. This is one respect, then, in which it is important
that Vernacularism may not actually be true. (What’s more, this is far from an
isolated example of Vernacularism being assumed by some, but not all, on
both sides in the debate.
26
But space doesn’t permit describing others.)
The second lesson is less direct. It is reasonable to wonder just how logical
form bearing non-linguistic things can be communicated, by using plain old
words and phrases. Reflection on incomplete descriptions provides rather more
than a clue, within a much more familiar domain. In discussing Russell’s
Theory of Descriptions, we saw how a set of objects can be salient, so that
this set can be the domain of quantification. Such saliency did not require that
the speaker intend, or the hearer recover, a linguistic description of the set.
27
In much the same way, a single object can be salient, not because the speaker
covertly, elliptically, etc., produced a description of it, and not because the
hearer recovered some description of it, whether used or not , but because
that object is presently in the environment of both the speaker and hearer. Once
the object itself is salient, there is nothing to stop the hearer from grasping a
proposition with a logical form: a proposition that contains the object as one
constituent and, say, the property red as the other, where the property-element
of the proposition is brought to her attention by a speaker’s production of an
ordinary word or phrase that expresses that property. This, we believe, is
precisely how non-sentential communication works.
To conclude: we have explained what Vernacularism says, and we have
argued, on the basis of non-sentential speech within arguments, that it may
well be false. The reason provided was that speakers can communicate pro-
positions, which can be used as premises, by producing ordinary words and
phrases. What they thereby communicate are not themselves linguistic items;
but they nevertheless have logical forms. We further argued that, in at least
one case (and, we believe, in very many more), it is implausible to suppose
that the thing-meant gets assigned a logical form on the basis of recovering a
linguistic item that encodes the thing-meant. So not only are there non-
linguistic things which have logical forms, but some of them have their logical
26
Clearly, Russellians who adopt the explicit approach in their semantic analysis of incomplete
definite descriptions assume Vernacularism. Interestingly, some critics of that approach also
assume Vernacularism, e.g. Schiffer (1995) and Wettstein (1981). On their view, the speaker
of a referential utterance of a sentence of the form, ‘the Fis G’, where ‘the F’is an
incomplete definite description, means an object-dependent proposition, one whose logical
form is derivative upon the logical form of the very sentence that the speaker uttered—it
is just that, on their view, the logical form of a referentially used descriptive sentence is
that of an indexical sentence. Other critics of the explicit approach reject that version of
Vernacularism as well, e.g. Recanati (1997), Sperber and Wilson (1981, pp. 191–193).
27
In Reimer (1998), Marga Reimer argues that contextually salient properties can delimit the
intended domain of quantification for referential utterances of the form, ‘The Fis G’.
Blackwell Publishers Ltd. 2001

422 R. Elugardo and R.J. Stainton
forms fundamentally. Or so it seems. We ended by looking at a case, dis-
cussions of incompleteness and Russell’s Theory of Descriptions, in which
both parties in the dispute had assumed Vernacularism. It became clear that,
once Vernacularism was rejected, the Russellian could tell a more plausible
story. One reason why the falsehood of Vernacularism is important is precisely
that it is, here and elsewhere, assumed as a suppressed premise: a suppressed
premise, indeed, which can blind one to more plausible treatments. Our overall
conclusion, then, is that Vernacularism may well be false, and that it’s rather
important if it is.
Department of Philosophy
University of Oklahoma
Department of Philosophy
Carleton University, Ottawa
References
Bach, K. 1987: Thought and Reference. Oxford University Press.
Bach, K. 2000: Quantification, qualification and context: a reply to Stanley and Szabo.
Mind & Language 15, 262–283.
Barber, A. 1998: The pleonasticity of talk about concepts. Philosophical Studies, 89,
56–86.
Barber, A. 1997: Deflated concepts: a reply to Stainton. Critica, 29, 83–105.
Bar-On, D. 1994: Conceptual relativism and translation. In G. Preyer, et al. (eds.),
Language, Mind and Epistemology. Dordrecht: Kluwers, 145–170.
Barton, E. 1991: Nonsentential constituents and theories of phrase structure. In K.
Leffel and D. Bouchard (eds.), Views on Phrase Structure. Dordrecht: Kluwer.
Carnap, R. 1967: Meaning and Necessity. Chicago: University of Chicago Press.
Carruthers, P. 1998: Thinking in language? Evolution and a modularist possibility. In
P. Carruthers and J. Boucher (eds.), Language and Thought: Interdisciplinary Themes.
Cambridge University Press, 94–119.
Carruthers, P. 1996: Language, Thought, and Consciousness: An Essay in Philosophical
Psychology. Cambridge University Press.
Davidson, D. 1975: Thought and talk. In S. Guttenplan (ed.), Mind and Language.
Oxford University Press. Reprinted in Inquiries into Truth & Interpretation. Oxford
University Press, 1984, 155–170.
Dummett, M. 1981: The Interpretation of Frege’s Philosophy. London: Duckworth.
Dummett, M. 1973: Frege: Philosophy of Language.Cambridge,MA:Harvard
University Press.
Elugardo, R. 1997: Descriptions, indexicals and speaker meaning. Protosociology, 10,
155–189.
Blackwell Publishers Ltd. 2001

Logical Form and the Vernacular 423
Evans, G. 1976: Semantic structure and logical form. In G. Evans and J. McDowell
(eds.), Truth and Meaning: Essays in Semantics. Oxford University Press, 199–222.
Fodor, J. 1978: Propositional attitudes. The Monist, 61, 501–523.
Fodor, J. 1975: The Language of Thought. New York: Crowell.
Frege, G. 1884: The Foundations of Arithmetic. Trans. J.L. Austin. Oxford: Blackwell.
Geach, P. 1957: Mental Acts. London: Routledge and Kegan Paul.
Grice, H.P. 1989: Studies in the Ways of Words. Cambridge, MA: Harvard
University Press.
Katz, J.J. 1981: Language and Other Abstract Objects. Oxford: Blackwell.
May, R. 1991: Linguistic theory and the naturalist approach to semantics. In D.J.
Napoli and J. Kegl (eds.), Bridges Between Psychology and Linguistics: A Swarthmore
Festschrisft for Lila Gleitman. Hillsdale: Lawrence Erlbaum Associates, 269–288.
McDowell, J. 1994: Mind and World. Cambridge, MA: Harvard University Press.
Moravcsik, J.M. 1998: Meaning, Creativity, and the Partial Inscrutability of the Human
Mind. Stanford: CSLI Publications.
Neale, S. 1990: Descriptions. Cambridge, MA: MIT Press.
Neale, S. 2000: On being explicit: comments on Stanley and Szabo, and on Bach.
Mind & Language, 15, 284–294.
Quine, W.V.O. 1979: Intensions revisited. In P. French, et al. (eds.), Contemporary
Perspectives in the Philosophy of Language. Minneapolis: University of Minnesota
Press, 268–274.
Recanati, F. 1997: Direct Reference: From Language to Thought. Oxford: Blackwell.
Reimer, M. 1998: Quantification and context. Linguistics and Philosophy, 21, 95–115.
Reimer, M. 1992: Incomplete descriptions. Erkenntnis, 37, 347–363.
Russell, B. 1905: On denoting. Mind, 14, 479–493.
Schiffer, S. 1997: Meanings and concepts. New York University, ms.
Schiffer, S. 1996: Language-created, language independent entities. Philosophical Topics,
24, 149–167.
Schiffer, S. 1995: Descriptions, indexicals, and belief reports: some dilemmas (but not
the ones you expect). Mind, 104, 107–131.
Schiffer, S. 1994: A paradox of meaning. Nous, 28, 279–324.
Sellars, W. 1963: Some reflections on language games. In his Science, Perception, and
Reality. London: Routledge and Kegan Paul, 321–358.
Sperber, D. and Wilson, D. 1998: The mapping between the mental and the public
lexicon. In P. Carruthers and J. Boucher (eds.), Language and Thought: Interdisciplin-
ary Themes. Cambridge University Press, 184–200.
Sperber, D. and Wilson, D. 1986: Relevance: Communication and Cognition. Cambridge,
MA: Harvard University Press.
Stainton, R.J. 1999: Interrogatives and sets of answers. Forthcoming in Critica.
Stainton, R.J. 1998: Quantifier phrases, meaningfulness ‘in isolation’and ellipsis.
Linguistics and Philosophy, 21, 311–340.
Stainton, R.J. 1997: Utterance meaning and syntactic ellipsis. Pragmatics and Cognition,
5, 49–76.
Blackwell Publishers Ltd. 2001

424 R. Elugardo and R.J. Stainton
Stainton, R.J. 1995: Non-sentential assertions and semantic ellipsis. Linguistics and
Philosophy, 18, 281–296.
Stainton, R.J. 1994: Using non-sentences: an application of relevance theory. Pragmatics
and Cognition, 2(2), 269–284.
Stainton, R.J. 1989/92: Discourse in a Bilingual Setting: Working Papers at LE CAMP.
Two volumes. Toronto: Applied Linguistics Research Working Group (ALRWG),
Glendon College of York University.
Stainton, R.J. 1990/91: A LE CAMP Sourcebook. Two volumes. Toronto: Applied
Linguistics Research Working Group (ALRWG), Glendon College of York
University.
Stanley, J. and Szabo, Z.G. 2000: On quantifier domain restriction. Mind & Language,
15, 219–261.
Strawson, P. 1950: On referring. Mind, 59, 320–344.
Wettstein, H. 1981: Demonstrative reference and definite descriptions. Philosophical
Studies, 40, 241–257.
Wittgenstein, L. 1953: Philosophical Investigations. Oxford: Basil Blackwell.
Blackwell Publishers Ltd. 2001