Professional Documents
Culture Documents
Here I speak with an object (the ring) to speak about that object. I transmit information by inducing a
perception identifying what I verbally refer to. This simple speech act is a paradigm of something we
do or observe most every day from infancy on, something natural and foundational. We do this as well
when the object we talk with is within the sentence referring to the object, as when we say:
(3) O is a circle
Sentence-embedded objects like those underlined above we'll call displays. The ring and comparable
No one thinks the prop-ring is a word or term. It's an object appropriated by an utterance of (1),
not a syntactic component of the sentence. It's a referent of a term in (1) we'll call a director. The
director, this figure, refers to the prop-ring. The prop does not denote or mean anything at all.
Historically, despite the evident semantic equivalence of uttering (2) and uttering (1) with its
prop, for some 2500 years, linguists and logicians, East and West, from Panini on, with rare exception,
have considered displays some sort of word, a kind of term that refers to itself (or its kind), something
properly called an autonym (Carnap (1937)). Autonymic conceptions of displays are diverse, but the
2
core idea is that, semantically and syntactically, displays are self-referential syntactic components.
After Tarski's (1932) formalization of a metalanguage, autonymic conceptions have been challenged by
Autonymic conceptions have yet to justify positing display self-reference when what can be said
with a display might be said with a prop, a referent that doesn’t refer. Paratactic conceptions have yet to
justify disregarding the evident name-like syntax of displays that enables display sentences to say what
This essay's Part I (Fonseca v. Autonyms & Parataxis) explains an historical compromise
conception. Semantically, the d3isplays of (2) and (3) are referents like the prop-ring of (1); they don't
denote or predicate. Syntactically, they are name-like sentence components whose position enables
their prop-like semantics. Displays and props are presentations of an object, not representations of
themselves. We present an object to an audience to induce their perception of it and direct their
attention to it. Its presentation refers them to it without its representing or referring. Unlike props,
displays are presented syntactically by positioning them where a term for the object would denote it.
something said), and display marking from quotation marking, and explains how these distinctions
structure the explanation of the semantics and syntax of sentences like (5) and (6) wherein quotations
are displayed.
Part III (Display Formalization) explains the misconception of displays in Alfred Tarski's
(1932) formalization of a metalanguage by stipulating that flanking an expression with a pair of lines
3
creates a new expression denoting the flanked expression (or its kind). We'll call this new expression an
intranym because it refers to its interior, unlike an autonym that refers to itself. Tarski's formalization
represents displays with intranyms, and misrepresents the form and content of our natural
metalanguage.
2. Framing the Project: The guiding premise of this project is that sometimes what is said with a
display may be said with a prop, and sometimes it can’t. The semantic (propositional, informational)
content of utterances pairs like (1) and (2) may be the same.
Of course their semantic content differs if they are read as referring to distinct concrete particulars, but
they may also be read as referring to the same property, a kind of shape. In this respect, (1) is like (2b)
and (2r):
So it seems that the location of the circle inside or outside the sentence has no more semantic import
However, many display embedding utterances have no prop-using utterance counterparts. One
example is (4): His name is Nick. A starker example: an utterance of (4A) (Nick is his name) is not well
matched by pointing at a prop Nick while saying is his name. Attaching a prop to an uttered sentence
fragment doesn’t make a grammatically whole sentence. We’ll later see other display utterances, like
4
(6)(Bertie thought we are “but brief meat and brains”), that can’t be said with props.
In essence, this project aims to explain such facts, and some of the implications of that
explanation. That explanation is shaped by the assumption that these matters have a history. My hunch
is that our use of displays is best understood as developing from our use of props. Our use and
understanding of props seem basically pre-linguistic: we couldn’t acquire our first language if we
didn’t naturally understand props. Displays are sentence components, creatures of language that come
later. (Nonhuman animal speech does not evidence utterances embedding displays.) Variations in
display syntax seem to be later developments; some of them seem due to peculiarities of modern
3. Presentation v. Representation: When I hand you a ring while saying sentence (1) the ring
is a prop; the indexical descriptor, this figure, is a director. This director refers to its prop (or a property
of it). My handing you the prop also refers you to it. I use it to induce a perception of it, to make it an
object of your attention and thought. That presentation identifies the director's extension perceptually
while the director identifies the presented prop verbally. You witness my showing you what I'm talking
about.
Representations of a thing direct our attention to it in its absence. Propositional representations rest
upon presentations. Since naming requires identification of the referent, fixing the referents of our
names must begin with a preverbal way of identifying things. Names are first learned by coordinate
presentations of two perceptual objects, a name (Mama, dog, water, red) and its referent (an individual,
property, relation, event, location, etc.) The name's referent is identified by a presentation inducing a
5
perception of it.
Presentations take many forms. Phonemes may be presented by a parent or babbling baby. A
prop may be presented by producing it, pointing to it, posting words on it, or passively exploiting
circumstances priming the audience to fix on the prop when presented its name or a verbal reference to
it. Properties, types, kinds and other abstracta may be presented by presenting their perceptible
instances, for they are identified by directing attention at their instances. So too, perceptible individuals
Yet, many things are not presentable. Many terms cannot be taught ostensively -- and most that
can aren't. Eventually, most of our words are learned purely verbally, with words explaining other
words.
Still, the importance of presentation is gauged by its primacy, not its relative frequency.
However wide and tight our net of verbal definitions, none of our words represent anything unless
some of their reference is communally coordinated by joint perceptions of a name presented in the
presence of its referent (Davidson (1973), Engelland (2014)). We can refer to things, perceptible and
imperceptible, past, present or future, in actual and imaginable worlds, because we can refer with some
perceptible objects in our world. Language gets to be about the world by incorporating the world within
The role of presentations in our acquisition of terms and our knowledge of their referents is both
causal and epistemic. Presentations are our ultimate evidence of our (imperfectly) shared
understanding of our language and our world. Whatever way we learned a word, our best test of our
referring to the same perceptual thing is our shared reference to it in its presence. Such verifications of
6
word-object correlations don't silence all possible interpretive skepticism. Nothing could do that. But
normally, given the usual speaker-audience psychological similarities, presentations provide reason
I may introduce you to a thing or its name by showing you thing and saying its name. I may
introduce to you to a property or predicate by using the same object and the same words. Which I am
doing may depend on the context. Often, as in this essay, I do both at once. (What could be wrong with
my wanting my readers to take my term introductions both ways?) We may first define a term by
presenting both the term and its referent. With definitions established, we may present an object and
represent with our terms. I may point at your shirt while saying: My kitchen is that color. By identifying
the verbal referent as the perceptual referent the content of your understanding is enhanced.
So is your cognitive standing. Perceptions of an object are evidence of the object’s existence
and perceptual properties. Presentational reference is defeasible evidence for its intensional contents.
Your acquaintance with a referent informs your references to it, and authorizes them. You now know
what you are talking about. Representational reference has no such epistemic import. Someone’s
affirming a description may be evidence of its denotation, the description itself is not. Special
circumstances aside, a name is no evidence of any of its referent's properties or its existence.
4. Props: Props are extra-sentential objects used like an appositive term, for they are
primarily identifiers of the subject their director identifies. But, appositive terms are alternative
representations of the subject. Props are not representations. They don’t “mention” anything. They
identify the director's denotation by being it, or instancing it, or exemplifying it, or replicating it -- and
by their being perceived as doing it. They are constituents of speech acts that contribute to the act’s
7
informational content, but not to the sentence’s semantic content. They contribute information that is
Sometimes the same proposition is expressed with a replicant object within a sentence.
In essentials, prop presentation is biological communication. It takes many forms, some of them
linguistically sophisticated, but generally they are best conceived as basically a kind of pre-linguistic
demonstrative. Pointing is an apt paradigm. We naturally use and read gestural pointers as directors of
While the core of prop presentation is pre-conventional, it mixes with representational and
conventional elements when, for example, an object is presented by presenting a picture of it, or, more
importantly, when the object presented is itself a conventional, representational object, like a map,
diagram, word, numeral, or sentence. Conventions and representations operate in the presentation and
the perception of representational objects, for we commonly present such objects to be not merely
(semantic) properties. The graphemes (types and tokens), Nick, Nick and Nick are three versions of
the same name. Their material doesn’t matter: such objects may be embodied in graphite, wood, plastic,
pixels, etc. Even their perceptual resemblance is inessential: they don't resemble the phoneme, Nick,
that they represent by doing visually what it does audially. The various graphemes and phonemes are
not different names of the person. Culture may make perceptually diverse objects identify the same
name by making them be the same name identifying the same person. We are taught to see some shapes
Our perceptions are culturally conditioned. But the core of prop presentations is a fact of our
nature: We are referred to things by their presentation. We read perceptions induced by presentations
accompanying an utterance about the perceptual object. We read them identifying the verbal referent.
Presentations have that effect on us and our intensionality. If we didn’t naturally read presentations we
5. Display Conceptions & History: Displays are like props fitted to function in a syntactic
structure. Props acquire their semantic role by some extra-syntactic, semantic event like pointing.
Nothing extra-syntactic directs attention to a display. Instead, displays are presented syntactically by
being positioned where a genuine name for the director's extension would denote that extension.
[W]hether you write the following whole [sentence], This figure O is a circle, or only this,
O is a circle, in neither writing will that O be the subject of the proposition, understood in the
Fonseca's discussion is short and skeletal. The nub is that the displayed O is a pseudo- or quasi-term
that only seems self-referential in (3) (O is a circle). Actually (3) is short for (2) (This figure O is a
circle), whose display is the semantic equivalent of the prop of (1) (This figure is a circle). In (1)-(3),
Theorists don't criticize Fonseca's proposal. It doesn't occur to them. Somehow, this elegant
insight of the once famed “Portuguese Aristotle” has gone little known and never appreciated for near
five centuries.1 Logicians and linguistic theorists haven't denied that (a) sentences using props always
1 I chanced upon my predecessor briefly described and dismissed in Ashworth (1974). John Doyle
generously further educated me about Fonseca and provided translations.
9
have a director, (b) sentences using displays very commonly have an explicit director, (c) an utterance
presenting a solo display (= no evident director) says the same as some utterance with an explicit
director, and (d) some term must be operating when a display operates. Increasingly over the last eight
decades, theorists, unaware of Fonseca, have observed that a display is semantically prop-like, not
term-like. More recently a few theorists began attending to the operations of a director. (Cf. Jackendoff
Some theorists see that a display is not a term semantically, but they don't confront the puzzles
of how a non-linguistic object could be a syntactic component, and how it could be referenced without
some term referring to it. (Cf. Reach (1938), Whiteley (1957), Christensen (1967), Searle (1969)
Binkley (1970), Clark & Gerrig (1990), Read (1997), Recanati (2001) de Brabanter (2005).) Like their
autonymic foils, they (almost) never mention directors and never recognize their operation.
Davidson (1979) recognized that displays aren't terms. And he noticed the need for a term. But,
he thought that need filled by the Tarskian (1932) assignment of an essential semantic role for
quotation marks flanking an object. Davidson took the marks to be a single, universal demonstrative
director: this expression (more fully: the expression type of this expression token) . Sellars (1950) had
seen that expressions can be individuated both functionally and materially, so an additional director in
Conceptions like these have been called demonstrative because of their (oversimplified but
rectifiable) semantic thesis that displays are referents of demonstratives (as though no other term could
serve).They are better called paratactic conceptions because their distinctive thesis is the syntactic
claim that displays are "paratactic" referents: i.e., not syntactic components of a sentence. Paratactic
10
conceptions consider a display a referent of a term within the matrix sentence by presuming the
propriety of Tarski's formalized metalanguage. The assumption is that our natural language display
marks are the equivalent of the Tarskian marks stipulated to create a referring expression, an intranym,
a referential device designating its innards. That stipulation invites -- but doesn't demand -- a paratactic
construal.
However construed, while Tarski's referential marks are perceptually like the punctuation
natural languages now often use to mark displays (and to mark an odd lot of other things we’ll discuss
later), our display markers don’t create a display or any referring expression. They are disambiguators,
helpful interpretive aids, but semantically and grammatically inessential -- like capitalized initial
sentence letters. We couldn't create or understand display punctuation unless we already used and
metalanguage.
For natural languages, what has seemed obvious (to minds uncorrupted by formalist
conceptions) is that displays are autonyms that don't need directors. Compare (4) and (7):
(7) He is Nick.
These are apparently comparable, complete grammatical sentences. (7) has two terms denoting the
same person, Nick. Surely, (4) must have two terms denoting the same name, Nick. Some term other
than His name must be denoting Nick, -- and what else could it be but the displayed Nick? The idea that
(4)'s Nick names the name, Nick , just as (7)'s Nick names the person, Nick, is altogether natural -- and
so compelling that even displays of nonlinguistic objects, like the circles in (2) and (3), look like a term
11
Actually, like props, displays are presented objects. Unlike props, they appear to represent
themselves. That appearance is a halo effect of incorporating a prop within a syntactic structure. The
integrity of a sentence allows props to become sentence components only if they are like the matrix
words materially, perceptually, replicationally, and like names syntactically. So they look like names
semantically.
can be virtually any sentence-independent object (event, property, location, etc.) perceptible by the
speaker and audience concurrent with the presentation of a sentence, because a vocal utterance is an
action readily coordinated with actions directing attention to such objects. Written sentences are
primarily products intended to persist apart from their production and detached from the objects
currently perceptible to a reader, so their props are generally limited to colored shapes on the same or a
facing page, or to things adjacent to signs about them: Slippery when wet.
presenting the sentence. Displays needn't be words or linguistic objects but they must be materially and
Compatibility with the embedding sentence's material and perceptual properties is a vague
requirement subject to cultural conditioning. Our conception of the displayable gets challenged by
children's books, advertising, art, and innovations in sentence embodiment. Some cases are clear; some
are not. C-sharp is displayable in monotonic sentences, but not written ones. Color patches are displays
in otherwise monochromatic sentences; they are props when waved while voicing the sentence. Some
12
purists might insist that a fabric swatch inside a greeting card sentence is a prop, not a display; few
Tarski (1965) predicted that if the sentence this stone is blue has the words this stone replaced
with a small blue stone, we would not regard the result as a sentence. Most folks, I'd bet, would read the
whole as a kosher sentence when is blue is written in similar small stones -- and when such a stone-
(Language has lots more aspects than formalist logicians attend to.)
The physical limits of display are relative to a linguistic medium. Displays can be any material
matrix words can be. Smells, tastes, tickles and pains can't be words of audial or visual sentences, so
they aren't displayable there; they can only be props for oral sentences. Gestures can be displays in
sentences of ASL but only props or pointers presenting a prop of oral sentences. Morse code can
(A topic for another time: What can be presented in Mentalese, the Language of Thought? If
conditioned by the requirement that displays be reproducible along with the reproduction of their
matrix sentence. Props are sentence-independent concrete particulars that needn't be replicable.
Displays recur with the recurrence of their sentence, and as a component of semantically related
(8) Every figure O is a circle, and every circle is two-dimensional, so every figure O is
two-dimensional
13
(9) This figure O is a circle and every circle is two-dimensional, so this figure O is
two- dimensional.
Parallel arguments using props are equivocations unless a single concrete prop is presented twice. (8)
and (9) would be equivocations if their display tokens were their material (ink, pixels, whatever), but
we read the recurring displays like recurring terms. Words and displays can be individuated by their
matter, but far more commonly and naturally we individuate them by their perceptual, positional and
functional properties. A display’s material resembles the matrix material, whatever that be; beyond that,
Generally, sentence reproductions (including here transcriptions and translations) may vary
widely in font, color, size, style, medium, etc. without affecting sentence meaning. Generally, displays
(and especially displays of matrix language words) are reproduced consistent with the reproductions of
the matrix words. Thus, the standard reproduction of (2) (This figure O is a circle) for a page of red
While they can be read the same, we naturally read (2c)-(2e) to say something else.
Consistency of a display's perceptual properties with those of the matrix is the norm. Those are default
properties of the display. They have no informational import. What attracts attention is the perceptual
14
peculiarities and divergences from the matrix. The color and style of the (2)/(2r)/(2b) displays are those
of their matrix words. We take the display (the sentence subject) of (2)/(2r)/(2b) to be simply the shape;
only the shape has any functional relevance or informational import. The displays of (2c) and (2d) are
conspicuously colored; (2e)'s display is conspicuously sized. We naturally read the (2c)/(2d) displays --
and the subjects of the sentence -- to be specifically a black circular shape. We take (2c) and (2e) to be
semantically the same, and unlike (2), (2b), (2r) and (2e).
perceptual properties, and also to the perceptual properties of the sentence context. (2e) is ambiguous.
We'd normally take its display to be, specifically, a circular shape of that size. That leaves open whether
(2d)'s display is an instance of an absolute size replicated in (2b) in a page of 18 point print, or instead
an instance of a size relative to its matrix, unlike displays in all the other sentences. Consider also that a
(2c) in a page of red print might be replicated by a (2) in a page of black print.
A display's perceptual consistency with its matrix and context may be necessary for replicating
sentence meaning and truth. Some perceptual consistency may instead be inconsistent with semantic
replication. Consistency of font in (2f) alters the display's shape, and must be sacrificed for semantic
consistency in (2g):
Some of the other variants of (2) may differ referentially, but they are all true, whereas (2f) is false.
Or so we might say in many contexts. But if I free hand write (2f), my audience might well
charitably see my display as a circle, just because I am saying that it is a circle. We do much the same
15
when, as is the norm, we charitably read all the other above variants as true. Those other seven displays
much more closely approximate a geometer's circle than (2f)’s display does.They may be close enough
to be seen as circles in many circumstances, yet a geometry text might be faulted for affirming any of
those sentences.
So, display tokens of the semantically same type in synonymous matrices may differ, in kind or
degree, materially or perceptually to adjust for material or perceptual differences in their matrix or its
context, or for the constant matrix content. So too, materially and perceptually identical objects may be
display tokens differing semantically because of perceptual or semantic differences in their matrix or its
context. Further, the kind and degree of acceptable or required variation may depend on the speaker's
intention, and speech context -- and our standards of accuracy of reproduction, which properly depend
To vivify the plasticity of a display's material and perceptual character, consider: What,
precisely, is the referent of an utterance of This figure O is a circle? Does that sentence printed on an
off-round balloon shift its reference and sense and gradually become a different sentence making a
false statement when the expanding balloon distorts the display? Or do the balloon words, This figure,
continue to refer to its display's original, deflated shape? (What if This figure O is black is a neon sign
whose colors change?) Questions about the identity of a display -- what are its essential, defining,
perceptual and material features -- might not have a determinate answer. Our intentions may be
inchoate in various ways and degrees. Our standards of acceptability of reproduction may conflict, for
adjusting the display’s perceptual properties, but must preserve its sentence position. A prop can be
located anywhere within the speaker's and audience's shared perceptual space concurrent with the
uttered sentence; suffice that attention to it can be captured by some coordinate perceptual act or
circumstance. Displays are presented by their syntactic position. Nothing inside or outside the sentence
points at a display. Apprehension of it is effected by its position within the sentence about it.
Specifically, displays are positioned like terms, because they function like appositive terms: they
Paratactic conceptions deny this and enjoy some plausibility, because display syntax seems
semantically irrelevant with sentences like (2) whose content is expressible with an utterance of a
display-less counterpart like (1) accompanied by a presentation of an equivalent prop. Actually, while
some displays may be replicated by a sentence-external element of discourse, many cannot, or not
For example, Davidson's (1979) only director is his display marks acting as the indexical
descriptor: This expression. He cannot account for -- or even recognize-- unmarked displays and their
legitimate transpositions. Nor can he really account for transpositions of marked displays. His
transformation of marked displays with an explicit director is a senseless stutter: (E'B') becomes (BEE):
More generally, even Davidson's paradigm transpositions flout the standard punctuation practice of
marking a paratactic display when its embedded replicas would be marked. Our current norms restate
('B') with ('B'E). Style books don't recommend the Davidsonian (BE).
17
(BE) would be properly punctuated if the word ‘Badass’ really were a prop, detached from the
displays as props. Such displays are not syntactically independent objects. While outside their matrix,
they remain syntactically attached to it by their adjacency. Sentences may refer to syntactically
detached words, but cannot presentationally refer to them. Speakers may present syntactically detached
words, but sentences cannot, because sentential presentations are essentially positional. Embedded or
Since Davidson sees displays as prop-like, as having both a director and some presentational
device, he thinks his display marks are both, yet he doesn't explain how something could be both a
demonstrative term and a pointing finger. He doesn't see that (a) props and displays have the same need
for a director, and (b) no punctuation could be a prop director so (c) no punctuation could be a display
director either.
Set Davidson aside. A display without syntactic presentation would be like a prop lacking an
extra-sentential presentation linking it with a director. The object would float free of the utterance's
content. Actually, a misplaced display is worse than a free-floating prop because it imperils the
intelligibility of its matrix sentence, as in: *O this figure is a circle/*This figure is O a circle/*Is a circle
O.
18
Paratactic conceptions disregard such evident facts of display syntax. And many others.
Consider:
(PL) Some t’s weren’t crossed and many i’s went undotted.
Most glaringly, paratactic conceptions exhibit a well-nigh willful blindness to the manifest
association of displays with directors. Obviously, displays are very commonly referenced by explicit
verbal elements in sentences. Positing display reference by some punctuation -- which may be utterly
invisible -- seems wholly gratuitous. What purpose could referential marks be serving? They are
redundant even as disambiguators because displays with explicit directors can rarely be read as other
than displays. Here, referential marks can only and do only get in the way as a stutter: This expression
this expression.
The semantic significance of display syntax will appear when we later see that, despite the
extensional equivalence of props and displays, the syntactic difference has modal consequences.
'NF’T: Nick fed Fred is true if and only if Nick fed Fred.
Props are extra-syntactic elements, so they cannot be essential components of a truth due solely to
syntactic form. For now, just recall the point of the previous section about the inferential possibilities
Then, recall another of that previous section's morals. Unlike props, a display's contribution to a
sentence's semantic content depends on, not just its perceptual properties and sentence position, but
19
also on the relations of its perceptual properties to the matrix content and to the perceptual properties of
the matrix and its context. These relations are another dimension of a display's integration with its
matrix. Evidently, sentence meaning may be composed from, not only the meanings of its individual
components and their positional relations, but also from the phonetic relations of the components. We
might conclude that some phonetic structures have semantic effects or that some syntactic relations
9. Contra Autonyms: The syntactic role of displays is specially vivid when they appear
unmarked and solo. We accept (3) (O is a circle) as a complete sentence. A displayed circle is
sufficiently syntactic to complete an English sentence. A presented prop is not. Eliding the director of
Solo displays seem the best case for displays being autonyms. Since (3) may say the same as (2)
(This figure O is a circle), (2)'s director may look dispensable, semantically redundant and syntactically
optional. The display seems to operate independently, referring to itself on its own. The sheer frequency
of solo displays might heighten this appearance, but it probably shouldn’t since the implicit director is
commonly so obvious that its absence is readily explicable by its evident superfluity.
For that same reason, we should not infer that displays are director-independent autonyms from
the bare fact that the interjection of an implicit director may sound plainly unnatural:
(10a) ??? Tom said the words (or, sentence, or whatever) 'Ted slept'
20
We should expect insertion of a director to be awkward, a redundant stutter, when the director is
The (b) sentences are puzzling stutters. The (c) sentences seem to unpack the (a) sentences. (Cf.
director term is an expectable consequence when the term is implicit in a cognate matrix predicate.
Why would -- how could -- such interjection be awkward in just these cases if there weren't an implicit
director?
The awkwardness of interjecting a director isn't probative. What matters is whether display
utterances say the same as a counterpart with an explicit director. The evidence here, and more that
later appears, suggest that display utterances don’t all have a single structure. We shouldn’t suppose
that display utterances must be diagrammed with the display appositional to a director. To make sense
of displays as presentational adjuncts of a director, it may suffice that displays be like an appositive in
21
We'd best be wary of reading data as evidence of display independence, for it may seem that if a
display is referenced without a director, the display must be referenced by itself. But, whatever the
worries of positing unspoken directors operating in the semantic structure of sentences with solo
displays, positing autonyms is way more problematic. It's fully on a par with positing self-referential
Logicians have thought that words must have names if we're to talk about them, just as we need
names for other objects. Yet we need names primarily to refer to things in their absence. Unlike other
things, linguistic objects are "by definition" presentable in sentences. They are essentially displayable
objects.
Any word is so readily referenced and identified by displaying it that we have no call to name
or represent it. We sometimes have reason to identify a word with a demonstrative or descriptor, but
any word you might refer to by name you can better refer to by presenting it. We do name word
sequences (proofs, poems, plays) whose intra-sentential presentation is impractical, but there's scant
reason to refer to a word with a term as long as the word. It's not unintelligible; just unmotivated.
Positing autonyms seems unproblematic because we can make any phoneme or grapheme refer
to any other object we choose, and we can make any object be the referent of any other sound or shape
we choose. We can, for example, make a symbol 'E' refer to the symbol E. So we're well primed to
presume we can make the expression, E, refer to E: we need but intend it.
Yet, we can't well intend a prop-ring to refer to itself. So, why suppose we can intend to use a
word autonymously? That may seem intelligible, because we are familiar with the self-reference of
22
persons and sentences, and the self-denotation of predicates. But these aren't precedents for autonyms.
Sentences like I am English, This sentence is English ,'English' is English can express substantive
truths due to the semantic peculiarities of specific words like I, this, English. But if any word can self-
Is self-reference some kind of natural semantic property of objects? If that seems unlikely or
unintelligible, how do you get an object to self-refer? Do you baptize it, mentally pointing at it and
vowing that it thereafter refer to itself? Could we agree to use glonk as a self-referring word, and have
our dictionaries define the word glonk as meaning, simply and solely, the word glonk? Why we would
‘Nick’ names the actor Nolte and others; ‘circle’ is a predicate of shape, ‘I’ refers to its speaker.
These are cultural facts that are prior to and explanatory of the semantic contents of utterances in which
the terms appear. No sense can be made of object self-reference as a semantic property fixed to
sentence-independent items in our lexicons. Object self-reference seems at all intelligible only as a
situational property of an object in a sentence mentioning it. In fact, objects never appear to self-refer
because the object is in fact referenced and is being used to effect that reference by inducing a
But with a genuine name we refer to a specific bearer only if we so intend the name. Our
intending a presented object to self-refer does nothing. The audience is referred to the object by its
presentation, whether or not anyone intends the object to refer to anything. The presentation explains
both the reference and the appearance of self-reference, and renders any self-reference redundant,
23
or utterance. Instead, that posit only distorts the logical form of display sentences. We’ll see that the
famous controversies and conundra about the logic of displays derive from treating displays as terms.
9. Director Semantics & Syntax: Positing unspoken directors avoids the vacuity of
autonyomy, but if reference to the display is explained by its presentation, why is a director needed?
I cannot fully answer that. It seems to me plausible that while a complete declarative sentence
may have unspoken terms, the statement expressed is explicit in a sentence with at least two terms. But
I am close to clueless how to justify such a principle. We posit the operation of unspoken sentential
elements -- and other unobserved phenomena -- when it seems to fit in some explanation we favor. My
conception of the explanation I am developing is largely inchoate. I argue that displays are best
conceived as presentational adjuncts of a director, but I have little of a larger theory of language that
would make specific sense of that idea. Frankly, I don’t know what a sentence is, or much else that is
assumed in my arguments.
I assume that utterances embedding solo displays say what is said by some utterance with an
explicit director, for I see no counter examples. I assume that what is said with a solo display is best
explained by explaining what is said with an explicit director, because, with props and displays alike,
variations in what is said with variant directors is best explained by the director variance. I don’t see
how to explain variations in what is said with a solo display without assuming the operation of variant
implicit directors.
earlier. His core insight becomes more compelling by noting that (3) (O is a circle) might be short for a
Any object has countless properties. Virtually any predicable of a displayable object could be a
director of its display. The implicit director of (3) might be figure, numeral, letter, title, enclosure,
name, sound, ring, etc. Recall that perceptually diverse objects may be tokens of a semantic type
serving the same director. The flip of this is that perceptually identical objects might serve semantically
diverse directors. What the object is and what it identifies are identified and determined by its director.
Further, as John Buridan (1986) noted near eight centuries ago: I can certainly say “Every term
‘Socrates’ is a singular term”, and “Some term ‘Socrates’ is a singular term”, and “A term Socrates’ is
a singular term”, and “This term ‘Socrates’ is a singular term”. The first of these is universal, the
second is particular, the third is indefinite, and the fourth is singular. Director predicates operate in
diverse syntactic constructions. The syntax of a display varies with its director's syntax.
Evidently, without some assumed director, the thought expressed by a sentence embedding a
solo display has an indeterminate content and structure. Any reading of (a sentence with) a solo display
assumes the content and structure of some director. Let’s elaborate on this a little.
Those four displays are materially and perceptually indistinguishable. Yet, letters and numerals are
distinct and independent kinds of objects. Is the object displayed in (3) the numeral zero or the English
letter pronounced oh? Or is it the general, unspecified shape of (2)? Abstractly it might be any of them;
25
The syntactic options are starker contrasts. (3) may be short for (2D):
Display directors needn’t be demonstratives. Other modes of singular reference may be more suited for
reference in other contexts or to other kinds of referents.The demonstrative this of (2) favors reading
2 A director's identification of a display is a large subject. Outside displays, a particular circle is a letter
(oL) is true and (oN) false because their referent constituent circles function, not as displays, but as
letters, constituents of words. What is predicated doesn't affect their function. Both (2L) and (2N) are
true because their referent constituent circles function as displays. Normally displays are most sensibly
read to be what their matrix says of them. Detached from a director, the circle in (3) is neither a number
nor a letter.
We presume (14) true, because we presume its implicit director is (something like) the word, so the
display is read as a sequence of letters. With a function-neutral director, the display needn't be a letter:
This director lets the object be what the matrix says of it. Displays are objects subserving a sentence, a
the descriptor, this figure, as referring to the adjacent sentence component, whereas (2D)'s definite
article the favors taking the topic to be some abstraction, a kind, type, class, property. (Compare: This
zebra is a striped animal/The zebra is a striped animal.) But of course the may be used to refer to
concreta and this may be used for reference to abstracta, and either may be used to refer to tropes,
The possible referents of directors are ontologically diverse. Their props and displays don't refer
to any of them, but presenting a perceptual object can direct attention upon the individual or a distinct
individual identified by perception of it: a property, type, trope or replica. These ontological
distinctions have great importance, yet commonly they are informationally and conversationally
irrelevant. Commonly, when we’re talking about words and their semantic and syntactic properties and
relations, we needn’t be anxious whether we’re talking about types or tokens, since we sensibly
attribute many the same semantic and syntactic properties and relations to tokens and types alike.
Directors of displays and props are primarily singular terms. Prop directors are predominately
Still, display plasticity is limited. Sentence predicates and interpretive principles can’t make the
(16) is simply false. (17) is not so simple. The (17) on this page is false because it is a mechanically
produced oval. If I free-hand write such a sentence it might properly pass as true. (18) is a curious case
of a failure of reference. It seems more nonsensical than false (despite its possibly stating a truth).
27
demonstrative. Display directors can be demonstrative, but most commonly they are definite
As noted earlier, since displays are referenced by their sentential presentations, we generally have no
reason to name displayable objects. But sometimes we do, so sometimes displays are referenced by
(21) Do you know “Side by Side”, Oh we ain’t got a barrel of money ...?
Paratactic conceptions have insisted that displays are referenced by demonstratives because they
think of displays as props, whose presentations are pointing-like preverbal demonstratives, and whose
directors are predominately demonstratives like this or demonstrative descriptors like this figure. The
The demonstrative this may be cognitively more primitive than a determiner the, but once we’ve
mastered a language, our talk about displayed words is most naturally expressed with a definite
determiner: the word (phrase, sentence, etc.) Generally, we don’t much say things like
unless there’s reason to focus on the particular instance of the name. Despite recognizing that our talk
with displayed words is most commonly talking about the type, paratacticians presume that such
reference requires demonstrative reference to the concrete display because that’s the format of
reference to props. (Why they suppose the display demonstrative must have a descriptor, and cannot be
a bare this is yet another question worth pressing.) Their conviction is self-protective: start regarding
displays as extra-syntactic objects, and you will soon cease taking seriously their syntactic features.
28
Here (more obviously than in (2D)) the director is a quantified predicate with an extension but no
referent for the display to identify. This is vivified in English, where displays normally succeed their
explicit director, but may transpose as quasi-predicates immediately preceding their director:
In their predicate-like function displays look more like common names than proper names. Props are
Another option is directors with referents that are not the display:
Still other options have been proposed. Recanati, for example, as argued that in a large class of
cases he calls "open quotations", displays aren’t referenced or mentioned or predicated either. However,
Sentence Meaning & Speaker Intention Many a conception of display is incompatible with
some of these grammatical possibilities. Many cannot well account for any of them. Saka (2007)
rightly emphasizes the semantic variability of displays, and explains variations of what is said with a
solo display as due to the speaker’s intentions. Certainly, whether (2) or (2L) or (2N) or (2D) or (2Q))
29
However, those intentions don’t explain the syntactic and semantic properties and relations of
the displays in those explicit sentences, (2)-(2Q). We automatically read the circles in sentences (2)-
(2Q) as displays; their presence in those sentences is otherwise random static, inexplicable and
senseless. We assume speakers intend the circles to be displays just as we assume speakers intend the
word circle to be a predicate of shape. That is, we assume they are speaking English, and intend to be
speaking English.
We commonly take such sentences to be types, abstracted from utterances and speaker-variable
intentions. They are constructions whose components are abstractions assigned determinate semantic
and syntactic properties and relations by a system of linguistic norms (syntactic, semantic, etc.) There
Indeed, even without an explicit director, there may be little possibility of wondering what, exactly, a
speaker might be mean by uttering a sentence like (4): His name is Nick.
Still, even with an explicit director, displays are more liable than words to serve diverse
intentions, so speaker intentions have a larger role in fixing a display’s contribution to an utterance’s
semantic content. However, speaker intentions have a far larger role fixing a prop’s contribution to
semantic content. As sentence independent objects, props have much the same properties before and
after the brief moment when they are presented with some utterance about them. Their transitory
semantic role is assigned by the speaker’s intentions. They have semantic properties only while they
are presented along with an utterance. The content may be whatever be the prop’s contribution to an
30
Displays are essentially sentence components. They have no reality independent of the sentence
presenting them. The linguistic properties of such components are largely a function of the sentence
fixtures of a language by being displayed but their reality as display is a creature of the language of its
explicit matrix. Which objects get displayed is a fact of speech. The grammar and semantics of their
display is fixed by the language of their explicit matrix, which may allow little room for speaker
variations.
excluding them, for my hunch is that our speaking with objects is best conceived as originating in a
natural, primitive form of reference with props identifying singular directors, and then developing
alternative formats. My hypothesis is that speaking with objects is an essential, natural dimension of
natural language and natural speech. I take it that human languages could never have developed if we
did not naturally -- without instruction, training, conventions or deliberate intentions -- use and
understand props and displays. We are first introduced to a language with co-presentations of words
and referent objects. When we can communicate in syntactically structured sentences, we have enough
to understand and use props. We don’t need instruction in the semantics of pointing fingers. We seem to
We don’t receive instruction in some semantic rule That semantic relation integrates the perceptual
31
information
a pre-conventional use and understanding of props, and then props are naturally incorporated into
sentences as displays. So, displays might be basically identifiers of a singular term's referent because
props are (nearly) exclusively that. Props don't lend themselves to looking predicative. They don’t
pluralize. They aren't easily "transposed" to form counterparts of sentences like (2P) (This O figure is a
circle).
Put props aside. Referent displays seem to be basic because when displays aren't referents, the
utterance seems to say the same as some utterance with a referent display. Utterances like (2Q) with a
Not so the converse. The singular (2) (This figure O is a circle) isn’t restatable with some
I speak of the basic director as a singular term whose referent is the display. We can say the
display is the referent when the reference is to some abstraction (a property the display exemplifies, a
kind it instances, a type it tokens.) We refer to the abstract -- we identify it -- (if possible) by presenting
the concrete because we are referred to the abstract by perceiving the concrete as instances of a kind.
Quite another kind of evidence of referent display primacy comes from languages that mark
displays verbally, in speech and writing alike, with particles. Ancient Greek fronted displays with a
definite article, usually a neuter whatever the display's gender (Barnes (2003)).3 Displays in Arabic
3 My thanks to John Cooper for further edification on this, and reference to the examples in Kühner,
R., and B. Gerth. (1898–1904).
32
(Carter (1981)) and Sanskrit (Staal (1975), Seaghdha (2004)) are succeeded by a demonstrative.
logicians, like Buridan, standardly marked displays with a nota materialitatis by prefixing iste terminus
(this term) or ly (or li), a particle derived from the French definite article le, which derived from Latin
ille. Later, Leibniz sometimes preceded displays with the Greek definite article (Spade (1996)). (That
Frege was the first to insist on display markers in formal work is one of many myths due to Quine
(1940).)
These conventions are not linguistic universals, but such particles seem grammatically
perspicuous, unlike our display punctuation. They look like natural expressions of displays’ being
Elsewhere, this can be a stand alone term, so This is a circle is a sentence, so there’s little
Here a director predicate seems unnecessary, not because it is normally understood (like The word
implicit in ‘Der’ means the), but because it may contribute nothing to the sentence content, as in:
(2t) seems no less determinate than (2t), and it seems more natural and idiomatic.
But, unlike this, the is not a term, and The is a circle is not a sentence. Thus, while (2d)
and such apparent quantifications of a display need to be read as quantifications of a director predicate.
The determiner the does not appear apart from some term, but it may determine that what it governs be
(23) The when and the how of his arrival were unknown.
These interrogatives are terms referring to what the question they ask is about: the date/time and the
(24) has a curious ambiguity peculiar words like ‘hello’. Like any word in any language, it can be
displayed and appear to refer to itself or something it instantiates or replicates. Unlike other word in
any language, it is uttered by itself to perform an act of greeting, and it is also a term referring to such
acts. The latter use comes close to being a case of object self-reference -- even closer than the term use
of interrogatives.
Unlike the, such quantifiers don’t so well determine a non-term to look like a term-- unless it is a
display: ??? A (any, each, every) when and how of her tour was considered.
Some predicate (explicit or implicit) seems operative for much the same reasons that proper names
seem to presuppose some metaphysical categorization of the referent. There are difficulties
34
understanding how we could identify or refer to some thing without our harboring presumptions and
Natural language display markers call for further empirical research. Cross-linguistic study of
displays is lacking. The little there is suggests that the subject has surprises and complexities. (Coulmas
(1986a) reports that Japanese has a display particle unlike those of Ancient Greece, Sanskrit and
Arabic.) But the little there is is much befouled by misconceptions that distort observations and
descriptions of the data. A prime example is this matter of marking. All the scholarly reports of display
particles automatically equate them with modern punctuational disambiguators. They never consider
Logicians are liable to be oblivious to the operation of a director because their formal work
generally assumes a single general director, like the expression. Paratactic conceptions look workable
in formalist discourse that studies some such single broad category of displayable objects. Such
conceptions are sorely challenged to recognize the evident great diversity of directors in common
speech. So too, autonymic conceptions are challenged to explain how an object manages to self-refer to
any of many kinds of objects, and be the referent of any of many perceptually different objects.
11. Origin of a Conflation: Logical and linguistic studies have long been impoverished by
regarding displays as autonyms, and disregarding the role of presentations in linguistic representations.
Modern scholars, misled by Quine, further cripple themselves by calling displays quotations. The cost
of this novel identification is habitual inattention to the differential roles of display and quotation, and
35
consequent confusions about direct and indirect speech. Direct speech is a display of speech. Displays
of quotations (= repetitions of prior speech) may be the most common and consequential displays of
speech. Their structure and function are incomprehensible without rigorously distinguishing displays
from quotations.
As a term for displays, quotation first appears in print in Quine (1940): "The name of a name or
other expression is commonly formed by putting the named expression in single quotation marks; the
whole, called a quotation, denotes its interior". Evidently, Quine thought some people had so-used
quotation. He did not credit himself with creating this usage: his "Autobiography" (Schillp (1986))
does not list quotation among his many successful neologisms. Nor did he ever explain the choice of
this label. In 1932-3 he began working with Tarski, whose Tarski (1932) refers to displays as "the so-
called quotation-mark names". Likely, displays were so-called by his teacher, Stanislaw Leśniewski,
whom Tarski credited for most of his ideas about displays (Ibid.). Polish logicians were impressed by
Frege's insistent marking of displays, so they might have been influenced by the connection with
Pre-Tarski (1932) publications contain no hint of anyone imagining that a kind of term was
"formed" or created by flanking an expression with quotation marks. Such marks (in variant shapes)
had been increasingly used since the 16th Century to mark word sequences as having the punctuation-
independent property of repeating prior speech. By the 19th Century such marks were occasionally
development explained below). By that century's end, Frege had declared the need for rigorously
Tarski initiated the abbreviation of quotation-mark name by contracting it in the same essay to
quotation name and contracting quotation-mark expression (which denotes quotation functions along
with quotation names) to quotation expression. Subsequently, Tarski and Quine may have fallen into
the conversational habit of contracting further to quotation. Or perhaps Tarski didn't object when Quine
used the contraction. But Tarski never took to Quine's usage in his writings. Nor did their
contemporaries: Carnap, Church, Reichenbach, et. al. Quotation is the jargon of generations brought up
on Quine. Signs of wide acceptance of the neologism come only after Quine's Methods of Logic (1950)
Many of my readers have lived their scholarly lives in this linguistic community happily
oblivious to any peculiarity in their habituated usage of the word quotation. They may be unaware of
its Quinean coinage, or at least unquestioning of its aptness. They may be unaware or simply
unconcerned that previous logicians, linguists and laymen did not liken displays to quotations or
They may be unaware or simply unconcerned that, for a century or so, every respectable writing
manual and style book makes a point of distinguishing displays and quotations, and distinguishing the
proper punctuation of each. Many mid-20th Century English manuals campaigned for the clarity of
italicizing displays and reserving flanking lines for quotation. (Recent manuals drop that proposal,
which, we'll see, is unworkable.) Never mind that the manuals all botch the characterization of displays
(they say displays are autonyms). They are plenty clear that display and quotation are disparate
categories.
From Quine on, logic texts don't deny this distinction; they just don't deign to mention it. They
37
introduce the term quotation to refer to displays. They preach the practice of marking displays with
quotation marks to distinguish displayed expressions from expressions in their matrix uses. The intro
texts do not mention any distinction between marking displays and marking quotations. Nor does the
scholarly literature on displays. That literature sometimes recognizes (as writing manuals regularly do)
that so-called square quotes are another matter, but markers of quotation are rarely recognized at all.
That literature offers various classifications of "varieties of quotation". These seem intended as
classifications of displays. They do not distinguish displays from quotations. A popular category is
popularly called direct quotation. This is not distinguished from what has more commonly been called
direct speech or direct discourse. Direct speech (discourse) is a display of possible speech, something
sayable. More fully, direct speech identifies the product of someone's (possible) act of speech (or
thought) by displaying it. Properly speaking, direct quotations are direct speech reports only of actual,
past speech.
(We'd best be entirely rid of talk of direct quotation. It cannot but breed confusion. Bad enough
it leads to conflating the formal category of direct speech with a historical subcategory of displayed
prior speech. Worse, while direct quotation is indeed a species of quotation differing from unreferenced
quotations, the term legitimates the incoherent term, indirect quotation, that refers, not to a kind of
quotation, but to an identification of a speech product in indirect speech. We say, She said, and here I
This matter is far from purely verbal. Many readers have come to think that displays are
properly called quotations because they really are quotations. This is a perilous habit of talk and
thought. It demonstrably distorts how we view the data. Consider: it has long passed as a scholarly
38
platitude that we mark displays in speech, mimicking inverted commas with air quotes (wagging
Actually, displays get air quotes only when and because they are quotations. (You might say His name
is quote Nick unquote when air quoting scare quotes, a variant of quotation marks.)
The prevailing myth of display air quotes is compelling evidence of endemic inattention to the
whopping contrast between marking displays and marking quotations This chronic inattention is
evident throughout the literature on something allegedly "anomalous" about marked quotations within
indirect speech, as in (6) (Bertie thought we are "but brief meat and brains"). It is evident, for example,
in the common misconception that the display in (6) is not semantically a referent or syntactically a
noun.
To make progress with the famous puzzles about the formal properties of displays, we need to
understand our history, how we came to be writing and thinking as we do. We need to appreciate that
display and quotation are formally disparate categories, whose whopping contrast makes it remarkable
that people could come to think them comparable. We need also appreciate that, however whopping the
contrast, it is no accident that these categories got conflated. The development was unplanned,
unanticipated, and unnoticed, but in hindsight it seems all too natural and predictable.
Scholars of language and logic, East and West, have discussed displays for 2500 years. Only
recently has anyone thought of displays as quotations. Only recently has anyone associated displays
with some punctuation. Previous scholars might sometimes display a quotation, but they didn't consider
39
displaying a kind of quoting. Many had disambiguated displays, not with punctuation, but with some
adjacent demonstrative term that flags the display of an object without altering its properties or creating
a new term. Scholars did not confuse these terms with quotation marks.
No one did, or could, consider displays quotations -- no one would think displays are aptly
named quotation mark names -- before they confused display markers with quotation markers. That's
because display and quotation are such disparate properties as to defy confusion until both they are
Modern conceptions of display are marred by considering displays quotations, but the display-
quotation contrast cuts across any contrasts between conceptions of display. Displays are always
conceived as a class of sentence components distinguished by certain peculiar semantic and syntactic
properties. No one supposes that a prior history is a precondition of being a display. Some display
conceptions may confine themselves to displays of linguistic objects; some may insist that the objects
In contrast, quotation is an essentially historical category that cuts across every formal linguistic
category, syntactic, semantic, phonetic or whatever. Virtually any word sequence of any sentence could
be a quotation. That is, it could be uttered in repetition of a prior utterance. The innards of both displays
and quotations may cut across linguistic categories. However, the display of the words depends, not on
their history, but on their position within some sentence about it (the matrix syntax and semantics). The
quotation of the sequence depends on its history, unconstrained by matrix syntax and semantics.
Quotation is not a property of sentences or their components, but only of their utterances.
Quotation is not mere repetition. Otherwise virtually every utterance would consist entirely of
quotations. To ask What is truth? is not ipso facto to quote Pilate. A quotation is a word sequence
whose utterance is explained partly by its repeating some prior utterance. The relation is causal but
needn't be intentional. You might quote Mencken ignorant of his -- or anyone's-- originating the phrase
you favor. Whatever your understanding of its origins, suffice that your utterance of the phrase is
traceable to Mencken's utterance. Intentionality of repetition is not essential. Nor enough. Reciting is
not quoting. We may quote lyrics or dramatic lines, but not while singing the song or performing the
part. Adding sincerity doesn't help: we may quote pledges and prayers, but not while authentically
pledging or fervently praying. Quoting repeats the words, not their performance. And while repeating is
transitive, quotation is not: when I quote Mencken quoting Veblen, I repeat Mencken and Veblen, but I
Again, virtually any word sequence of any sentence could be a quotation, repeating a prior
utterance, but however much the words match a prior utterance, the later utterance is not a quotation if
If the above isn't obvious, that because display marking is confused with quotation marking.
Before protesting that there are counter-examples to the above contrasts, one should recognize that
neither displays nor quotations need be marked as such, so the alleged counter-example better be vetted
by removing any markers. Better also be reminded of a familiar, often forgotten fact. One clue of the
source of display/quotation confusion appears in Frege's (1893) words alluded to earlier: “It can also
happen … that one wishes to talk about the words themselves. This happens, for instance, when the
words of another are quoted”. Frege's "for instance" rightly implies that words needn't be quoted to be
41
displayed and referenced. But he also here says that when we quote the words of another, we talk about
the words themselves, and thus he betrays a blinkering of quotation that associates it solely with direct
speech ascriptions. Experience teaches that Frege is not alone in this lapse. In saying or writing:
I quote Wittgenstein by using (the standard translation of) his words (an unnecessary shuffle) much as
he used them, for I mean to mean what he meant. He didn't denote or display his words, and neither do
I. Quoting is repeating words, not referencing them. Why, nonetheless, think of quoting as referencing
words? That comes from marking quotations, for that marking does reference the words.
Current histories on these matters are instructive about the distant origins of our current
punctuational practices. (They are less enlightening about the post-Reformation evolution that brought
us to our current practices, because they lack the requisite formal distinctions relevant to linguistic and
logical theories.) Historians trace our current complex use of what we currently call quotation marks
back to their origins, before they acquired their current shape and came to be called quotation marks.
Their precusors are marks used by medieval christian scribes to flag a special category of
quotation: passages of Sacred Scripture, the Christian Holy Writ. Subsequent scribes expanded this
practice and used similar marks for words of revered Church Fathers. Then came the Reformation and
the printing press, and the practice got "democratized" by printers using such marks for quotations
The core message of the markers of quotation -- the QM (quotation marker) message -- became
QM: This is a quotation: i.e., someone said this/this was said by someone
42
Put aside for now that what we call quotation marks are now used to say and/or do various other things,
sometimes instead of marking a quotation, sometimes in addition. Focus on what remains (outside the
discourse of linguistic and formal studies) their most common use: delivering just the QM message.
This snippet of history makes evident the essential historical character of what is marked.
Surely no modern logicians supposes that all and only Biblical passages are distinguished by some
formal linguistic feature (absent even in contemporaneous writings.) We can readily imagine various
motivations for monastic scribes to set off the words of Holy Scripture. So too it seems quite
understandable how monks might be moved to extend the marking to the words of those who spoke
most authoritatively about the sacred words. And we can see the spirit of the Reformation and
Renaissance and the power of the new printing press inspiring the great transformation to marking
everyone's words.
Quotation contains nothing of concern to theorists of linguistic structure, but it has no shortage
of social significance. Relaying another's message is a central, essential function of human speech, as it
is in much of animal "speech". Humans have all kinds of interests (epistemological, political, sexual,
legal, religious, etc.) in who said what and what was said by whom. Human relations often times pivot
on such matters. Our interests in all this are practical, not theoretical. They are heightened by the fact
that, while being a quotation does not itself imply anything about the content or structure the words,
your choosing to express yourself with some quoted words rather than your own may mean something,
and commonly means quite a lot. It bears directly on our understanding of what you are expressing, and
what you mean to be expressing. That's so if only because normally quoted words are meant to mean
Considering the enormous significance of quotation, and its utter lack of any formal features,
nothing could be more natural than speech communities developing extra-formal means of signaling
quoted words. In contrast, outside formal studies, displays rarely need disambiguating markers,
because, outside such studies, their matrix syntax and semantics rarely allows their being read as
anything other than displays. This is specially so when a display director is explicit. (The rare exception
in English occurs with interjections addressing the audience: His name, Bob, is short.) Even solo
displays are rarely readable as other than displays. First off, displayed non-linguistic objects can't be
(read as) sentential elements outside a display. Secondly, since displays are syntactically term-like, only
displayed terms have syntactically viable alternative readings. Finally, the syntactic possibility of
ambiguity is closed semantically for most terms since their denotata are generally categorially unlike
themselves. Few sentences are like (21) and (22) in admitting dual readings.
Few predicates are like preceded and rhymes with in being predicable of both a term and its denotata.
Displays in the discourse of linguistics, logic, and mathematics are exceptionally liable to ambiguity
and consequent misreading because the names of linguistic categories are commonly members of the
category or have many similar properties, and our systems of numeration construct terms with
mathematical properties that enable the discovery and representation of mathematical relations by
operations upon numerals. Logicians like Buridan, Leibniz and Frege marked displays in their writings,
Consequently, when quotation marking was introduced and achieved wide acceptance there was
44
no marking of displays that could be confused with it. Displays were marked only in the arcane
writings of a few formal theorists, and their verbal markers were unreadable as markers of quotation.
Of course, quotation can be flagged in any language by ascribing it, predicating its historical
property of being spoken: the Bible says, Astrid asked, etc. However, such ascriptions are often
unwanted intrusions. Then too, they are often insufficient because languages commonly express both
direct and indirect speech with the very same words, like: Ned said neon is inert. It seems a natural and
nearly inevitable feature of natural languages that speech is identifiable by presenting the words and by
using them to represent their content. Many languages have no verbal devices distinguishing the two.
Many have devices that are unreliable indicators: e.g., proper English often drops the complementizer
that for indirect discourse and sometimes inserts it with direct discourse.
Again, quotation isn't evidenced by its formal features, so its importance has prompted the
In daily discourse, the primary source of display ambiguity is due to a large class of speech predicates
Apparently, while the look of the marks varied from country to country, the function was the
same: the marks marked words someone had said (spoken or written). Outside the writings of some
logicians, there was no practice of marking displays, per se, but only quotations.
I have seen no evidence of the initiators of this practice having intended or even anticipated the
consequences of their practice. I wouldn't expect them to predict the consequences, or to have much
45
imagined the possiblities. In hindsight, much of the complex consequences seems virtually inevitable.
Quotation marks, from their very beginning, have been a peculiar, perhaps unique, linguistic
phenomenon. Prosodic and behavioral cues of direct speech and quotation are commonly found, but
these are essentially phenomena of speech, elements of utterance performances. In our speech they are
not properties of the uttered sentence. We rightly regard our punctuation as elements attached to a
sentence. Other punctuation (periods, commas, colons, capitals, question marks, exclamation marks,
etc.) signal
Quotation marks are peculiar punctuation because they are actually utterance markers, not sentence
markers.
That's because they mark words for an historical property of their utterance, not for any linguistic
Indeed, we not uncommonly quote some words quite unaware that we are quoting. Sometimes
we do refer to the words by attributing the quotation, by saying (e.g.) She said "...". Sometimes we may
Attributing quotations and marking them are distinct ways of displaying words, because
attributing the words makes the reference within the matrix, so the historical predication is in the
matrix message message, while marking them makes the reference and historical predication outside
The markers did serve as a disambiguator of sorts, for though it didnt distinguish between
distinct semantic contents of the words, it distinguished bewteen direct and indirect discourse, between
In this latter capacity, the markers of quotation became markers of display. Still, the marks of
(25) could be quotational. And, I take it, originally they were only that. But, as (26) and (27) illustrate,
the marks couldn't remain quotational when direct speech gets negated or questioned. Consider the
(28d) Pete did report 'Paul snored' or Pete did not report Paul snored.
Since quotation is an historical property of uttered words independent of their grammar and semantics,
their marking is bound to be problematic when direct speech gets negated or questioned. Such marking
was bound to give birth to display marking because, as we'll see shortly, it creates a display.
It's well-nigh logical impossibly to write a coherent history of quotation marking while calling displays
quotations. Nor will it help to use the terminology of the orthodox taxonomies of quotation. Their
Any association of displaying with quoting is utterly alien to anything in the prior two millennia of
thought about displays. The association is traceable to Tarski's taking displays to be names
distinguished by the quotation marks defining their reference
Perhaps because , you're thinking of quotation as essentially a display. And you're primed to think of
displays as referencing words by saying them. Something else, I conjecture, something more insidious
has been afoot.
12. Quotation:
Humans have all kinds of keen interests (epistemological, legal, romantic, economic, etc.) in
who said what and what was said by whom. The diversity of interests is due partly to the diverse
readings of what was said and the diversity of possible referents. What was said may be identified by a
presentation of the words uttered or by a description of their content
We naturally adjust our standards for accuracy and adequacy of quotation to suit our current interests.
Sometimes close paraphrase passes for quotation. Sometimes deictic rules are transgressed. Sometimes
quotations are translations. Other times verbatim replication is essential but insufficient, and prosodic
factors, along with stutters and stammers, must be replicated. Quotation attributions can go wrong all
kinds of ways, and our concerns with imperfections are no less variable. So, generalities about
quotation (like many that follow) are liable to "exceptions" because what qualifies as a proper
quotation -- which features of the original must be replicated -- is contextually variable and liable to
controversy.
Our current conception of quotation, is intriguingly contoured. How much is culturally peculiar
is unknown. One peculiarity is documented. We, the educated people of the modern world, have
institutionalized the practice initiated by sixteenth century European printers of punctuating words
someone said as quotations, usually by flanking them with one or two lines (Parkes (1993), Finnegan
(2011)). That innovation has been unexpectedly consequential. One predictable consequence is that we
can now produce sentences like (6) (Bertie thought we are "but brief meat and brains") wherein
quotation marking can create ambiguity and alter propositional content.
Quotation is a primitive, primary communicational activity. Much animal speech relays
messages. Pre-print people produce quotations to attribute them in direct speech. Direct speech (aka
direct discourse, oratio (di)recta) attributes to a speaker a possible (not necessarily previous) speech
act, and identifies its product by presenting it. Direct speech attributes speech by displaying it. Indirect
speech attributes speech, but it identifies speech products by a representation of the product's
propositional content. The difference between a presentation of someone's words and a representation
of their content is commonly a matter of intense human concern; human relations frequently pivot on
such matters.
That concern contends with the fact that direct speech and indirect speech can take the identical
form:
(22) Ed said Ned fed Fred
(22) has two distinct, inter-related, and essentially equivalent readings that appear in the truism: To say
Ned fed Fred is to say that Ned fed Fred. For many, though not all, verbs of speech, what was said, the
product of the saying, may be or be identified with a propositional content, so a display of words may
be equivalent to a representation of their content.
Human languages evidence our concern to distinguish direct and indirect speech. Oral direct
speech is often marked by prosodic, paralinguistic and kinesic behaviors (Cf. Wilkinson (2010)). Many
48
languages also have grammatical markers of direct speech and indirect speech, like:
(25) Ed said that Ned fed Fred.
(Cf. Coulmas, 1986, Kvavik (1986), Klewitz and Couper-Kuhlen (1999)). Modern writing often marks
direct speech with punctuational markers of display. This punctuation commonly gets mistaken for
markers of quotation.
Current writing often marks quotations, but this is grammatically and semantically optional.
Compare:
(20) Cosmological arguments are an unnecessary shuffle
(20m) Cosmological arguments are "an unnecessary shuffle"
The core QM (Quotation Marking) message is a bare historical claim:
QM: This is a quotation: i.e., these words were said/someone said this.
The QM message is an extra, off-stage, historical side-comment on the words used. It is not a
component of the quoted message or the matrix message. The truth or falsity of the historical claim
doesn't touch (20) or (20m)'s meaning or truth. Substitution of a synonym into the marked quotation of
(20m) may alter the QM message's reference and truth, yet the matrix sentence meaning and truth are
unaffected by synonym subbing in (20) and (20m) alike. Quotation marking is extra-linguistic, added
after all the words and other punctuation are in place, so the sentence meaning is already fixed.
A complication here. Absent any QM message (by punctuation or matrix term), quoted
indexicals have the referent of their use in the sentence matrix, not the referent of their original use.
Marking a quotation relates the words to a prior use, so marked quoted indexicals normally have the
reference of their original use, not their matrix use. More generally, when other factors in the sentence
or speech content identify the original speaker or the circumstances of the utterance, the marks may
suggest a particular reading of the quoted words. However, the marks don't enforce reading replication
of intent. I don't misquote you when I purposely intend something other than you meant by your words.
Display marks do enforce a reading, and nothing more. The DM (Display Marking) message is
more a directive than a predication:
DM: This is a display. Read this as a referent of this sentence.
Contrast:
(23) Peace preceded prosperity
(23m) 'Peace' preceded 'prosperity'
The nouns of (23) could be presented objects or names of situations. The marks of (23m) enforce the
display reading and eliminate the alternative. Such punctuation is like capitalizing initial sentence
letters: a typographical device facilitating identification of an intended reading which it presupposes
and does not alter.
14. Marks of displayed speech: Let's call the director of a direct speech display a speech
director. It is the (explicit or implicit) direct object of a verb of speech (or thought):
(25) Pete reported [made the report] 'Paul snored'
(26) No one ever said [the words] 'I am a transsexual Pope'.
(27) Has anyone ever thought [the thought] 'I am a transsexual Pope'?
Absent an explicit speech director and the display marks, (25)- (27) may be ambiguous over direct and
49
indirect speech.
(25)-(27) illustrate marks of display commonly considered marks of quotation. Yet, if the
marks of (26) signaled quotation, they would contradict the matrix sentence message; (26) would then
express inconsistent messages. Similarly, in (27) quotation marking would preclude the questioning of
the QM message by the matrix sentence. In (25) the QM message is contained in the matrix message,
delivered by the speech predicate on the display; here quotation marking is permissible, but redundant.4
Still, the marks of (25) could be quotational. And, I take it, originally they were only that. But,
as (26) and (27) illustrate, the marks couldn't remain quotational when direct speech gets negated or
questioned. Consider the awkwardness of the disjunctive:
(28d) Pete did report 'Paul snored' or Pete did not report Paul snored.
Since quotation is an historical property of uttered words independent of their grammar and semantics,
their marking is bound to be problematic when direct speech gets negated or questioned. Such marking
was bound to give birth to display marking because, as we'll see shortly, it creates a display.
Quotation marking is truly informative only where the QM message is otherwise absent or
underspecified by the matrix sentence, as in (20m) (Cosmological arguments are "an unnecessary
shuffle"). Absent the markers, nothing suggests that the words are quoted.
Now consider:
(29) Bertie thought we are but brief meat and brains
The matrix predicate implies that some words within its scope could be quoted, but not that any are.
(29) could be purely direct speech, just as in:
(30) Bertie thought "We are but brief meat and brains"
or purely indirect speech, as in:
(31) Bertie thought that we are but brief meat and brains
or indirect speech by direct speech as in (6) (Bertie thought we are "but brief meat and brains"). And
(6), in turn, could be either (6I) or (6D):
(23I) Bertie thought that we are "but brief meat and brains"
(23D) Bertie thought "We are "but brief meat and brains""
Whether as (6I) or (6D), (6) is said to be puzzling, in a way that (30) and (31) are not, because
the marked words of (6) are both "used" and "mentioned", so they seem to have two different meanings
in the very same reading. Actually, displayed words have the same meaning(s) and referent(s) that they
have in their matrix uses.
Displayed words are referents of a sentence. Like any sentence referent, (a) reference to them
does not alter their properties, and (b) which of their properties affect or are relevant to the meaning or
truth of a sentence depends on the sentence predicates. To say Ned fed Fred is to say that Ned fed Fred,
precisely because the words have same meaning in their display and matrix occurrences. The displayed
terms in (26m) ('Peace' preceded 'prosperity') and (27m) ('Terry' rhymes with 'Keri') are not divested
4 Displays and quotations are standardly explicitly distinguished in writers manuals and style books
(Rarely, if ever, in texts of logic or semantic theory). Mid-20th Century English manuals frequently
urged italicizing displays and reserving flanking lines for quotation. Recent manuals omit that proposal,
perhaps because of persistent confusions, for they all assume displays are autonyms, and their
paradigms of quotation marking are display markings like (25).
50
of their meaning or reference; rather, their semantic properties are simply irrelevant to the semantic
properties of the sentence. The component words of (25) (Ed said Ned fed Fred) have the same
meanings in direct and indirect speech. The distinct uses of Ned fed Fred are syntactic, not semantic.
The puzzles about (6) dissipate by recognizing that while its marks don't say This is a display ,
they do imply it, for they make it true. The words are displayed by and within the QM message: These
words, ..., were said. Within that message the display is syntactically like an appositive noun. That
message is outside the matrix message, so the displayed word sequence needn't be noun-like in its use
in that matrix.
The QM message may stay outside the sentence message when the marks fall within the scope
of the speech verb as in (6) (Bertie thought we are "but brief meat and brains"). (6) may be true just
whenever the unmarked (29) is true, like (20)/(20m). But, by creating a display within indirect speech,
(6)'s marks create a potential referent of a speech director: Bertie's thought. The marks don't themselves
create the reference. The speaker does.
Speakers mark quotations from various motives, and may mean various things by the marking.
Someone might mark (29) just to disavow credit for Thomas Kinsella's striking alliteration that
captures Lord Russell's thinking. If so, the QM message is extraneous to the matrix, so (6) may state
the same fact as (29).
Instead, more commonly quotations in indirect speech are marked to indicate that the speech
referenced contained the quoted words. That's the natural reading if we are unfamiliar with Kinsella's
line. Here (6)'s speaker identifies Bertie's thought by representing its content partly by presenting
Bertie's words encased in quotation marking. Now the off-stage, Someone said these words, becomes
the on-stage: Bertie said these words. More fully, here the utterance of (6) has the form: Bertie thought
we are but brief meat and brains, partly by thinking the words "but brief meat and brains". If you think
Russell knew nothing of Kinsella's words, you might deny (6) with: No, he didn't.
Quotation marking creates a distinctively modern channel of self-reference and communication,
with new content and concerns and kinds of irony and self-consciousness.
16. Autonym v. Intranym: Calling displays quotations has distorted the study of speech and thought.
Calling both displays and intranyms quotations has derailed research on meaning and truth.
Tarski (1932) aspired to found a scientific study of truth by formalizing a fragment of language
with sentences referring to sentences and their components. Formalized languages are "artificially
constructed languages in which the sense of every expression is unambiguously determined by its
form" (Tarski (1932)). A Tarskian formal language -- call it Formalese -- is an artifact whose every
51
feature is fixed by explicit stipulations and their entailments. Formalese goes metalinguistic by
FI: An expression, 'E', consisting of marks flanking an expression, E, denotes E (or its kind).5
FI creates a species of expressions best called intranyms because they are terms denoting the
expression within themselves. Absent a version of FI, Formalese has no systematic means of
identifying and referring to the sentences and components that semantic theory studies. There is no
From Tarski (1932) onwards, logicians have assumed that the FI stipulated for Formalese is also
a description of natural language marked displays. Actually, (aside, perhaps, from the idiolects of some
logicians) intranyms are used only in Formalese. Natural languages have no use for intranyms, and no
use of them. There is no evidence of anything like an intranym in any language or theory of language
prior to Tarski. Yet now intranyms and marked displays get "mentioned" interchangeably. From Quine
1940) onwards, logicians have come to use the term quotation to refer indiscriminately to the intranyms
of Formalese and the marked displays of natural languages -- and thus to regard marked displays as
Typically,they are presented as though stating a Formalese formation rule. Yet they get used as though
FI is a description of natural language displays. (Cf. Smullyan (1957), Richard (1986), Ludwig (1998),
5Tarski (1932) defines quotation mark name as a name of an expression “which consists of quotation
marks, left- and right-hand, and the expression which lies between them, and which (the expression) is
the object denoted by the name in question”. The Quine (1940) definition of quotation quoted above
restates this.
52
Soames (1999).) 6
Through a glass darkly, Davidson (1979) saw that Tarski, Quine, et. al. didn't distinguish
intranyms from marked autonyms. Both terms refer to their interior. But the interior of a marked
autonym is an autonym. Autonyms (if at all intelligble) are terms, self-referential syntactic components,
whether or not they are marked as such. Autonym marks are mere disambiguators just like display
marks. Intranym marks don't disambiguate.They don’t discriminate between pre-existing options. They
create a new kind term that would not otherwise exist. Its referent is its interior.
To confuse intranyms with marked displays is to confuse FI with DD, our rule for display
disambiguation:
Since autonyms are misconceived displays, to confuse intranyms with marked autonyms is to confuse
FI, DD and DA are starkly contrasting readings of a marked expression, ‘E’. Yet they are readily
6 Capellen and Lepore (2007) defend the descriptive adequacy of their version of FI by declaring it
immune to disproof by any evidence of unmarked displays: a theory of display needn't consider
unmarked displays! Why not? Allegedly because unmarked displays are optional: a natural language
could (somehow) get along without unmarked displays, or reject them as Formalese does. Capellen and
Lepore don’t notice that this would make a stronger reason why a theory of display needn’t consider
marked displays. They do not explain how (as things are) we could come to understand and use marked
confused. After all, none of them debar the intranym interior from being (read as) a display or autonym.
Little wonder that people don't notice their conception of “quotation” shifting among FI, DA, and DD.
DA and DD are not inconsistent with FI, but they are incompatible with Formalese. Formalese
cannot have disambiguators because disambiguation recognizes the legitimacy of alternative readings.
Expressions can be displayed or used autonymously in Formalese only if such use has been stipulated
there, but that stipulation licenses systematic ambiguities and equivocations, and thereby violates a
basic requirement of a formal language. Formalese cannot contain any ambiguity for FI to eliminate or
cancel.
Formal languages cannot allow displays or autonyms, but natural languages can include
intranyms. The idiolects of some logicians might actually be regulated by FI. But this innovation
introduces a distinct third use of a punctuation that has long marked repetitions of speech, and more
recently also marks displays. Quine thought he followed Frege in promoting the rigorous practice of
using such marks as disambiguators to preclude confusion of word and object, term and referent, use
and mention. That ambition betrays his own word/object, term/referent confusion. Frege promoted the
rigorous application of an existing convention to mark displayed expressions to preclude reading them
representationally. The intranym marks Tarski and Quine touted do preclude a representational reading,
but they preempt that ambiguity by introducing a more insidious ambiguity, because the intranym these
marks create looks just like a marked display. But, an intranym refers to an object displayed in a
marked display of it. Thus does an intranym term get confused with its displayed referent, .
Intranyms within a formal language create no ambiguity and may have real utility. Introducing
intranyms into our scholarly discourse about languages and metalanguages creates an ambiguity and
54
consequent confusion. People were already prone to confuse display marking with quotation marking.
Now they are further prone to confuse display marking with intranym marking, because intranym
marking is like quotation marking: with both, the result of adding flanking lines is an expression that
refers to its interior. However, an intranym refers to its interior and doesn't predicate anything of it; that
reference is the intranym's contribution to the sentence's semantic content. Quotation marking refers to
its interior and predicates an historical property of it, but that reference and predication are outside the
sentence's semantic content. Display marks refer to the display but that reference only enforces a
reading it does not create or alter. The introduction of intranyms has been a cause and consequence of
17. Term/Object Confusions: The consensus amongst the cognoscenti on the topic is that the
conception of displays as self-referring names is a horse decades dead, not needing further beating
(Cappelen & Lepore (2012)). Yet, logicians and linguists still standardly call both displays and
intranyms names. Even Davidson unapologetically retained Tarski's talk of marked displays being
names whenever he explained his transformation of Tarski’s theory of truth into a truth-conditional
theory of meaning. Meanwhile, Davidson and everyone else applies quotation to intranyms and
Actually, the old autonymic misconception of displays remains alive, while the new intranymic
misconceptions sometimes collide but often happily coexist because at bottom both are confusions of a
origins. Kaplan (1968) is hardly alone in thinking that Frege thought that “quotation marks produce an
oblique context”. Actually, that idea is the Tarskian innovation Quine promoted. Actually, displays are
"oblique contexts" whether marked or not, and Frege never suggested otherwise. Like nearly all his
predecessors, Frege thought displays are autonymously used expressions, and, like many before, he
thought his formal work demanded disambiguating markers. That's unmistakable in his famous
declaration of a policy of marking displays because there was then no such regular practice. It's
unmistakable in his comments on the innocuous absence of display marking in Riemann’s writings
(Frege ,1898?)). That absence may be regrettable, but Frege would find it risible to be told that
Riemann's sentences having unmarked displays are "not merely untrue but ungrammatical and
Frege's statements about language are all made in natural language sentences using
language or considered a formal counterpart of displays. He and two millennia of predecessors left no
traces of any truck with intranyms or any inklings of display marks being essential constituents of
displays. Why? Because absent the formalization of a metalanguage, there's no reason -- not even a bad
reason -- to imagine anything like an intranym. (Frege warned against incautiously presuming that
Frege thought he used names of words and symbols, but he never did. His autonymic
misconception did lead him further astray than his predecessors, because Frege (1892) likened displays
to intensional contexts as cases of “referential obliqueness’ since in both ‘HP’ and SHP, replacing
‘Hesperus’ with the coextensional term ‘the Morning Star’ may turn a truth into a falsehood.
56
(Predecessors back to Panini had noted this feature of displays, but not likened it to intensional
contexts.) Regarding these cases as truly comparable comes from mistaking displays for terms.
Actually, display opacity (=obliqueness) is the opacity of a prop, a referent object, not a term.
Props and displays are not referential contexts. Display substitutability is the substitutability of an
object, not a term. 'Greece is boldface' and 'Greece is boldface' display the same term but different
kinds of object. 'Hellas is boldface' displays a different term but the same kind of object. Which objects
replace a display without altering a sentence's semantic properties depends on which display properties
are semantically relevant, which depends on the sentence meaning, especially the predications on the
display. Further, recall that which objects can replicate a display in a reproduction of its matrix sentence
depends also on the perceptual properties of the matrix and its context. Neither of these constraints on
display equivalence have any counterparts in the equivalence of intensional or modal contexts. The
Frege's autonymic misconception of display opacity does not appear to distort his conception of
intensional opacity, or his semantic theory generally -- any more than autonymic misconceptions distort
previous semantic theories. It's the peculiar post-Tarskian conception of opacity that inspires Quine's
(1953) semantic reductionism, where displays become "the referentially opaque context par
excellence". Again, with displays and autonyms, the marks are irrelevant to the opacity. Intranym
opacity is created by its marks. More precisely, intranyms and their opacity are created by the
57
stipulation of FI. FI invites the idea of an intranym interior being an extra-syntactic "orthographic
accident" (Quine (1953)). Kaplan (1968) saw that idea leading to Quine's (1953) considering displays
"the referentially opaque context par excellence", and explaining away other forms of referential
opacity by way of display opacity. Frege's autonyms are no orthographic accident. Marked or not, their
reference is themselves, not some sub-syntactic snippet of themselves. Their opacity is not "par
excellence"; it doesn't distort Frege's conception of intensional contexts as intranym opacity distorts
Quine's.
Quine's misconception of display opacity culminates in his "disquotationalism": the thesis that
predicates like true and names "cancel" the semantic import of display marks. Recall that display
substitutability governed by semantic predicates (true, names) is just like term substitutability: only the
displayed object's semantic properties are relevant to the sentence's semantics, so objects with the same
semantic properties freely intersubstitute. Quine explains this referential transparency of displays
governed by semantic predicates by supposing that the marks create an opacity that semantic predicates
Consequently, with semantic predicates like 'true', display substitutability is like term substitutablity in
referentially transparent contexts: objects with the same semantic properties freely intersubstitute
Actually, display marks have no semantic import to cancel. (Nor do autonym marks.) Only intranym
marks have any semantic import. Only Formalese semantic predicates could be disquotational.
"deflationism" about truth (e.g., Frege, Ramsey) makes any mention of display marks. Post-Tarski,
disquotationalism may never have been the dominant view, but even its many critics don't question its
A final point on opacity. Tarski (1932) and Quine (1940) assure us that problems of intranym
descriptions identifying an expression by its spelling. The adequacy of this alternative went
unexplained and undefended. Some skepticism seems in order. After all, how could the spelling
alternative avoid the semantical problems if it is truly semantically equivalent? Now recall: 'Greece is
boldface'/'Greece is boldface'. Here the reward of avoiding referential opacity is encoding a truth and a
falsehood with the same spelling description. Consider also embedded intranyms: ''' '' 'Is an intranym'
is an intranym'' is an intranym'''. What is the equivalent spelling description? Consider also intranym
variants, like Quine's (1940) quasi-quotation marks and Sellars' (1950) dot quotation marks. All have
opaque innards, but cannot all be replaced with spelling designators. Logicians and linguists invent an
ever growing array of markers, because they need to refer to arrays of formally different kinds of
objects – objects natural language displays with its flexible system of directors. What would be the
spelling-designator variants for all the intranym variants? Finally, it needs asking: Suppose spelling-
designators were extensionally equivalent to intranyms. Still, are they fit for the practice of science
despite being cognitively inequivalent and infeasible? Like displayed words, intranyms can be read.
Words represented by a spelling description must be deciphered to be read, a formidable procedure far
more liable to error. Is a science of language really furthered by resorting to imperspicuous and
Anyway, as replacements of displays, spelling designators are no better than intranyms . Neither
can replace displays, because they presuppose displays, because Formalese terms have referents and
Formalese sentences have truth values only by assignments grounded in our use of props and displays.
That aside, even as an uninterpreted system, using intranyms or spelling designators, Formalese
For most folks, this formal entailment is an amusing absurdity canceled by extra-formal
disambiguation. In practice it's inconsequential, an innocuous linguistic curiosity. Yet that entailment
challenges the very idea of a formalized metalanguage. Formally entailing an absurdity is fatal for a
language lacking extra-formal means of disarming syntactic equivocations. So too, the equivocation in
(40) is essentially the duality in sentences like HN (He is Nick), which may introduce a person or his
name. Formalese cannot admit such dual readings; natural languages cannot avoid them.
displays in his writings. If marking displays became as standard as capitalizing initial sentence letters,
sentences with unmarked displays would still be no more defective, grammatically or semantically,
than sentences with lower case initial letters. (40) would still be perfectly grammatical and meaningful,
albeit obviously absurd. (Actually, Frege's practice is not universalizable. We must use and understand
displays before we understand and use display markers.) The legitimacy of (40) doesn't threaten Frege,
for the fragment of language he formalized was not metalinguistic. It does threaten a language aspiring
60
to formalize Nick is his name. Formalization cannot recognize (40) as well-formed, albeit absurd,
because the preclusion of ambiguity and equivocation is a primary precondition of a formal language.7
Tarski and Quine the deny of (40). For them it was axiomatic that "a statement about an object
must contain a name of the object rather than the object itself" (Quine (1940)). Their denial was
emphatic. It was unprecedented. And it remains undefended. No one before had need of it, and no one
Formalists don't recognize that they are denying the legitimacy of our natural metalanguage of
displays. They recognize the reality of semantic paradox and aim to avoid it by restricting their object
language semantically. They see that intranyms save their metalanguage from ambiguity, equivocation
and absurdity, which they regard as regrettable defects of a natural language that are costlessly
eliminated in a properly regimented formal language. Formalists don't recognize that they preclude
syntactic paradox by replacing displays with intranyms and restricting the syntax of their metalanguage
19. Display v. Intranym Syntax: Formalese cannot state the formal truths expressed by
7 Recall Tarski's defining formal languages as "artificially constructed languages in which the sense of
every expression is unambiguously determined by its form". Per Frege,“the first requirement” of an
ideal language for logic and science is that the symbols be univocal, unambiguous (Frege (1882a)). To
make "logical" transitions from one judgment to another, “we may not use the same symbols with a
double meaning in the same context” (Frege (1882b)). A Formalese symbol may be assigned multiple
meanings only when this doesn't legitimize sentence ambiguities or equivocations.
61
Metalogical sentences are comparable, not to analytical sentences (like Squares are quadrilateral), but
Both logical and metalogical sentences can express both formal, necessary truths and contingent,
empirical truths and falsehoods. Whether their utterance expresses a necessity or a contingency is itself
a contingency dependent on whether the speaker is using two occurrences of the semantically same
word(s). Languages require repeatable objects, perceptual types with semantically identical tokens. Yet,
So, speech would be epistemologically impossible without the interpretive principle that the
normal, default reading of sentence tokens is that perceptually identical sentence components are
semantically identical.8 Necessarily, that's the standard reading of logical and metalogical sentences.
Presentations of the name, Nick, might refer to various usages of the name with various referents. Still,
our normal, default reading of ('N'N) and (NN) utterances is that, whatever either Nick-token may
name, they both name the same. So read, ('N'N) and (NN) are true, necessarily.
The flip side of this is the normal reading of sentences like: Nick is not Nick, Nick might not be
Nick, 'Nick' does not name Nick, 'Nick' might not name Nick. Rather than read them as denials of
obvious necessities (or denials of their obvious necessity), we normally assume that their occurrences
8 Gallois (2005) alleges that this principle is merely a convention, and thus self-identities are
Whatever our difficulties explaining it, we readily read and understand formula like:
(n) n names n
Displays are significantly like variables, and can be positioned like variables, because they take the
position of a referring expression but do not themselves refer to anything. They do not form a complete,
closed sentence with just a predicate. Sentence completion requires replacing variables with terms, and
supplying directors for displays. Displays are peculiar variables because matrix semantic predicate
words are read as the representational use of the words displayed, and the words displayed are read as
presentations of the words used representationally in the matrix semantic predicate. (Cf. Beaney,
Harman)
Displays aren't quantifiable, just as props aren't, because they are objects, not terms denoting objects.
Instead, a display's director may be quantifiable like any term, with or without a display. Thus a
sentence like (14) (Every i was dotted, but some t's went uncrossed) is formalized like (14F), not
(14F*):
(14F) (x) (Lix -> Dx) & - (x) (Lt x -> Ux) -- where ‘Li’ abbreviates 'letter i’ and ‘Lt ’ abbreviates
‘letter t’
(14F*) (x)(ix -> Dx) & - (x)(tx -> Ux) -- where ‘i’ and ‘t’ are displayed
Consequently, when the display's implicit director is something like, the expression, so-called
"substitutional quantification" of displays makes sense (Dunn (1968), Marcus (1972)).Since a display is
an object, not a term,
Now consider the much vexed question of quantification into or over displays. Evidently, what
it gets quantified is the director, not the display. Director terms are quantifiable like any term, with or
without a display. Displays aren't quantifiable, just as props aren't, because they are objects, not terms.
Perhaps that's why, while we can say
we more commonly appear to quantify over displays with explicit directors as Buridan does. In any
case, so-called substitutional quantification of displays makes sense when and because the display's
63
(implicit or explicit) director is something like: the expression. (Cf. Dunn (1968), Marcus (1972)).
Display opacity is of a piece with its quantifiability. Displays aren't quantifiable, just as props
aren't, because they are objects, not terms. However, a display's director may be quantifiable like any
term, with or without a display. Consequently, when the display's implicit director is something like,
the expression, so-called "substitutional quantification" of displays makes sense (Dunn (1968), Marcus
(1972)).
expression and its matrix use, between a presentation and its representational use. As with the sentence
form, a=b, any sentence of the form, (The name) 'a' names b, is true just when its two terms co-
designate. Sentences of either form are formally necessary just when the co-reference is free of
semantic contingency -- and thus, just when the two names are the same name. Thus, 'a' names a is a
syntactically distinctive form of 'a' names b because its truth is due to its structure independent of its
referential relations.
Elsewhere (......) I argue that (contra MacFarlane (2000)) a=a is a paradigm truth due to
logico-syntactic form. The logician's '=' is not the relational predicate of equality. It is comparable, not
to '>' and '<', but rather to a sentence connective like '.' that forms sentences that are true just when their
component sentences are true. The identity sign forms sentences that are true just when their terms
corefer, whatever the (extra-semantic) properties of the referent, so their truth is formally necessary just
when the constituent terms are the same. (Quine (1987) betrays his own use-mention confusion when
Earlier we saw that matrix predicates may require altering a display's perceptual properties to
64
metalogical sentence must have the same truth-securing syntax and a semantically equivalent semantic
object. So a proper translation of 'S'S must be like 'G'G and not like 'S'G:
'S'S's syntax demands that its translation switch its ostensible referent by translating the displayed
expression the same as its matrix occurrence in the semantic predicate, besagt Schnee schmilzt.
Elsewhere, extensional equivalence is essential for proper translation, but here the referent's role
explaining the truth ostensibly about it is subordinate to syntax. Like objectual logical truths, a
metalogical truth is translinguistic. Its truth is explained by a structural feature of sentences with a
The Formalese 'S'S is not a formal truth. It states the contingent fact of the semantic relation of
a term to its referent, a relation created by the stipulation of FI. No term has any syntactic, formal,
logical relation to its denotation. The intranym referent might be the semantically same words as the
matrix component in the semantic predicate, but that coincidence is an artifact of a formally arbitrary
stipulation. So, a Formalese 'S'S and 'G'G are structurally like the inter-linguistic contingency, 'S'G. So,
in Formalese, 'S'S is semantically equivalent to 'S'G, not 'G'G. (Church (1954), Salmon (2002). Cp.
Olders & Sas (2001).) Formalese metalogical sentences state necessities comparable to Greece is
Hellas and Greeks are Hellenes. They are necessities explained by semantic contingencies.
Formalese cannot represent the logical form of metalogical truth. If Formalese takes displays to
65
be paratactic, it takes a sentence that is true due solely to its syntactic form to be dependent on an extra-
syntactic element. If Formalese takes displays to be autonymic or intranymic, it takes a truth due solely
Further, Formalese differs from our natural metalanguage in its subject matter as well as its
form. In Formalese the referent of a metalogical sentence needn't be semantically the words in the
contained in Formalese.
Formalese semantic truths are indexed to the language of the words referenced. Our metalogical truths
aren't indexed to any language: their displayed words replicate the words of the semantic predicate,
This might seem otherwise, because, again, a metalogical sentence utterance might not express
a metalogical truth. Whether it does depends on the speaker's intent, which may be uncertain. So
sometimes there's need for the display's director to specify the display as the words of some language
or as used on some occasion. Sometimes we can infer that p from an utterance of 'p' is true only on the
empirical assumption that the referent sentence means what we now mean by its utterance.
Still, if 'p' is true normally needed the premise 'p' says p to imply p, it would need the further
premise:
'p' says p says 'p' says p says 'p' says p says 'p' says p.
66
Like objectual modus ponens, metalogical truths are principles of reasoning and representation, not
premises of it. (If that weren't the default reading of metalogical sentences, they wouldn't sometimes
state empirical contingencies.) So when we talk about truth by talking with some language, normally
we're not talking about any language. We're predicating a translinguistic property of a translinguistic
object.
67
Bibliography
Ashworth, E.J. (1974). Logic and Language in the Post-Medieval Period. Dordrecht: Reidel.
Buridan, J. (1986). Quaestiones in Porphyrii Isagogen. In: R. Tatarzynski, Jan Buridan, Kommentarz
Dunn, M. & Belnap N. (1968). The Substitution Interpretation of the Quantifiers. Nous 2 (2), 177-185.
Cappelen, H. & Lepore, E. (2007). Language Turned On Itself. New York: Oxford University Press.
Cappelen, H. and Lepore, E. (2012). Quotation. In E. N. Zalta (Ed.), The Stanford Encyclopedia of
Philosophy. http://plato.stanford.edu/archives/spr2012/entries/quotation.
Carter, M.G. (1981). The use of proper names as a testing device in Sibawayhi's Kitab.
Christensen, N. (1967). The alleged distinction between use and mention. Philosophical Review. 76(3),
358–67.
Church, A. (1954). Intensional Isomorphism and the Identity of Belief. Philosophical Studies 5(5),
68
65-73.
Coulmas, F. (Ed.), (1986). Direct & Indirect Speech. New York : Mouton de Gruyter.
Utterances. In A. Dey et al. (Eds.), Modeling and Using Context. Berlin: Springer.
Engelland, C. (2014). Ostension: Word Learning and the Embodied Mind, Cambridge, MIT.
Fara, D. (2011). You can call me 'stupid', ... just don't call me stupid. Analysis.71(3), 492-501.
Finnegen, R. (2011). Why Do We Quote: The Culture and History of Quotation, Cambridge,
Open Court.
Fonseca, P. (1564). Institutionum Dialecticarum Libri Octo. J. Ferreira Gomes (Ed.), (1964), Coimbra:
Universida de Coimbra.
Frege, G. (1892). On Sense and Reference, in P. Geach & M. Black (Eds.) (1952), Translations from
Frege, G. (1893). The Fundamental Laws of Arithmetic, in M. Furth (Ed.) (1964), The Basic Laws of
69
Frege, G. (1898?). Logical Defects in Mathematics. In H. Hermes, et al., (Eds.) (1969), Gottlob
Frege, G. (1918). Thoughts: A Logical Inquiry. In P. Geach & R. Stoothof (Eds.) (1977), Logical
Frege, G. (1924?). Logical Generality. In H. Hermes, et al. (Eds.), Gottlob Frege: Posthumous
Gallois, A. (2005). The Simplicity of Identity. The Journal of Philosophy.102(6), 273 – 302.
Houston, K. (2013). Shady Characters: The Secret Life of Punctuation, Symbols, and Other
Jackendoff, R. (1984). On the Phrase the Phrase 'The Phrase', Natural Language & Linguistic
Theory. 2(1), 25-37.
Kaplan, D. (1968). Quantifying In, Synthese.19(1/2), 178-214.
pages/anglistik/publikationen/inlist.
Kühner, R. & Gerth, B. (1898–1904), Ausführliche Grammatik der griechischen Sprache., Zweiter
Teil: Satzlehre 2.
Kvavik, K. (1986). Characteristics of direct and reported speech prosody: Evidence from Spanish. In
Coulmas (1986).
Ludwig, K. & Ray, G. (1998). Semantics for Opaque Contexts. Philosophical Perspectives.12, 141-66.
70
MacFarlane, J. (2000). What Does It Mean to Say That Logic is Formal?. Dissertation, University
of Pittsburgh.
Merchant, J. (2004). Fragments and Ellipsis. Linguistics and Philosophy. 27(6), 661-738.
Moltmann, F. (2013). Abstract Objects and the Semantics of Natural Language. Oxford: Oxford
University Press.
Olders, D. & Sas, P. (2001). Lifting the Church-Ban on Quotational Analysis. Journal for General
Parkes, M.B. (1993). Pause and Effect, Berkeley: University of California Press.
Partee, B. (1973), The Syntax and Semantics of Quotation, A Festschrift for Morris Halle, San
Pennington, H. (2007). The Problem with Biodiversity, London Review of Books, 29 (9), 31-2.
Quine, W. (1953). Three Grades of Modal Involvement. Proceedings of the XI International Congress
Quine, W. (1986). Autobiography. In L.E. Hahn & P.A. Schilpp (Eds.), The Philosophy of W.V.
71
Ramsey, F.(1931). The Foundations of Mathematics and Other Logical Essays, London: Routledge.
Reach, K. (1938). The name relation and the logical antinomies, Journal of Symbolic Logic. 3(2),
97-111.
Read, S. (1997), Quotation and Reach's Puzzle. Acta Analytica 19, 9-20.
Richard, M. (1986). Quotation, Grammar , and Opacity, Linguistics and Philosophy. 9(3), 383-403.
Salmon, N. (2002). The Very Possibility of Language. In M. Zeleny & C. A. Anderson (Eds.),
Spade, P. (1996). Thoughts, Words and Things: An Introduction to Late Medieval Logic and
72
Staal, F. (1975). The Concept of Metalanguage and its Indian Background, Journal of Indian
Tarski, A. (1932). The Concept of Truth in Formalized Languages. In J. H. Woodger, (Ed. & Trans.)
Tarski, A. (1965). Introduction to Logic and to the Methodology of Deductive Sciences. New York:
Wilkinson, R., et. al. (2010). Formulating Actions and Events With Limited Linguistic Resources:
Enactment and Iconicity in Agrammatic Aphasic Talk. Research on Language and Social