EROS CORAZZA
ON THE ALLEGED AMBIGUITY OF ‘NOW’ AND ‘HERE’
ABSTRACT. It is argued that, in order to account for examples where the indexicals ‘now’
and ‘here’ do not refer to the time and location of the utterance, we do not have to assume
(pace Quentin Smith) that they have different characters (reference-fixing rules), governed
by a single metarule or metacharacter. The traditional, the fixed character view is defended:
‘now’ and ‘here’ always refer to the time and location of the utterance. It is shown that
when their referent does not correspond to the time and/or location of the utterance, ‘now’
and ‘here’ work in an anaphoric way, inheriting their reference from another noun phrase.
The latter may be explicit or implicit in the discourse. It is also shown that ‘now’ and
‘here’ can inherit their reference from a presupposed or tacit reference. In that case, they
are coreferential with what will be labeled a ‘tacit initiator’. This anaphoric interpretation
has the merit of fitting within the Kaplanian distinction between pure indexicals (‘now’,
‘here’, ‘today’, etc.) and demonstratives (‘this’, ‘that’, ‘she’, etc.).
According to a widely accepted theory popularized by Kaplan (1977) and
Perry (1977), indexicals (‘I’, ‘today’, ‘this’, ‘now’, etc.) are directly referential expressions. As such, they contribute the referent itself to what
is said (i.e., the proposition expressed). Thus, an utterance containing an
indexical expresses a singular proposition. This theory has been labeled
the direct reference theory. It holds that an indexical expression has a fixed
linguistic meaning (character), which can be represented as a function from
context of utterance to referent (content). The character or meaning is,
roughly, a rule explaining how reference is fixed – it is what we find in
a dictionary for an indexical. Under the entry ‘I’, for instance, one reads
that it refers to the speaker or writer, while ‘today’ refers to the day it is
uttered, etc. If we change the context, we are likely to change the content.
‘I’ said by you does not refer to the same individual as when it is uttered
by me and ‘today’ uttered today does not refer to the same day as ‘today’
uttered yesterday. If today we desire to say what we said yesterday using ‘today’, we use ‘yesterday’. The character, however, remains constant
across contexts. Let us call this the fixed character thesis.
The fixed character thesis has been criticized. It has been argued
(Quentin Smith; 1989) that it cannot account for all the relevant data. We
use, for instance, present tense locutions to refer to past or future events.
Synthese 138: 289–313, 2004.
© 2004 Kluwer Academic Publishers. Printed in the Netherlands.
290
EROS CORAZZA
In these cases, a use of ‘now’ does not pick out the time of utterance. It
may be used to refer to a time differing from the time of the utterance. In
lecturing about the Nazi invasion of Paris, for example, a teacher can say,
“Now Hitler takes control of Paris”. In that case, ‘now’ does not refer to the
time of the utterance but to the time when the Nazis invaded Paris. We can
tell a similar story concerning ‘here’: ‘here’ in “It’s Monday September 11.
I’m walking around Yosemite; there are few grizzly bears here”, may pick
out neither the location where it was written nor where it is read (the writer
may have written this note when she arrived back home in San Francisco,
or she may simply be in her office writing it as a piece of fiction, for
example). ‘Here’ does not refer to San Francisco or the location where
the note is read, however; it refers to the Yosemite national park. If this
statement appeared in a novel, ‘here’ could even refer to an imaginary
location, which happens to be called ‘Yosemite’.
Examples like this prompt Quentin Smith to claim that indexicals do
not have a fixed character, for in different situations the same indexical expression may be used in rather different ways. An indexical character does
not remain constant across uses. Let us call this the multiple characters
thesis.
According to the multiple characters thesis, an indexical is an ambiguous expression, for it can have different meanings (i.e., the differing
characters).
In this paper I shall challenge the multiple characters thesis. To do so, I
shall focus on the indexicals ‘now’ and ‘here’ and argue that their alleged
ambiguity is best dealt with by adopting a framework that assumes the
fixed character thesis. In so doing I shall argue that when temporal indexicals like ‘now’, ‘today’ etc. and spatial indexicals like ‘here’, ‘there’, etc. do
not refer to the time and location of utterance, they work as anaphoric terms
inheriting their value from an antecedent to which they are linked. One may
argue that this is a multiple characters thesis too, for there are two characters: the usual one (the character of ‘now’ is represented by a function from
a context c to the time of c) and the anaphoric one (the character of ‘now’
is represented by a constant function yielding the semantic value of the
antecedent). On this view, however, the supposed indexical ambiguity of
‘now’ and ‘here’ reduces to an ambiguity between indexical use/meaning
and anaphoric use/meaning. This is similar to the ambiguity of pronouns:
‘s/he’ can be used either as a demonstrative (“S/he [pointing to someone]
is my best friend”) or as an anaphoric pronoun (“Janei1 thinks that shed1
is my best friend” – where the subscripts ‘1’ signal coreferentiality, while
the superscript ‘d’ signal that the NP is semantically dependent and the
superscript ‘i’ that the NP is semantically independent). However, this is
ON THE ALLEGED AMBIGUITY OF ‘NOW’ AND ‘HERE’
291
not the kind of ambiguity Quentin Smith has in mind. To put it slightly
differently, while I assume that indexicals are systematically ambiguous,
i.e., they can have both an indexical and an anaphoric reading, Smith assumes that they are accidentally ambiguous. The difference between the
position I defend and Smith’s can also be stated as follows. I claim that
indexicals like ‘now’ and ‘here’ play different grammatical roles. When
they are used as indexicals, they always pick out the time/location of the
utterance, while when they work as anaphors they inherit their reference
from another NP and, as such, they may pick out a time/location different
from the time/location of the utterance. On the other hand, Smith assumes
that ‘now’ and ‘here’ always play the same grammatical role but they are
ambiguous regarding their referents.
1. THE DATA AND THE MULTIPLE CHARACTERS THESIS
To begin with, let us consider cases of so-called historical time. It is
claimed that in these cases ‘now’ does not refer to the time of utterance
and thus that its character must have changed from that of ordinary uses of
‘now’. In an utterance such as:
(1)
Last Sunday, Jon visited his mother, now an old sick woman
‘now’ is likely to refer to the time of its utterance, not to the Sunday before,
whereas in an utterance like:
(2)
In 1834 Jon visited his mother, the once famous actress, now an
old, sick woman
‘now’ does not pick out the time of the utterance. Instead, it picks out the
time (1834) when Jon visited his mother. In this example, the time picked
out by ‘now’ is explicitly stated by the utterance, but in some cases it can
simply be presupposed.
Most interesting cases can be found with the historical present. If, during a history class discussing the Nazi invasion of the USSR, the lecturer
says:
(3)
Now Hitler begins his invasion of the USSR
‘now’ does not refer to the time of the utterance – it refers to the time of the
Nazi invasion. The time picked out, however, is not explicitly stated in the
utterance (and it need not even have occurred in previous utterances). If the
292
EROS CORAZZA
lecturer assumes that all her students know the period under discussion, she
does not need to mention the relevant date, i.e., June 22, 1941. If, whilst
watching a recording of one of the last European Champions’ League final,
one utters:
(4)
(5)
Now Bayern are dictating the pace
Now you see why Bayern won the game
the occurrences of ‘now’ do not pick out the same time. In (4), ‘now’ refers
to a time during the European Champions’ League final, whilst in (5) ‘now’
refers to the time of the utterance. In other cases, ‘now’ can also be used
to refer to imaginary or fictional times. In watching a movie containing
flashbacks, a member of the audience may be confused and ask:
(6)
Is this happening now?
i.e., is the event happening at the same imaginary time as other represented
situations (the time the movie represents as the present) or at a previous
imaginary time? One could answer:
(7)
No, it happened two years ago
meaning two years before the represented fictional time. In other cases,
‘now’ can even be used to refer to a nontemporal item. If, in the process of
proving a theorem, one states:
(8)
Now I prove lemma
‘now’ does not pick out the time of utterance.1 At this very moment one
may stop one’s lecture and continue proving the theorem during next week
lecture. In these cases ‘now’ seems to go proxy for something like ‘at this
point of the proof’.2
On the basis of examples like these, it is claimed that:
The data regarding these present tense indexicals . . . show these indexicals do not have an
invariable character, role or meaning. In some instances they are governed by the referencefixing rule that they refer to the time of their tokening, in other instances they obey the
reference-fixing rule that they refer to an earlier time, in still others to a nontemporal item
such as the point in the argument at which they are tokened. (Q. Smith 1989, 176)
The same (or a similar) phenomenon occurs with spatial indexicals such as
‘here’. If we consider the following passage from California: The Ultimate
Guidebook (Riegert, 1990):3
(9)
If an entire neighbourhood could qualify as an outdoor museum, the Mount Washington district would probably charge
admission. Here, just Northwest of downtown, are several
picture-book expressions of desert culture within a few blocks
ON THE ALLEGED AMBIGUITY OF ‘NOW’ AND ‘HERE’
293
‘here’ does not refer to the location of the author, Ray Riegert, when he
wrote (9), or to the location where the guidebook is read. It refers without
doubt to the Mount Washington district. Similarly, pointing to Paris on a
map one can say:
(10)
We spent last weekend here
and refer to Paris and not the location at which (10) is uttered. ‘Here’ can
also be used to refer to an imaginary location: watching a movie that takes
place on an imaginary distant planet, one can say:
(11)
The Martians’ ancestors first landed here.
In other uses, ‘here’ does not even refer or purport to refer to a spatial
location, as in (12):
(12)
Here Prokophiev always stops playing
for one is likely to mean that Prokophiev stops playing at that particular
point in the composition. Notice that one could reach the same goal in
using ‘now’ instead of ‘here’.
Examples such as these led Quentin Smith to distinguish between the
character of an indexical and a metacharacter or metarule. The former
(reference-fixing) rule varies with different uses, whereas the latter remains
constant:
These first-order rules are themselves governed by a second-order rule of use, a rule-fixing
rule of use or a metarule. It is not the reference-fixing rule of use that remains constant
from use to use, but the metarule. By remaining constant from context to context, the
metarule (or “metacharacter”) is able to determine which reference-fixing rule (character)
governs the indexical in each context. (Q. Smith 1989, 168)
In short, Quentin Smith assumes that the temporal indexical ‘now’ and the
spatial indexical ‘here’ are ambiguous. We have a metarule that helps us
choose which of the reference-fixing rules (i.e., which character) comes
into play in fixing the reference of the indexical in such-and-such a situation. A natural question could be whether Quentin Smith’s position can
be extended to account for words like ‘bank’. Do all ambiguous expressions have a metarule to decide which word is used in such-and-such a
situation? Could the account also be extended to homonyms? Is there a
metarule governing all the reference-fixing rules at work with the name
‘Aristotle’, i.e., selecting the reference-fixing rule for the philosopher or
the reference-fixing rule for Onassis?
One could claim that the multiple characters thesis is unmotivated, for
it suffices to assume that ‘now’ and ‘here’ are abbreviations of ‘this time’
294
EROS CORAZZA
and ‘this location’.4 Along these lines, one can argue that they have a fixed
character which is, nonetheless, less restrictive. If ‘now’ and ‘here’ mean
something like ‘this time’ and ‘this location’, then the examples proposed
can be dealt with in postulating a single character. The latter, though, would
be less limiting than the character of ‘now’ and ‘here’. In particular, the
relevant contextual parameters would not be the time and location of the
utterance, but some time and location picked out by ‘this’. According to
this suggestion one need not distinguish, pace Quentin Smith, between
characters (reference-fixing rules) and metacharacters (metarules) either.
This suggestion, though, blurs the traditional distinction between pure
indexicals and demonstratives. Kaplan (1977) distinguishes between pure
indexicals (‘I’, ‘now’, ‘today’, . . . ) and demonstratives (‘this’, ‘she’, . . . ).
The former, unlike the latter, do not need a demonstration to secure the
reference.5 To put it in a nutshell, while the reference of a pure indexical
is secured by the contextual parameters such as the agent, time and location of the utterance, the reference of a demonstrative is fixed by these
parameters plus something else, such as a pointing gesture, ostension or
demonstration.6 In his (1989) paper, Kaplan argues that a demonstrative
refers to what the directing intention directs us to.7 As I understand it, the
directing intention does not coincide with the individual one intends to talk
about, i.e., the individual one has in mind. If, in believing that the man in
front of me is Bush, I claim “That man is the president of the USA”, I have
Bush in mind and I intend to say something about Bush. If, unbeknownst
to me, the man in front of me is not Bush but Tony Blair, I refer to Tony
Blair regardless of my having Bush in mind. For my primary intention is
to refer to the man in front of me, not to the man I have in mind. It is
this communicative intention that I take to be the directing intention. For
this reason it may be better to label it directing attention. The practice of
directing attention can be viewed as resting on the communal practice of
using demonstrative expressions. One masters the use of a demonstrative
expression when one is able to exploit both the linguistic features and the
features available from one’s surroundings in order to single out an object
of discourse. In using ‘she’, for instance, a speaker expects to refer to a
female (this is given by the meaning of the demonstrative, i.e., by the fact
that ‘she’ is gender-sensitive). To single out the relevant object, the speaker
must also know how to exploit features available from the context in which
the linguistic interchange occurs, such as a physical gesture, a glance, a
previous remark, etc. In a word, when using a demonstrative expression the
speaker is responsible for that use, for fixing its reference; our competent
speaker makes the best use of the traits available from the context. With
pure indexicals, the situation is more straightforward insofar as what is
ON THE ALLEGED AMBIGUITY OF ‘NOW’ AND ‘HERE’
295
relevant in fixing the reference is given by the utterance itself. That is to
say, the utterance gives us the relevant contextual parameters such as the
agent, time and location. The latter are what indexical expressions direct us
to. When one uses a pure indexical, one automatically refers to the relevant
item. If one uses ‘today’, one refers to the relevant day regardless of one’s
intention. The simple act of uttering that word suffices to fix the reference.
If we assume, for the argument’s sake, that Kaplan’s distinction between
pure indexicals and demonstratives is sound, then ’here’ and ‘now’ are not
reducible to ‘this location’ and ‘this time’. The latter are demonstrative
terms referring to whatever the pointing gesture demonstrates or the directing attention singles out, while the former refer to the time and location
of the utterance. To be precise, the latter would be complex demonstratives,
i.e., expressions of the form ‘this/that F’.8 The multiple characters thesis
has the merit of fitting into the pure indexicals/demonstratives distinction.
As such, it deserves to be taken seriously. If one believes that pure indexicals reduce to (complex) demonstratives, however, one can avoid paying
attention to Smith’s position. Be that as it may, for the sake of my argument I invite you to assume the framework according to which there is a
principled distinction between pure indexicals and demonstratives.
Having said that, the multiple characters thesis does not seem to be a
very economical position. We should resist positing ambiguity beyond necessity: a position that accounts for these data without positing ambiguity
should be preferred. I will now propose a picture that deals with our data
and yet does not give up the fixed character thesis, thus avoiding the need
to posit ambiguity. In other words, in assuming the multiple characters
thesis, Quentin Smith’s position turns out to be more complicated that
the position I advocate. In particular, in positing metarules, the multiple
characters thesis seems to suggest that a competent speaker masters the
metarule and in a given context of use she is able to choose the relevant
character to fix the reference. The fixed character thesis, on the other hand,
simply suggests that a competent speaker masters the character of the indexical and this capacity is all one needs in order to competently use and
understand an indexical expression.
2. THE DATA AND THE FIXED CHARACTER THESIS
I am now going to show how the fixed character thesis can be defended and
yet account for the data proposed in favor of the multiple characters thesis.
In doing so, I shall defend the following anaphoric interpretation thesis:
296
•
EROS CORAZZA
AI
When a use of ‘now’/‘here’ does not select the time/location of
the utterance as referent, ‘now’/‘here’ works like an anaphoric
term inheriting its reference from another noun phrase.
In other words, when ‘now’ and ‘here’ are not used as indexicals picking
out the time and location of the utterance, they do not have referential independence; they work as anaphors inheriting their value from an antecedent
to which they are linked and coindexed. This can easily be illustrated if we
consider (2) [In 1834, Jon visited his mother the once famous actress, now
an old sick woman]. As we have seen, ‘now’ does not pick out the time
of the utterance; it refers to 1834. How does it pick out 1834? The most
natural answer is to claim that ‘now’ works like an anaphora, inheriting its
reference from ‘1834’. The reference of ‘now’ in (2) can thus be viewed
as depending upon the antecedent to which it is linked and thus coindexed
(‘1834’). On this view, we can represent (2) as:
(2)a.
In 1834i1 Jon visited his mother, the once famous actress, nowd1
an old, sick woman
where it is stressed how the reference of ‘now’ depends on the reference
of its antecedent, ‘1834’.9
Before going further, it may be worth emphasizing that, according to
the position I have in mind, a NP is always marked as (semantically)
independent or dependent, i.e., NPi/d . A proper name is always marked
as independent, while an indexical is ambiguous between being independent or dependent. If it is independent, then it is used deictically. If it is
dependent, then it is an anaphora (or a bound variable). In the latter case,
one must look for an antecedent. Within this framework, anaphoricity is
not represented by indices (subscripts), but by dependence/independence
markers (superscripts). In other words, I take the superscripts, not the
subscripts, to represent anaphoricity/non-anaphoricity, i.e., semantic dependence/independence, not coreference. To illustrate how coindexation
(and thus coreference) differs from anaphoricity, we can focus on identity
statements. In “Tully = Tully”, the two occurrences of the proper name
must be coindexed. This is constrained by the grammar of ‘=’. Think of
mathematical statements such as “2 + 3 = (2 − 1) + 4”, where the two
occurrences of ‘2’ must be coindexed. Yet they are not anaphoric upon
one another. The correct representation should be “2i1 + 3i2 = (2i1 − 1i3 ) +
4i4 ”. Since proper names are never anaphoric, they are always referentially
independent; yet they can be coindexed.
ON THE ALLEGED AMBIGUITY OF ‘NOW’ AND ‘HERE’
297
Let us now consider:
(3)
Now Hitler begins his invasion of the USSR
Since ‘now’ does not pick out the time of utterance it must, according to the
thesis I propose (AI), function as an anaphoric term; but then, where is the
antecedent? According to history, Hitler began his war against Russia on
June 22, 1941, so this fact should set the value of ‘now’; yet since there is
no term in our utterance explicitly referring to June 22 1941, where could
the value of ‘now’ come from? The solution that comes to my mind is
that ‘now’ inherits its reference from a tacit initiator, for it is presupposed
by the discourse situation that the facts described happened on June 22,
1941.10
The notion of tacit initiator I have in mind can be spelled out as follows.
An anaphora inherits its reference from the NP, the antecedent, to which
it is linked. The antecedent fixes the reference and initiates the anaphoric
chain. For this reason the antecedent can also be characterized as the initiator of the reference. An anaphoric chain too can be initiated by a tacit
reference, in which case the antecedent of the anaphora is not expressed.
As such, I characterize it as a tacit initiator. I could also characterize it
as a tacit antecedent. I adopt the following notation: if the antecedent is
explicit, the indexical has a non-zero number as a subscript, while if the
antecedent is implicit the subscript is a zero (NPd0 ).11
We can now return to our example (3). If the lecturer does not assume
that her students are aware of the importance of this date, she might instead
say:
(3)a.
It is June 22, 1941i0 . Nowd0 Hitler begins his invasion of the
URSS
where the bold ‘June 22, 1941’ can be viewed as the tacit initiator (for
this reason it get ‘0’ as a subscript) from which ‘now’ qua anaphoric term
inherits its value. It is a platitude that during a linguistic interchange a
great deal of information is unexpressed and is conveyed tacitly. Without
this phenomenon, communication would be extremely difficult and slow,
if not impossible, thus the existence of tacit initiators should not come as a
complete surprise. These initiators may be understood as what is taken for
granted during a speech act. If a cooperative speaker does not believe that
her audience is aware of this presupposed information, she would simply
express it in her utterance. Presuppositions are usually viewed as the “part
of discourse or speech” denoting propositions whose truth is taken for
granted. In uttering, for instance, “Jane’s hat is red”, one conveys two
298
EROS CORAZZA
propositions: the proposition that Jane has a hat and the proposition that it
is red. While the former proposition is presupposed, the latter is asserted.
In many cases, however, the presupposed propositions are merely taken
for granted and nothing in the utterance itself may trigger or bring them
to salience.12 As an approximation of the phenomenon I have in mind, I
follow Stalnaker when he argues that:
The notion of common background belief is the first approximation to the notion of pragmatic presupposition . . . A proposition P is a pragmatic presupposition of a speaker in
a given context just in case the speaker assumes or believes that P , assumes or believes
that his addressee assumes or believes that P , and assumes or believes that his addressee
recognizes that he is making these assumptions, or has these beliefs. (Stalnaker 1974, 49)
If the speaker does not assume that her audience is aware of the presuppositions in place, she can easily express them; thus, the tacit initiator could
be expressed by the speaker – this accounts for the difference between
(3) and (3a). (3) contains a tacit initiator, whilst (3a) contains an explicit
initiator, i.e., that which is presupposed in (3) is made explicit in (3a). Both
(3) and (3a) express the proposition that on June 22, 1941 Hitler begins his
invasion of the USSR.13
The same analysis applies to (4), which can be analyzed as:
(4)a.
It is the 2001 Champions’ League finali0 . Nowd0 Bayern are
dictating the pace.
Before going further, it is worth mentioning that one could object that
my claim (that ‘now’ and ‘here’ can work in an anaphoric way) does
not do justice to the way anaphors have been traditionally conceived,
especially within the generative grammar school. Kripke (1977), Lasnik
(1976) and Lewis (1979), to name but a few, could argue that the anaphoric interpretation I am proposing is nothing but a simple variant of
the indexical interpretation. For ‘now’ and ‘here’ in the example I discuss
work like indexicals picking out an object previously made salient. To be
precise, the authors mentioned do not make their point regarding ‘now’
and ‘here’.14 They are concerned with personal pronouns and claim that
when the pronoun ‘s/he’ is not bound, as in:
(13)
Jon won the lottery. He is now rich
it does not work like an anaphora. It merely works like a demonstrative
referring to an object previously made salient. In some cases, the referent
of a demonstrative such as ‘s/he’ is made salient by a pointing gesture. In
other cases, like in our example, it is made salient by a previous string of
the discourse. Yet, in some other cases, the relevant object may be made
ON THE ALLEGED AMBIGUITY OF ‘NOW’ AND ‘HERE’
299
salient by the setting in which the linguistic episode takes place. Let us
imagine, for instance, that Mary and Jane both dislike Jon and that they
are both aware of each other’s antipathy vis-à-vis Jon. During a meeting,
after several men (including Jon) have left, Mary can tell Sue that:
(14)
We can be relieved now that he has gone.
Given the discourse situation, ‘he’ refers to Jon even if Jon is not present
to be demonstrated or has not been previously raised to salience by the use
of a NP.
It should be stressed that if one adopts the Kripke-Lasnik-Lewis attitude
to the case of ‘now’ and ‘here’, one undermines the distinction between
pure indexicals and demonstratives. As we have seen, the character of a
demonstrative is incomplete and the demonstrative refers to whatever the
act of directing attention directs us toward. A pure indexical, on the other
hand, does not need a pointing gesture or an act of directing attention to
select a referent. The utterance itself gives us all the relevant contextual
parameters (time, location and agent) needed to secure their reference:
reference is automatic. If one claims that ‘now’ and ‘here’ refer to a time
and location previously raised to salience in the example I proposed, then
‘now’ and ‘here’ do not refer to the time and location of the utterance.
As such, they do not work like pure indexicals, but instead like demonstratives. The position I am proposing, on the other hand, can handle the
examples discussed without giving up the pure indexical/demonstrative
distinction.
Before going further, it may be useful to stress that, unlike the third
person pronoun ‘s/he’, ‘now’ and ‘here’ cannot act as bound anaphors.
They cannot behave as bound variables either. Consider:
(15)a. Jane won the lottery and she is now rich
b. Jane thinks that she is rich
In (15a) ‘she’ is not bound by ‘Jane’, whereas it is in (15b). To stress this
point we can replace ‘Jane’ by a quantifier. We would thus have:
(15)c. Everyone won the lottery and she is now rich
d. Everyone thinks that she (herself) is rich
In (15c) ‘she’ is not bound by the quantifier ‘everyone’: it works like a free
variable, while in (15d) it is bound and works as a bound variable.15 If we
consider ‘here’ and ‘now’, they are never bound the way ‘she’ can be in
(15b). Thus, if ‘here’ and ‘now’ work as anaphoric terms, they are unbound
300
EROS CORAZZA
anaphors.16 Some pronouns cannot be bound either. Yet, the intuitive and
default interpretation of the plural pronouns ‘them’ and ‘we’ in:
(16)a. Jon said that Sue expected them to win the competition
b. I told Jane that we won the lottery
is the anaphoric one. The natural and default interpretation of ‘them’ in
(16a) is that it is anaphoric on ‘Sue’ and ‘Jon’, while the default interpretation of ‘we’ in (16b) is that it is anaphoric on ‘I’ and ‘Jane’. Yet, they are
unbound – pronoun can never be bound by split antecedents.17 They are
unbound anaphors: (16a, b) could be represented as:
(16)c. Joni1 said that Suei2 expected themd1+2 to win the competition
d. Ii1 told Janei2 that wed1+2 won the lottery
To be sure, one could still claim that even in this case ‘them’ does not
work in an anaphoric way, but instead works as a demonstrative selecting Jon and Sue as referent because they have previously been raised to
salience. The same consideration would apply for ‘we’ in (16b). In that
case, a representation like (16c) and (16d) would not be appropriate, for
the superscript ‘d’ signals that the pronoun is referentially dependent, i.e.,
that it inherits its reference from another/others NP(s).
Needless to say, in the case of unbound anaphora the anaphoric chain
is not merely syntactically constrained: semantic and pragmatic considerations also enter into the picture.18 On this issue one can quote recent
studies, notably Huang (2000), which convincingly show that:
(i) syntax, and pragmatics are interconnected to determine many of the processes of
anaphora that are thought to fall within the province of grammar, and (ii) the extent to
which syntax, semantics, and pragmatics interact varies typologically. (Huang 2000, 205)
There seems to exist a class of languages (such as Chinese, Japanese, and Korean) where
pragmatics appears to play a central role which in familiar European languages (such as
English, French, and German) has hitherto been alleged to be played by grammar. In these
‘pragmatic’ languages, many of the constraints on the alleged grammatical processes are in
fact primarily due to principles of language use rather than rules of grammatical structure.
(Huang 2000, 213)
Be that as it may, if one follows my proposal and accepts the existence
of unbound anaphora inheriting their reference from another NP, one can
maintain that ‘now’ and ‘here’ are systematically ambiguous: they are
either pure indexicals or unbound anaphors. On the other hand, one could
claim that coreferentiality (and thus coindexation) in the examples I proposed is somewhat accidental, i.e., the relevant NP is coreferential with
ON THE ALLEGED AMBIGUITY OF ‘NOW’ AND ‘HERE’
301
another NP not in virtue of inheriting its reference but simply because the
latter has been previously raised to salience (they are thus referentially
independent). One is then forced to embrace the thesis that ‘now’ and
‘here’ in the example discussed, like all alleged unbound anaphors, actually work like demonstrative expressions selecting as referent a location
and time differing from the time and location of the utterance. To summarize, the anaphoric interpretation I proposed, unlike the demonstrative
interpretation: (i) maintains that there is a structural difference between
unbound anaphora and demonstrative reference (the former is referentially
dependent while the latter is referentially independent) and (ii) fits within
the distinction between pure indexicals and demonstratives. Besides, as
we shall soon see, the position I advocate does not commit itself to the
view that ‘now’ and ‘here’ are contextually ambiguous. The anaphoric
interpretation is thus cheaper than the demonstrative interpretation. Concerning tacit initiators it should also be pointed out that there are many
potential initiators that could successfully fulfill the communicative aim.
The position I put forward can also be stated as follows: utterances like (3)
and (4) are shortcuts to utterances like (3a) and (4a). Indeed, there is no one
single way in which the former utterances can be successfully completed.
(4) could also be successfully completed as:
(4)b.
It is May 15 2001 the Champions’ League final matchi0 .
Nowd0 Bayern is dictating the pace
Before going further it is worth mentioning that one could claim, along
with Predelli, for instance, that the relevant contextual parameter to which
the character of ‘now’ and ‘here’ is sensitive is not the location and time
of the utterance, but an intentional time and/or location. One could thus
maintain that the referent of ‘now’ and ‘here’ corresponds to the intended
contextual parameter. In that case, the character takes as argument a relevant contextual aspect and gives as value the relevant (intended) referent.
Thus, the contextual parameters to which ‘now’ and ‘here’ are sensitive
may differ from the time and location of the utterance and so their referent may also differ from the location and/or time where the utterance
occurs.19 Although this position does not distinguish between the indexical
and the anaphoric use of ‘now’ and ‘here’, like the position I attribute to
Kripke, Lasnik and Lewis, commits itself to the view that a single token
of ‘now’/‘here’ may be sensitive, in principle, to infinitely many contexts
(times and locations). Since the relevant time and location qua parameters
of ‘now’ and ‘here’ may differ from the time and location of the utterance,
they must be contextually furnished. Following this suggestion, ‘now’ and
302
EROS CORAZZA
‘here’ can be viewed as contextually ambiguous. The ambiguity of the
indexicals is not at the level of character but at the level of context.
The position I propose presents the following advantage over the position advocated by Predelli and the position I attribute to Kripke, Lasnik
and Lewis. On my view ‘now’, ‘here’ (and similar indexicals) can be used
either as indexicals or as anaphoric terms. As such, they are systematically
ambiguous, but their ambiguity rests on their being used either deictically
or anaphorically. It does not rest on a multiplicity of contexts.20 Hence,
the position I advocate multiplies neither contexts nor characters. It simply
rests upon an existing common phenomenon, the phenomenon I characterized as unbound anaphora (sometimes labeled cross-sentential or discourse
anaphora). For instance, “Maryi1 believes that shed1 is lucky” contains an
intra-sentential anaphora, whilst in “Joni1 won a million dollars last night
in Monte Carlo. Even though hed1 won all that money, hed1 was still not very
pleased”, we have a case of cross-sentential or discourse anaphora, i.e.,
the anaphor ‘he’ inherits its value from an antecedent, ‘Jon’, appearing in
another sentence.
Since the picture I defend appeals to existing phenomena, it comes
more cheaply than the position claiming that indexicals are contextually
ambiguous. Thus, the very same argument I used against Smith’s position (which multiplies characters and thus violates Ockam’s razor by
positing unnecessary ambiguities) can be used against the Kripke-LasnikLewis-Predelli position. The latter views posit ambiguities by multiplying
contexts. To be sure, this position accepts the fixed character view, but
it handles the data in multiplying the contexts that these characters can
operate upon. The relevant context may thus differ from the context of
utterance. The moral is that the Kripke-Lasnik-Lewis-Predelli view also
violates a principle of parsimony and Ockam’s Razor, by multiplying contexts and ambiguities beyond necessity (just as Smith’s view multiplies
characters and ambiguities beyond necessity). To put it in a nutshell, the
conception I propose presents an advantage over the Kripke-Lasnik-LewisPredelli position by being more economical and positing less ambiguity
(the very same advantage it has vis-à-vis Smith’s multiple characters view).
As I have already mentioned, if our aim is to understand and explain
communicative interaction, we must take into consideration discourse anaphora, which may extend beyond utterances of a single speaker. Hence we
should also consider interpersonal anaphora, for it is not uncommon that,
in every day communication, anaphors that inherit their reference from an
antecedent uttered by someone else are often used. Consider:
(17)
Mary: “I’ll be in Parisi1 on New Year Eve”
Jon: “I’ll be thered1 anxiously waiting for you”
ON THE ALLEGED AMBIGUITY OF ‘NOW’ AND ‘HERE’
303
In this case, ‘there’ inherits its reference from a token made by someone
else, Mary. If we consider a telephone conversation such as:
(18)
Mary (in Paris): “It is still snowing”
Jon (in Lyon): “In that case, I won’t be able to be thered0 before
midnight”
we have ‘there’ inheriting its reference, Paris, from another phrase. In
that other phrase, though, the reference to Paris is not explicitly made –
it is for this reason that the pronoun ‘there’ has ‘0’ as a subscript. Mary
tacitly refers to Paris. To borrow Perry’s (1986) terminology, Paris is an
unarticulated constituent of what Mary says.21 If Mary uttered, “It is still
snowing here”, Paris would be an articulated constituent of what she says
and ‘there’ would be anaphoric and coreferential with ‘here’. Cases like
this suggest that we do have anaphoric terms inheriting their value from
a tacit initiator. I do not see how (18) could be interpreted without taking
‘there’ to be anaphoric on (and coreferential with) a presupposed tacit term
referring to Paris.
Cases where anaphors inherit their value from tacit initiators are not
confined to spatial and temporal indexicals. In:
(19)
Jane is hoping for a baby, whereas hed0 is far from exited by the
idea
‘he’ seems to inherit its value from a tacit initiator (NP like ‘Jane’s partner/husband/. . . ’ would suffice), for the speaker of (19) tacitly refers to
Jane’s partner. ‘Jane’ is relevant in helping the audience to select Jane’s
partner/husband/. . . as the object of discourse and thus in bringing to
salience the referent the tacit initiator stands for.
The same analysis applies to cases of reference to an imaginary time,
such as (6). Treating the ‘now’ that picks out an imaginary time as anaphoric on an antecedent referring to an imaginary time, (6) can be analyzed
as:
(6)a.
In the play, there is an imaginary time ti0 . Is this happening
nowd0 ?22
Notice that the imaginary time need not be specified, for it can be
quantified over.
304
EROS CORAZZA
A similar story can be told about ‘here’. In (9), ‘here’ is anaphoric on
‘the Mount Washington district’ and (9) could thus be represented as:
(9)a.
If an entire neighborhood could qualify as an outdoor museum, [the Mount Washington district]i1 would probably charge
admission. Hered1 , just Northwest of downtown, are several
picture-book expressions of desert culture within a few blocks
The same goes for cases like (11), where the speaker refers to an imaginary
location. (11) can be interpreted as:
(11)a. This/that is the imaginary planet XYi0 . Hered0 is where the
Martians’ ancestors first landed.
To stress this anaphoric interpretation, one can also claim that in such cases
‘here’ goes proxy for ‘there’ and that (11a) is equivalent with (11b):
(11)b. This/that is the imaginary planet XYi0 . Thered0 is where the
Martians’ ancestors first landed.
(11a) and (11b) express the proposition that the Martians’ ancestors first
landed on the imaginary planet XY.
In the case of pointing to a map (say pointing to the spot representing
Paris) and saying ‘here’, one can think of ‘here’ as going proxy for ‘this
location’ and thus that ‘here’ is used as a demonstrative picking out the
demonstrated location, instead of picking out the location where the utterance occurs. The anaphoric interpretation, however, allows us to reach
the very same conclusion without supposing that ‘here’ works as a demonstrative. For it can be argued that ‘here’ works as an anaphora, and that an
utterance like (10) should be analyzed as:
(10)a. This/that location is/represents/. . . Parisi0 . We spent last weekend hered0 23
Similarly, we can stress the anaphoric reading, taking (10a) as equivalent
to (10b):
(10)b. This/that location is/represents/. . . Parisi0 . We spent last weekend thered0
where ‘there’ stresses the anaphoric interpretation. (10a) and (10b) express
the proposition that we spent last weekend in Paris.
ON THE ALLEGED AMBIGUITY OF ‘NOW’ AND ‘HERE’
305
When ‘now’ and ‘here’ are not used to pick out a time and location,
they belong to the idiomatic use of language, i.e., that peculiar use which
cannot be captured by a general rule. As such they do not work in their
usual, indexical, way to single out the time/location of the utterance. It
is for this reason that ‘now’ and ‘here’ are often interchangeable without
any loss or gain in a communication. My New Shorter Oxford Dictionary
reads “at this point in an argument, a situation, etc.; at this juncture” under
the entry for ‘here’, while the entry for ‘now’ reads, “at an important or
noteworthy place in an argument or proof or in a series of statements”.24
Again, even in these cases, ‘now’ and ‘here’ can be interpreted as anaphors;
(8) and (12) can be analyzed as:
(8)a.
We are at this point of the argument/proof/. . . i0 . Now/hered0 I
am going to prove lemma
(12)b. We are at this point of the sonata/concert/. . . i0 . Here/nowd0
Prokophiev always stops playing
It is worth noticing that, if one favors the multiple characters thesis, one
must be able to spell out the different characters qua reference-fixing rules
at play in different examples. Since there are (logically) infinitely many
such examples, one ends up with infinity many characters. How can we
supply all the characters needed to account for the variegated uses of ‘now’
and ‘here’? It could be objected that the multiple characters thesis does not
presuppose a different reference-fixing rule (character) for each specific
example. Actually, Smith claims that the metarule delimits sorts of contexts in such a way that for each context delimited a corresponding fixing
rule is also delimited. As an example, we can quote historical context. In
that case a single fixing-rule suffices.25 Along these lines one can claim that
there are only finitely many fixing-rules. The characters vary according
to each different kind of example, not with each specific example. The
first question that comes to mind is how to group infinitely many possible
examples into kinds. That is to say, how do we group the various examples
together? Even in assuming that we have a definite number of kinds and
that each specific case can easily be classified as one kind or another, a
question remains to be answered. How could one possibly come to master
the use of these indexicals? One could, following Quentin Smith, argue
that there is a single metarule or metacharacter that must be mastered and
that it is this fixed rule that selects, in different occasions of use, the character qua reference-fixing rule. A question seems to remains: how does this
metacharacter select the reference-fixing character? Smith’s answer should
be: by directing the speaker/hearer toward the kind of context which will
306
EROS CORAZZA
enable them to select the relevant character. This picture commits itself to
the view that a competent speaker must master several rules governing the
use of indexicals. This may not be a knock down argument, but a picture
suggesting that a competent speaker must master a single rule or character
should be preferred insofar as it would be cognitively more economical.
Following the Kaplanian picture, an indexical’s character can be represented as a function taking as argument the context and giving as value
the content or referent. In the case of ‘I’, for instance, its character will be
represented by a rule such as: “The referent of ‘I’ is the speaker and/ or
writer”. In a more formal way:
•
C-‘I’: an individual i is the semantic value of ‘I’ with respect to
a context of utterance c iff i is the agent of c,26
whilst the character of ‘now’ and ‘here’ can be represented by the rules:
•
C-‘now’:
a time t is the semantic value of ‘now’ with respect to a context
of utterance c iff t is the time of c
•
C-‘here’:
a location l is the semantic value of ‘here’ with respect to a
context of utterance c iff l is the location of c
In all other cases, when the referent of ‘now’ and ‘here’ is not the time
or location of the context of utterance, ‘now’ and ‘here’ do not work as
indexicals, but instead as anaphors. This fact should not come as a surprise,
for the pronouns ‘he’, ‘she’, ‘they’, etc. can also work both as indexicals
and as anaphors. In:
(20)
She [pointing toward Jane] is my favorite student
‘she’ is an indexical, while in:
(21)
Janei1 is very bright; shed1 is my favorite student
‘she’ acts as an anaphoric pronoun and is thus coreferential with ‘Jane’,
the antecedent, from which it inherits its semantic value.
ON THE ALLEGED AMBIGUITY OF ‘NOW’ AND ‘HERE’
307
3. SOME ADVANTAGES OF THE FIXED CHARACTER THESIS OVER THE
MULTIPLE CHARACTERS THESIS
We can summarize the competing positions I presented as follows.
While the multiple character thesis distinguishes between character qua
reference-fixing rule and metacharacter, the fixed character position distinguishes between indexical uses and anaphoric uses and further claims
that this distinction is not confined to pronouns; it applies to the whole
spectrum of indexicals.
A central problem for the multiple characters position is to spell out
what the metacharacter or metarule of ‘now’ and ‘here’ is:
The fact that each of these rules and the corresponding sort of context are delimited by
the metarule implies that the metarule is considerable complex. This need not daunt us,
however, for the mastery of this metarule by a language-user does not require the user to
be able to explicitly formulate or verbalize this metarule. Clearly we do not “run though
in our mind” this metarule on each occasion that ‘now’ is used. Rather, this metarule
is implicitly comprehended and its comprehension is normally evinced by our ability to
determine correctly which reference-fixing rule governs ‘now’ on any particular occasion
of use, a determination that is itself evinced by our grasp of the referent of ‘now’ on that
occasion. (Q. Smith 1989, 178)
A competent speaker need not be able to spell out the linguistic rules she
follows, for her linguistic capacity or mastery is manifested in the way she
actually (successfully) follows these rules. Nonetheless, the semanticist or
linguist may wish to spell them out. A story must be told to explain why,
in a given occasion, the metarule selects one specific reference-fixing rule
and not another. This task, too, may turn out to be more complicated that
at first thought, for there are many reference-fixing rules.
If one adopts the fixed character picture I propose, one does not need
to posit metarules or multiple reference-fixing rules. A single simple rule,
such as those presented at the end of the previous section, suffices. There is
only one other way a temporal indexical like ‘now’, or a spatial indexical
like ‘here’, can work: when ‘here’ and ‘now’ do not select as referent the
time/location of the utterance, they work as anaphoric terms. It is not difficult, however, to propose grammatical rules governing anaphors, for the
very same rules governing unbound anaphoric uses of ‘she’, ‘it’, ‘they’,
etc. will also govern anaphoric uses of ‘now’ and ‘here’. In our semantics,
we already have all we need.27 Without entering into detail, we can say
that there are simple constraints at work (such as the one I indicated in
introducing the superscript ‘d’ to signal semantic dependence).
One could object that, while the multiple characters thesis friend posits a complicated metarule and several reference-fixing rules, I posit tacit
initiators, for in many utterances there is no expressed NP from which
308
EROS CORAZZA
the alleged anaphora can inherit a semantic value. This may be seen as
the major problem I face; perhaps the fixed character position I defend
does not come cheaper than the multiple characters thesis defended by
Quentin Smith. But is this really an insurmountable difficulty? In particular, would the existence of tacit initiators make the fixed character
position I present less economical than the multiple characters position,
which posits a metarule and several reference-fixing rules?
It is worth emphasizing that the multiple characters position assumes
that, in a specific situation, the speaker and hearer can easily select the
reference-fixing rule at work. This capacity is manifested in the way the
speaker and her audience succeed in singling out the relevant referent. I
have no objection to this assumption; indeed, most of the time the speaker
and audience do not face any difficulty individuating the relevant referent
and succeeding in their communicative interaction. However, the question
remains: do they succeed because they tacitly master a metarule or do they
succeed simply because the relevant information is presupposed and taken
for granted? Is their success due to selection of the right reference-fixing
rule or because the presence of a tacit initiator, from which ‘now’ and
‘here’ qua anaphoric terms inherit their reference, is taken for granted?
After all, if we ask the speaker and audience to make explicit what they
tacitly presupposed, they certainly end up uttering the tacit initiator or a
similar, equivalent, term. The analysis I have proposed captures this fact
in an elegant way, for the tacit initiator is a NP referring to the object the
speaker and her audience intend to talk about.28 I do not commit myself to
the view that there must be a specific tacit initiator. There are and can be
many tacit initiators selecting the relevant referent.
The advantage of the fixed character view I defend over the multiple
characters position can be summarized as follows: the fixed character position does not appeal to several reference-fixing rules. All that needs to be
spelled out are the reference-fixing rules for ‘now’ and ‘here’ and an appeal
to a theory of discourse anaphora. We do need the latter for independent
reasons; we need a theory of discourse anaphora to handle cases where, in
a discourse, people keep track of objects and individuals using anaphors.
I have not spelled out the details of a theory of discourse anaphora, for
it suffices to say that a correct theory would do the job required. My
argument regarding the introduction of tacit initiators to handle the data
proposed in favor of the multiple characters thesis does not rest and does
not presuppose a specific theory of discourse anaphora.
Both the rules governing indexicals and those governing anaphors are
needed in order to account for the phenomenon of indexicality in general.
Since we already have in place the distinction between indexicals and
ON THE ALLEGED AMBIGUITY OF ‘NOW’ AND ‘HERE’
309
anaphors, we do not have to complicate our semantics by assuming that
temporal and spatial indexicals such as ‘now’ and ‘here’ force a distinction between metacharacter and character. These indexicals are not, pace
Quentin Smith, ambiguous expressions. They have a simple and single
character. In some cases, however, they work as anaphoric terms. This
notion, as we have seen, is all we need to account for the data advanced in
favor of the multiple characters position.
ACKNOWLEDGEMENTS
An early version of this paper was presented at CREA (1, rue Descartes,
75005, Paris), at the University of Nottingham and at the University of
Budapest (in early spring 1999). Due to subsequent teaching and administrative duties, I have been unable to work on this subject until recently,
whilst on sabbatical (Spring to Autumn 2002) partly supported by the
AHRB research leave scheme. Thanks to the AHRB for the support.
Thanks also to the audiences for comments. For further comments and
discussions, I would like to thank Jérôme Dokic, Gábor Forrai, Jonathan
Gorvett, Adriano Palma, Jérôme Pelletier, François Recanati and Mark
Whitsey. This paper has also benefited from extensive comments and
criticisms by three anonymous referees for Synthese.
NOTES
1 Whether one takes the time of the utterance to correspond with the time of writing or
the time of producing and/or decoding the message does not matter. In this example ‘now’
picks out none of these times.
2 Similar examples can be found in Q. Smith (1989).
3 I am borrowing this example from Predelli (1998).
4 As a referee for this journal did.
5 Hintikka formulates the distinction as follows:
[T]he reference of a type two demonstrative [i.e., what I simply call demonstratives] is not
fixed by the parameters that suffice to specify the occasion in question, that is to say, by a
person and his or her spatiotemporal vantage point. (Hintikka 1998, 208)
6 I spelled out and defended the distinction between pure indexicals and demonstratives
in some detail in Corazza (2002). I argued in favor of three main features that distinguish
them: (i) The use of a pure indexical, unlike the use of a demonstrative, never requires a
pointing gesture to fix the reference, (ii) The use of a pure indexical, unlike the use of a
demonstrative, is not perception-based and (iii) Pure indexicals, unlike demonstratives, are
never vacuous terms.
310
EROS CORAZZA
7 “The directing intention is the element that differentiates the ‘meaning’ of one syntactic
occurrence of a demonstrative from another, creating the potential for distinct referents,
and creating the actuality of equivocation.” (Kaplan 1989, 588)
8 I discuss complex demonstratives and show how they differ from simple demonstratives
in Corazza (2002a).
9 It is worth mentioning that, while coindexation entails coreference, non-coindexation
does not entail non coreference. If, in answering the question “Who left?” one says:
(i)
He1 put Jon2 ’s coat on,
what is said may well be consistent with ‘he’ and ‘Jon’ being coreferential (see Fiengo and
May 1984, 3).
10 If the lecturer believes that the USSR invasion occurred on June 22, 1942, on the account
I have in mind we may face two possibilities. If the lecturer utters ‘1942’ ‘now’ refers to
1942, but if left it silent she refers to 1941, since this is the presupposed date. This does not
undermine my picture insofar as I am not committed to the thesis that a speaker’s intention
determines reference. More on this later.
11 So in our syntax we can represent the whole range of indexical NPs as follows:
Dependence + (non-zero) number index ⇒ anaphora with explicit initiator/antecedent (NPdn )
Dependence + zero index ⇒ anaphora with tacit initiator/antecedent (NPd0 )
Independence + (non-zero) number index ⇒ deictic term (NPin )
12 For a convincing and detailed account of presuppositions, see Stalnaker: “The distinction between presupposition and assertion should be drawn, not in term of the content of
the proposition expressed, but in terms of the situation in which the statement is made – the
attitudes and intentions of the speaker and his audience. Presuppositions, on this account,
are something like the background beliefs of the speaker – propositions whose truth he
takes for granted, or seems to take for granted, in making his statement.” (Stalnaker 1974,
48)
13 One could object that the notion of a tacit initiator qua anaphora antecedent cannot
be explained in terms of pragmatic presuppositions, for the notion of the antecedent of
an anaphor belongs to semantics, not pragmatics. This objection (raised by a referee of
this journal) rests on the idea that we have a clear-cut distinction between semantics and
pragmatics. I am not sure, however, that such a distinction is operative, or welcomed, when
one deals with natural language. Actually, if one follows the ordinary language school
inaugurated by Wittgenstein, one is keen to take on board features concerning the way
linguistic expressions are used. Within this camp, semantics can be understood as the study
of the rules for the correct use of language and the meaning of an expression type is determined by the ways in which it can be correctly used. Semantics, so understood, takes into
account the speaker’s competencies or, at least, an ideal speaker’s linguistic competences.
I am not sure that a clear-cut distinction between semantics and pragmatics can be made
at all within this framework. I tend to believe that semantics is pragmatically informed
and vice versa; to paraphrase a famous Kantian dictum, semantics without pragmatics is
empty, while pragmatics without semantics is blind. To put it more crudely, I suspect that
the semantic/pragmatic divide is neither a welcome nor a healthy distinction. At least, it
is unhelpful when one focuses on the study of natural or ordinary language. For these
ON THE ALLEGED AMBIGUITY OF ‘NOW’ AND ‘HERE’
311
reasons I believe that the notions of rules and conventions are more promising – or at least
more primitive – when trying to analyze the meanings of linguistic expressions. For these
very reasons, I do not believe that the notion of an anaphora’s antecedent (either tacit or
explicit) can be dealt with without taking into consideration issues that have traditionally
been considered to be non-semantic. Moreover, it has been argued (Huang 2000) that an
exhaustive study of anaphora cannot be done in purely semantic or syntactic terms, insofar
as an anaphora’s antecedent cannot be selected uniquely on the basis of syntactic and
semantic considerations. Pragmatics factors often operate in determining the antecedent
of an anaphor. This is particularly evident when one concentrates on Asiatic languages,
which rely more heavily on pragmatics than English, Italian, French, etc. I discuss this
issue in more detail below.
14 As we shall soon see Predelli (1998) can be interpreted as adopting this stance to the
case of ‘here’ and ‘now’.
15 Actually, (15c) and (15d) would be regimented as follows:
(15)e.
f.
∀x (x won the lottery) & y is now rich
∀x (x thinks that x is rich)
16 In particular, they do not fall into the category of anaphors according to the principles
A, B and C of Binding Theory:
Principle A: anaphors must be bound in their governing category.
Principle B: pronouns must be free in their governing category.
Principle C: other NPs must be free in all categories.
17 “Although pronouns with split antecedents are coindexed with those antecedents, they
are never bound by them. This predicts that the distribution of split antecedents will not be
limited by Binding Theory.” (Fiengo and May 1994, 39)
18 As an example of a pragmatic constraint on anaphora we can quote the general pattern of anaphora, i.e., the fact that “reduced, semantically general anaphoric expressions
tend to favour locally coreferential interpretations; full, semantically specific anaphoric
expressions tend to favour locally non-coreferential interpretations” (Huang 2000, 214).
As examples, consider the following contrasts:
(i)a.
b.
d
Mozarti1 adored his1/2 music
d
Hei1 adored Mozart’s2/∗1 music
(ii)a.
The busi1 came trundling round the bend. The vehicled1 almost flattened a
pedestrian
b.
The vehiclei1 came trundling round the bend. The bus2/∗1 almost flattened a
pedestrian
d
19 Predelli (1998) defends a position similar to this. On his account, ‘now’ and ‘here’ may
not refer to the time and location of utterance, for they can be sensitive to an intentional
context, which may differ from the context of utterance. Following Predelli’s position, the
characters of ‘now’ and ‘here’, for instance, are sensitive to intentional parameters. He thus
rejects the thesis that contextual parameters such as time and location are always identical
with the time and location of the utterance.
312
EROS CORAZZA
20 Insofar as the present tense is interpreted using ‘now’, we can claim that it can work in
an anaphoric way as well. If, watching a video, one claims that “Arsenal are dominating
the game”, the present tense can be viewed as anaphoric on a tacit initiator referring to the
time the game was played. No doubt, more should be said on this issue, but time and space
prevent me from discussing it any further.
21 I discuss the notion of unarticulated constituents and Perry’s treatment of the phenomenon in Corazza (2003, forthcoming). Although I refer to Perry’s work, the view I am
putting forward is independent of whether one accepts the direct reference framework of
singular propositions championed by Parry, Kaplan and others. For an exhaustive criticism
of the direct reference picture, see Hintikka and Sandu (1995).
22 One could claim that (6a) is unnatural and artificial and that ‘now’ should be replaced
by ‘then’. In that case the fact that the original ‘now’ in (6) works like an anaphora is even
more striking.
23 We saw that in the case of (6a), if we replace ‘now’ by ‘then’ the reading is more natural.
The same thing happens here. A use of ‘there’ instead of ‘here’ sounds more appropriate.
Again, we would thus stress how the original ‘here’ in examples like (10) is best interpreted
as an anaphora.
24 If I am right in considering these examples as idiomatic uses of language, it may be
better not to treat them in the same way as the other examples. It is by mere accident that
we use spatio-temporal terms to describe non-temporal items. Is it simply because it takes
time to run through a proof or a sonata that we adopt the temporal framework. When the
proof and sonata are written, we adopt the spatial framework for similar reasons. These
phenomena are linked to our cognitive architecture; they are not intrinsic to the semantics
of our language. Thanks to Predelli for this suggestion.
25 In that case the reference-fixing rule for ‘now’ would be: “ ‘now’ refers to the historical
time the context indicates the speaker/writer wishes to emphasis and take as the chronological point of reference of the event reported in the other relevant portions of the discourse.”
(Smith 1989, 177)
26 It may be worth mentioning that the agent may not be the writer or speaker. In the case
of answering machines and similar devices, the agent need not be the writer/speaker of the
playback message. One may, for instance, use an answering machine message recorded
by someone else, or one can use a post-it note written by someone else. In cases like
these, the time and location of the utterance are rarely the time and location the token is
produced; one makes an utterance at a distance or, to borrow Sidelle’s (1991) terminology,
one makes a deferred utterance. In the case of an answering machine, the time of the
utterance corresponds with the time the recorded message is played back and the location
corresponds to the location it is played, while in the case of post-it notes the location/time
of utterance corresponds to the location/time the note is read. In a post-card, on the other
hand, the time picked out by the indexical is likely to be the time of the production of the
message. For an exhaustive discussion of these specific issues see Corazza et al. (2002).
27 For the sake of my argument I do not need to commit myself to a specific theory of
anaphora. A satisfactory theory of cross-sentential or unbound anaphora will do the job
required.
28 As we saw, though, the referent may not correspond with the object the speaker and/or
audience intend to talk about. If one falsely believes that today is the first of May 2002,
in using ‘today’ or tacitly referring to the day of utterance one picks out the day of the
utterance even if the latter differs from the day one intends to talk about.
ON THE ALLEGED AMBIGUITY OF ‘NOW’ AND ‘HERE’
313
REFERENCES
Corazza, E.: 2002, ‘Temporal Indexicals and Temporal Terms’, Synthese 130(3), 441–460.
Corazza, E.: 2002a, ‘ “She” and “He”: Politically Correct Pronouns’, Philosophical
Studies 111(2), 173–196.
Corazza, E.: 2003, ‘Thinking the Unthinkable: An Excursion into Z-land’, in M. O’Rourke
and C. Washington (eds.), Situating Semantics: Essays on the Philosophy of John Perry,
MIT Publications, Cambridge MA, forthcoming.
Corazza, E., Fish, W., and Gorvett, J.: 2002, ‘Who is I?’, Philosophical Studies 107(1),
1–21.
Hintikka, J. and Sandu, G.: 1995, ‘The Fallacies of the New Theory of Reference’, Synthese
104, 245–283.
Hintikka, J.: 1998, ‘Perspectival Identification, Demonstratives and “Small Worlds” ’,
Synthese 114(2), 203–232.
Huang, Y.: 2000, Anaphora: A Cross-Linguistic Study, Oxford University Press, Oxford.
Kaplan, D.: 1977, ‘Demonstratives’, in J. Almog et al. (eds.), Themes from Kaplan, Oxford
University Press, Oxford, 1989, pp. 481–563.
Kaplan, D.: 1989. ‘Afterthoughts’, in, J. Almog et al. (eds.), Themes from Kaplan, Oxford
University Press, Oxford, 1989, pp. 565–614.
Kripke, S.: 1977, ‘Speaker’s Reference and Semantic Reference’, in P. A. French, T. E.
Ueling and H. K. Wettstein (eds.), Contemporary Perspectives in the Philosophy of
Language, University of Minnesota Press, Minneapolis, 1979, pp. 6–27.
Lasnik, H.: 1976, ‘Remarks on Coreference’, Linguistic Analysis 2, 1–22. Reprint in
Lasnik, H.: 1989, Essays on Anaphora, Kluwer, Dordrecht, pp. 90–109.
Lewis, D.: 1979, ‘Scorekeeping in a Language Game’, in D. Lewis (ed.), Philosophical
Papers, Vol. 1, Oxford University Press, Oxford, 1983, pp. 233–249.
Perry, J.: 1977, ‘Frege on Demonstratives’, The Philosophical Review 86(4), 474–97. Reprinted in Perry, J.: 2000, The Problem of the Essential Indexical and Other Essays,
CSLI Publications, Stanford, CA, pp. 1–26.
Perry, J.: 1986, ‘Thoughts without Representation’, Proceeding of the Aristotelian Society
60, 137–152. Reprinted in Perry, J.: 1993, The Problem of the Essential Indexical and
Other Essays, Oxford University Press, Oxford, pp. 205–225.
Predelli, S.: 1998, ‘Utterance, Interpretation, and the Logic of Indexicals, Mind and
Language’, 13(3), 400–414.
Sidelle, A.: 1991. ‘The Answering Machine Paradox’, Canadian Journal of Philosophy
81(4), 525–539.
Smith, Quentin: 1989, ‘The Multiple Use of Indexicals’, Synthese 78, 167–191.
Stalnaker, R.: 1974, ‘Pragmatic Presuppositions’, in M. Munitz and P. Unger (eds.), Semantics and Philosophy, New York University Press, New York. Reprinted in Stalnaker,
R.: 1998, Context and Content, Oxford University Press, Oxford, pp. 47–62.
Department of Philosophy
The University of Nottingham
University Park
Nottingham, NG7 2RD
U.K.
E-mail: eros.corazza@nottingham.ac.uk