Open quotation revisited
A pragmaticist feels the tug of semantics: Recanati’s “Open quotation revisited”
Philippe De Brabanter
Université Libre de Bruxelles
As Recanati explains at the end of chapter 7, “Open quotation revisited” was originally
written with a view to doing mainly to things: (i) respond to papers on open quotation
published after 2001, notably in the collection I edited as volume 17 of the BelgianJournal of Linguistics (De Brabanter 2005a), and (ii) reexamine the more complex
account of context-shifts that he had provided in Oratio Obliqua, Oratio Recta. As
Recanati himself acknowledges, chapter 8 has a more semantic flavour than the nearly
unconditionally pragmatic chapter 7. Towards the end of this paper, I will devote two
sections to explaining why I think concessions to the semanticist are better avoided.
I will begin by reviewing the three main topics of chapter 8, though in a differentorder. First, I discuss Recanati’s treatment of the widely debated question of the
cancellability of the utterance ascription to the reportee in MQ. Second, I devote some
time to an assessment of Recanati’s response to the strong objections that Bart Geurts &
Emar Maier (2005) voiced against ‘two-dimensional’ theories of hybrid quotation
(among which they included Recanati’s). Third, I comment on Recanati’s reappraisal of
his own account of context-shifts.
1 Cancellability and ambiguity
Several authors hold that it is part of the semantic contribution of MQ that the quoted
string is ascribed to the reportee (the referent of the subject of the reporting verb). This
view is expressed notably by Benbaji (2005), Cappelen & Lepore (1997, 2005),
McCullagh (2007). (I shall focus on Cappelen & Lepore, as their analysis has been by
far the most influential.) These authors take this ‘utterance ascription to the reportee’ to
be an entailment of MQ. This may seem surprising, since there are apparently quite
straightforward counterexamples to this claim. It appears easy to cancel the putative
entailment, which therefore turns out to be a pragmatic inference instead. Take the
following pair of examples. In (1), it is very tempting to ascribe the quoted words to
Alice. Yet, the acceptability of (2) suggests that that ascription is cancellable:
(1) Alice said that life
‘is difficult to understand’.
(2) Alice said that life
‘is difficult to understand’, to use Rupert’s favourite phrase.
In (2) the quoted words are ascribed to Rupert, not to the reportee, Alice.
In the face of examples like these, how can one continue to defend the‘entailment’ story? By invoking the critic’s failure to distinguish between cancellation
and recourse to another sense of an ambiguous expression. The argument then is that
At least — and this is what matters — this reading is possible. But perhaps the quoted words can be1
simultaneously ascribed to both Alice and Rupert.
Ambiguity is a vague term, covering both the linguist’s ‘polysemy’ and ‘homonymy’. I shall stick to it,
however, because most discussions of this kind in the philosophical literature are framed in terms ofambiguity.
the metalinguistic comment shows that the words between inverted commas are not
mixed-quoted, but, say, scare-quoted instead.
Generally, ambiguity-based analyses are dispreferred, because they are at oddswith Grice’s ‘Modified Occam’s Razor’, which states that one should avoid multiplying
senses if their postulation does no more descriptive or explanatory work than an
independently justified general pragmatic mechanism. The principle is a difficult one to
apply, though (cf. Sperber & Wilson 2005: 469), and judgments as to which account
does more useful work may be tricky to make. So it is not entirely surprising that the
literature on quotation does feature a few theories that appeal to ambiguity. The
question is this: other than the somewhat arbitrary application of Modified Occam’sRazor, how can one reject an ambiguity-based account? The best way to go, it seems, is
to attempt to show that the meaning distinction posited is ad hoc, or untenable in some
other way. This is exactly what Recanati sets out to do. First, he submits variants of an
example similar to (1), along the lines of:
(3) Alice said that life
‘is difficult to understand’, as she put it.
(4) Alice said that life
‘is difficult to understand’, to use her favourite phrase.
The quoted words are explicitly ascribed to the reportee, Alice, by means of a
metalinguistic comment. So far, these examples are compatible with both Recanati’s
account (explicitation of a pragmatic inference) and Cappelen & Lepore’s (double
ascription of the utterance to Alice). But then, Recanati offers minimal variations on (3)
(5) Alice said that life
‘is difficult to understand’, as Rupert would put it.
(2) [repeated] Alice said that life
‘is difficult to understand’, to use Rupert’s favourite
Now the quoted utterance is ascribed to Rupert. Recanati’s reasoning goes like this:there seems to be every reason to put (3) and (4) on a par with (1): all of them involve
MQ. Furthermore, since (2) and (5) only differ from (3) and (4) by one NP, they should
be judged to exhibit MQ as well. Yet, Cappelen & Lepore would (have to) say that they
involve scare quoting (ScQ) instead, a judgment that now appears ad hoc.
This is not a logically conclusive demonstration, but I take it that it does shift theonus on ambiguity theorists to show that their assumption that different meanings (or
uses) of the quotation marks are involved in (2) and (5) is not arbitrary. Overall, I
believe that the strategy that consists in questioning the boundaries between MQ and
ScQ is worth pursuing. In my opinion, too many authors — this includes Recanati —
To my knowledge, Gómez-Torrente (2005, 2011) is the only one to fully articulate a theory on which
quotation marks have several conventional meanings. However, both Cappelen & Lepore (2005) andBenbaji (2005) appeal to strategies that resemble appeals to ambiguity (although they are at pains to deny
postulations of ambiguity), and Geurts & Maier write that “quotation marks seem to be polysemous rather
than just ambiguous” (2005: 127).
Should you have any trouble with these examples, think of modalised sentences like Alice might say that
4 life ‘is difficult to understand’, to use your favourite expression.
MQ. Though interesting in their own right, notably because of their alleged truth-
conditional effects, the Cappelen & Lepore examples have resulted in theorists
sometimes ‘not seeing the forest for the trees’. From the perspective of an empirically
sound theory of quotation, there is no particular reason to give Cappelen & Lepore’s
MQ precedence over other hybrid cases. I will add that that is so even if one’s main
concern is with truth-conditions. Thus, it is not clear that the quotations in examples like
(6) and (7) below have no impact on truth-conditions. None the less, they are certainly
not amenable to a Cappelen & Lepore-type analysis, because they do not come under
the scope of an obvious reporting verb:
(6) Chateaubriand returned to France in 1800,
‘with the century’. (Recanati 2010: 272)
(7) Mrs. Obama described herself as a “110-percenter,” which is how much she said she
gives of herself to both her family and her job, which means she always feels “
The hybrid quotation in (7) is especially interesting because it includes an indexical, I.
Clearly, it has a truth-conditional effect here (it shifts the context). Left with a choice
between semantic MQ and pragmatic ScQ, Cappelen & Lepore would not be able to
account for this. In the end, I believe there are good reasons to hold that MQ and ScQ
do not exhaust the domain of hybrid open quotations. It is not even clear that MQ and
ScQ can be neatly distinguished in a non ad-hoc manner. I conclude that appealing to
ambiguity in order to claim there is no cancellation of the ‘utterance ascription to thereportee’ in MQ is a strategy that fails.
In their 2005 paper, Bart Geurts and Emar Maier outline a presuppositional account of
hybrid quotation (which they call ‘mixed’). This is deliberately intended as a one-
dimensional theory, because, as they see it, two-dimensional theories face major
difficulties. In their critique, they focus on Potts (2007), the most fully worked out
account from a formal point of view, but they add that “although we will confine our
attention to one particular version [of two-dimensionality], our criticism is directed
against the whole family of 2D theories” (2005: 111), and this includes Recanati (2001)
and Predelli (2003), to which I would add Cappelen & Lepore (1997, 2005) and García-
Consider:(8) When in Santa Cruz, Peter orders
‘[eɪ]pricots’at the local market.
For a more detailed critical assessment of the distinction between MQ and ScQ, see De Brabanter (2010:
115-117). García-Carpintero is among the few authors who have made allowances for “cases in betweenmixed quotations and scare quotes, i.e., cases in which there is in the background a direct-discourse
ascription to a speaker, but the utterance itself is not a saying-ascription” (2011: 129).
Potts’s (2007) theory states that two propositions are expressed, one independent of the
quotation — the ‘use’ line below —, the other reflecting the ‘speech report’, the
‘mention’ line below.
Use:When in Santa Cruz, Peter orders apricots at the local market.
Mention: Peter utters ‘[eɪ]pricots’.
According to Geurts & Maier, a first problem for this account is that it doesn’t capture
the most obvious interpretation of (8), namely that “when Peter is in Santa Cruz andbuys apricots at the local market, he says ‘[eɪ]pricots’. They suggest that the additional
restriction might result from the quotation being in focus: “the restriction observed in
this case coincides with the complete regular meaning (in Potts’s terms) of ‘Peter orders
‘[eɪ]pricots’ at the local market’” (2005: 112), and not as is usual, with “backgrounded
information in [the] scope [of the quantifier]” (2005). They are sceptical that the two-
dimensional analysis can account for the focus effect observed.
What they see as an even more pressing problem arises in connection with anexample like (9), in which the most natural reading has it that each soldier said
‘mommy’ to refer to his own mother, something which is not captured by the mention
(9) Every soldier said he longed to go home to his
Every soldier said he longed to go home to his mommy.
Mention: Every soldier uttered ‘mommy’.
Geurts & Maier point out further problems with metalinguistic negation and concludethat the two-dimensional analysis is ill-equipped to account for cases, like the above,
where there is rich interaction between the two postulated levels of meaning.
In addressing Geurts & Maier’s critique, Recanati acknowledges from the outsetthat the theory set out in chapter 7 is multi-dimensional:
On the version of this view I put forward in chapter 7, any of [several
examples containing a hybrid quotation] compositionally expresses a certain
proposition — the same it would express without the quotation marks —
and use-conditionally expresses a further proposition to the effect that the
speaker is R-ing the enclosed words. In addition the utterance pragmatically
conveys an array of propositions having to do with the speaker’s point in R-
ing the enclosed words. (2010: 275)
In this citation, Recanati uses “R-ing” as a placeholder for a relation to be specified later
in the chapter, so as to temporarily leave open the question of the conventional meaning
of quotation marks. As we shall see in section 3, the choice will be between the broad
I have transformed Potts’s fully explicit formulas into plain English.
I showed in the essay devoted to chapter 7 that Recanati recognises the richinteraction between the meaning of a quotation and truth-conditional content. Unlike
Potts (2007), he has available a couple of tools that are designed to capture the
interactions between levels, namely free pragmatic enrichment and context-shifts. These
two mechanisms are quite capable of explaining indirect effects of pragmatic meaning
upon an utterance’s (intuitive) truth-conditions.
With respect to free enrichment, Recanati basically repeats his analysis of MQ:if the quotational point is to make the addressee understand that the quoted words were
uttered by the agent of the speech event, that point will enrich the truth-conditions of the
utterance. In other words, the contextual meaning of the quotation affects the truth-
conditional content. His conclusion: “One may deny that free enrichment exists, but if it
exists, then it provides an explanation of the interaction of semantic content and
quotational meaning in [MQ] that is fully compatible both with multi-dimensionalism
and with a pragmatic approach to open quotation” (2010: 279).
Now it seems to me that Geurts & Maier’s criticism goes further and concernsaspects of meaning that are perhaps less easy to deal with than basic MQ. In the next
couple of paragraphs, I will therefore propose what I take to be a Recanati-style analysis
of examples (8)-(9), and see how it fares. Let us begin with (8). Geurts & Maier’s idea
was that, as a result of the quotation being in focus, there is an extra domain restriction
affecting its interpretation, and that Potts’s separation between the regular and the
quotational meaning makes it difficult for him to capture this interaction. The difference
between Potts and Recanati is that the latter makes allowances for pragmatic intrusions
into truth-conditions. Elsewhere, he has provided analyses of quantifier domain
restriction in terms of free pragmatic enrichment (2004: 87-88, 2010: section 6.1 of
chapter 3), and I shall do likewise here: probably helped by the intonational focus on[eɪ]pricots, the addressee understands that one major aspect of the meaning of (8) —
closely linked with the quotational point — is the West Coast pronunciation of
7 apricots. This may help him see that (8) does not convey that “all the time that Peter is
in Santa Cruz, he is (constantly) buying apricots at the local market”. The inference
about the quotational point together with our world knowledge (people are not normally
buying fruit all the time) seems enough to trigger the extra contextual restriction which
affects the truth-conditions of (8).
I have little reason to assume that Geurts & Maier would disagree with this sortof account. Note that they themselves offered no detailed explanation of how the extra
domain restriction arose in the first place (merely suggesting the influence of
intonational focus), and they might therefore be open to the free enrichment account. An
extra argument stems from the following observation: the disquoted counterpart of (8)
seems to lend itself to at least two salient readings:
DISQ) When in Santa Cruz, Peter orders apricots at the local market.
The likely quotational point here is “getting the addressee to understand that people on the West Coast
of the USA pronounce apricots /ᴵeɪprɪkɒts/”.
Reading a: “every time Peter is in Santa Cruz, he goes to the local market and orders
Reading b: “every time Peter is in Santa Cruz and goes to the local market, he orders
I take it that the place of the intonational focus will favour one or the other reading. On
reading a, there is no extra domain restriction. But on reading b, there is. Now thisrestriction is different from that in (8), but that is not what matters here. The important
point is that there should be a restriction at all, in the absence of any quoting. The lesson
I wish to draw from this is that the requirement for a satisfactory account of both (8) and
DISQ) is, first and foremost, that it incorporate (something like) free enrichment. So,
where Geurts & Maier thought they had found fault with two-dimensional accounts of
quotation in general, the problem may actually have been with two-dimensional
frameworks that did not allow for free enrichment. I’d go even further than that. The
problem was probably not with two-dimensional accounts per se, but with those
frameworks that include nothing like free enrichment, irrespective of the number of
meaning dimensions they postulate.
Turning now to (9), we see that the extra meaning component posited by Geurts& Maier does not affect the truth-conditions of the utterance, which are neatly captured
by the use line. In Recanati’s parlance, the extra component is entirely a matter of the
pictorial meaning. The quotation marks (in speech, some intonational feature) indicate
that the word mommy is produced for demonstrative purposes. The addressee fleshes
this out by identifying an internal target: the word mommy is used echoically, mimicking
each soldier’s use of the word to refer to his own mother. This is the expected result.
Though impressionistic, the Recanati-like account probably gives a good idea of how
Geurts & Maier themselves hit upon the notion that each token of mommy was not
merely uttered, as the mention line of the Potts-like analysis suggests, but uttered to talk
about a particular person.
I conclude that Recanati’s two-dimensional theory is sufficiently different fromPotts’s to withstand Geurts & Maier’s objections. It is equipped with conceptual tools
that enable it to capture refinements that are inaccessible to a formally more precise but
pragmatically poorer two-dimensional picture like Potts’s.
3 Are context-shifts encoded in the conventional meaning of quotation marks after
Still, Recanati feels, a challenge remains. The fact that the tough examples brought up
by Geurts & Maier can be dealt with by his two-dimensional framework does not prove
that the latter is superior to Geurts & Maier’s one-dimensional semantic theory of
At first blush, unmarked focus on market would favour reading a, while reading b would be facilitated
by marked focus on apricots. The latter might signal a contrast with, say, peaches, rather than with anEast Coast pronunciation /ᴵæprɪkɒts/, as in the quotational example (8).
This is not meant to diminish the merits of Potts (2007), which offers a fully worked out grammar,
something that Recanati does not do. I’ll return to the issue of the differences between semantic andpragmatic accounts in the concluding remarks to this paper.
hybrids. Recanati is willing to face the challenge, and possibly to reconsider his views,
as we shall see.
Recanati remarks that the account of non-cumulative hybrids in chapter 7involves a revision of the classic Kaplanian framework (Kaplan 1989). As stipulated by
Kaplan, semantics assigns a character, namely a function from contexts to contents, to a
sentence. The problem with sentences containing non-cumulative hybrids is that no
single context (the ‘current’ one, or the ‘shifted/source’ one) yields the right content.
Now one may decide to give up the notion that whole sentences have characters, so that“only simple expressions will be assigned characters: for more complex expressions like
sentences, we will directly compose the contents determined by the characters of the
parts in their respective contexts” (2010: 283). But, Recanati points out, we may prefer
to keep the Kaplanian framework unchanged. One way of doing this is to make the
context-shift “internal to the character of the sentence [...] by assigning to the sub-
clausal quotation a metalinguistic character, which maps the context in which the sub-
clausal quotation occurs (viz. the current context) to the content expressed by theenclosed expression when interpreted in the source context” (ibid.). This analysis, put
forward in Oratio Obliqua, Oratio Recta, still leaves open the question whetherquotation marks are interpreted as the syntactic vehicle for the context-shifting operator.
If they are not, the operator is simply a tool of the metalanguage, and the context-shift
remains a fully pragmatic mechanism operating at the pre-semantic level. If they are,
however, then context-shifts are semanticised, in the sense that they are now effected by
a linguistic device, the quotation marks themselves.
This is an overstatement. There are cases in which the shifted context will yield the right content, albeit10
in a manner that cannot satisfy the theorist. Consider a variation on (6):
(6’) ‘Quine’ wants to have breakfast.
The current context would yield the wrong content, since (6’) would be construed as being about the real
Quine. By contrast, the shifted context (James’s) yields the right content, provided the other words in the
sentence receive the same interpretation in ordinary English and in James’s idiolect/from James’s
perspective (this was the reason for removing the indexical us from the example). Still, the latter analysis
is unsatisfactory because it does not differentiate between an utterance of (6’) and one, say, of:
(6’’) ‘Quine wants to have breakfast’,
where the quotation takes scope over the whole sentence.
One more point: Recanati himself explicitly makes allowances for cases like (6’) when he describes some
context-shifts as ‘benign’ (2010: 286; see below).
Clausal/sub-clausal are terms from Potts (2007). I prefer hybrid/non-hybrid, however, because:
(i) there are instances of so-called ‘sub-clausal’ open quotation that are clausal after all, e.g. Her ideathat
“Feminist studies should, by definition, entail respect for the views and intentions of
authors”(238) ought, in fact, to have been extended to her discussion of other plays. (mclc.osu.edu/
(ii) Potts’s clausal applies to open and closed quotations, regardless;
(iii) there are instances of so-called ‘clausal’ open quotation that prove less than clausal, e.g.
fridge?’What on earth was it doing there?.
In this chapter, ‘echoic use’ and ‘context-shifting use’ are used interchangeably.
I have not been able to pin Geurts & Maier down to either position. Nowhere in their (2005) do they13
say explicitly that their meaning-shifts are triggered by the quotation marks. In later writings, though,
Maier writes that English (as opposed to e.g. Ancient Greek) requires quotation marks in order to achieve
meaning-shifts. (Maier 2012). In this case, it is quite clear that he is talking about syntactic quotation
marks. In previous work, he made a distinction between the latter and semantic quotation marks: “There
are well-known constructions, even in English, that are completely unmarked, intonationally and
orthographically, yet contain quotation marks semantically” (2007). I have not been able to determine if
these semantic quotation marks belong to the object-language or to the metalanguage.
Interestingly, whereas Recanati opted for the pragmatic alternative in OratioObliqua, Oratio Recta, he now concedes that the issue is controversial, and declares
himself ready to reassess the conventional meaning of quotation marks in hybrids.
He begins his assessment by showing that the pragmaticist cannot make use ofthe following objection: “though non-cumulative hybrids are (usually) echoic, they
involve no context-shift since they do not affect the truth-conditions of the utterance in
which they occur”. With reason, Recanati dismisses this objection: in modifying the
character of an expression, context-shifts need not affect their content; they may be
‘benign’. Now, if it turned out that all uses of quotation marks in hybrids are echoic, it
would be tempting to say that quotation-marks-as-used-in-hybrids conventionally
encode echoicity, rather than just a demonstrative intention.
In chapter 7, Recanati offered the quotation in (10) as an illustration of flatmention in hybrid quotation:
(10) A ‘fortnight’ is a period of fourteen days.
His revised opinion is that this sort of example alone cannot refute the view that
quotation marks encode echoicity. It does not seem illegitimate, at any rate, to say that
the display of fortnight echoes a sort of generic speaker, something like ‘the competent
English speaker’. Given the current state of our knowledge, Recanati judges himself
unable to determine whether there exist unmistakable examples of non-echoic hybrids.
If further empirical research should find no instances of non-echoic hybrids,then, Recanati volunteers, he would be ready to adopt Geurts & Maier’s
presuppositional analysis, which he deems very similar to the analysis in terms of a
metalinguistic character. That, however, would make context-shifts a semantic
phenomenon: in Geurts & Maier’s analysis, presupposition is a kind of anaphora that
needs to be resolved as part of semantic interpretation. Exit the pre-semantic analysis,
Still, Recanati feels that this concession to semanticists does not mean that a
proper treatment of quotation can do without two dimensions. One still needs to deal
with the quotational point. “If we leave aside what I called ‘the contextual meaning of
the quotation’, we get only a truncated account” (2010: 289).
I definitely agree with Recanati. However, I feel that few philosophers wouldnot. After all, even diehard semanticists-about-quotation like Cappelen & Lepore devote
entire pages to the pragmatics of quotation (2005: 55-57), and it is clear that they defend
a dual approach distinguishing between semantics and ‘speech-act heuristics’. There is
worse, I believe. If Recanati grants that the meaning of quotation marks in open
quotation is echoic while at the same time maintaining, correctly, that there are cases of
closed quotation that are not echoic, he tacitly admits that quotation marks are
4. Why resist the semanticisation of quotation marks?
I identify the multiplication of senses as one of the tools available to the semanticist
when it comes to accommodating phenomena or properties uncovered by the
the possibility of cancelling utterance ascription to the reportee in mixed quotation. In
retrospect, this looks like a Pyrrhic victory. Pending the results of further empirical
enquiry, Recanati is poised to accept that there are two meanings (or uses) of quotation
marks: one simply consists in signalling a demonstration, the other encodes echoicity,
i.e. takes the meaning of the quoted string ‘σ’ to be something like “what echoed
speaker X means by ‘σ’”. Though not negligible, the only difference with the distinct
meaning that Recanati rejected earlier is that the quoted utterance is ascribed to some
agent (to be contextually determined) rather than systematically to the reportee.
The first question to ask is whether there are non-echoic hybrids? It does seemthat examples like (10) — the fortnight type — behave differently from their closed
counterparts. Thus, the addition of a metalinguistic comment such as as X says (in L)
seems more felicitous in (10) than in (11):
(10’) A ‘fortnight’, as one says in English, is a period of fourteen days.(11) ‘Fortnight’ is a noun.
(11’) ? ‘Fortnight’, as one says in English, is a noun.
This possible difference in acceptability between (10’) and (11’) may be grist to the mill
of the theorist who leans towards generalised echoicity in hybrid quotation: in (10’), it is
the ‘generic English speaker’ who is echoed, and this may be what (10) implicitly
conveys too. However, the data are rather complex. Consider:
(12) ‘AWOL’ means “absent from one’s post but without intent to desert”.
(12’) ‘AWOL’, as they say in the army, means “absent from one’s post but without
intent to desert”.
In (12), we have a closed metalinguistic citation. Yet, a metalinguistic comment
suggesting an echoed speaker seems perfectly acceptable here. If anything, this
confirms Recanati’s claim that a thoroughgoing empirical study is needed.
Our current inability to settle the above question doesn’t have to mean that wemust leave the issue aside until further notice. I believe that a case can be made against
‘going the Geurts & Maier way’ even if it should turn out that there are no non-echoichybrid quotations. My main arguments will be that it is better (i) to steer clear of an
ambiguity theory of quotation marks, and (ii) to avoid splitting the theory of quotation
into one theory of written instances and another theory of spoken ones.
I hinted above that ambiguity theories are more typical of semantic thanpragmatic accounts. What all ambiguity accounts agree on is that quotation marks have
distinct conventional meanings. But they may regard the role of quotation marks as
being more or less important to the generation of quotations, i.e. they may be more or
less semantically-orientated. As I see it, the relevant positions are defined by the
answers to the following three questions:
try and devise an integrated theory of quotation that applies across the board to
spoken and written instances?
(b) do spoken quotations also have quotation marks (i.e. some intonational or
paralinguistic counterpart to quotation marks)?
(c) are quotation marks necessary to the generation of quotation, or are they optional?
To begin with question (a), I believe the theory laid out in chapter 7 is anintegrated theory of quotation, emphasising the pictorial dimension of the interpretation
of all quotational utterances. This theory, though illustrated with mainly written
examples is clearly also intended (actually even more strikingly so) to have relevance to
spoken quotations. It is also an account that gives pride of place to the quoter, and to
how the quoter exploits a variety of tools (only a few of them conventional) to mark a
string as quoted and to guide the addressee towards a correct apprehension of the
It seems to me that the ambiguity theorist is at risk of pushing the quoter into thebackground and of giving up on an integrated theory. Whether she does depends on how
she answers questions (b) and (c). Below, I sketch what I take to be the main theoretical
positions defined by answers to (b) and (c):
position 1: There is no quotation without quotation marks. Any putative counterexample
will be explained away as no more than apparent. There are several ways of dealing
with (apparently) unmarked quotations, which I can only hint at here: unmarked
‘quotations’ are simply not quotations (e.g. Cappelen & Lepore 2005); unmarkedquotations trigger blatantly false readings that require pragmatic repair (cf. García-
Carpintero 2004, Gómez-Torrente 2005); ‘unmarked’ quotations are marked in logical
form (and/or in syntax) even when the marks are invisible at the surface of things. The
surprisingly popular view of the indispensability of quotation marks is the most
profoundly semantic one that I can make out. When taken to apply only to written
instances, it has the additional consequence of ruling out any integrated account. When
taken to apply to both speech and writing, it is compatible with an integrated account,
albeit one that plays down the pictorial dimension and the role of the quoter: it is
quotation marks that ‘do the quoting’.
position 2: Quotation marks exist in both writing and speech. If tenable, this view is
consistent with an integrated theory. Moreover, if quotation marks are taken to be
optional, the theory does not have to downplay the role of the quoter. But is it tenable?
So far, no one has been able to show that speech uses anything like the conventional
quotation marks of writing. That doesn’t mean there never are any marks of quoting in
speech, there are, but these are variable and it’s unclear that they are more than
indicators triggering some pragmatic inference as to the occurrence of a quotation.
Admittedly, a lot more empirical work needs to be done into this question, but at thisstage it would be awkward for an ambiguity theorist to adopt position 2.
See De Brabanter, “Quoteless quotations” (in preparation), for details.
In the sparse literature on the subject, the conclusions tend to be negative: “[i]t would be an
overstatement to claim that prosodic marking is used systematically as a sign of reported speech in talkthe way quotation marks are in texts” (Klewitz & Couper-Kuhlen 1999: 473; see also Kasimir 2008)
position 3: quotation marks are optional in writing and have no conventional
counterpart in speech. This is certainly the more pragmatic position, and the one that I
suppose Recanati would opt for if he ended up adopting Geurts & Maier’s view on
hybrid quotation. Note, however, that Geurts & Maier seem to regard quotation marks
as necessary in hybrid instances. Indeed, a distinction could be made between optional
marking in closed and autonomous open cases, and compulsory marking in hybrid ones.
I do not think that Geurts & Maier are right on this point. It turns out that hybrid
quotations may not be marked by quotation marks (or any other device) at all (cf. De
Brabanter 2010: 113-115). This often happens in allusions, notably allusions to well-
known sayings or famous literary works. In those cases, the utterer may choose to
somehow help her addressee (by using quotation marks, or intonational marking in
speech, cf. (13)) or to make him go to the extra effort (and reward) of detecting the
presence of the allusion himself ((14, 15)):
(13) ... consider this, just how many of our immigrants are
yearning to breathe free’? And how many are coming for just the economic
(14) BMP files weren’t of good quality, and, since
beauty is in the eye of the
beholder, I’ve pulled out some of the screens that I like. (BNC, HAC 4519)
(15) So ended the attempts of these
poor, yearning, tired huddled massesto gain
asylum in the US. (New Statesman, 17/01/2000: 16)
In (13), the writer quotes a line from the poem by Emma Lazarus that is inscribed on the
base of the Statue of Liberty, the symbol for the United States’ hospitality towards
immigrants. In (14), even in the absence of any signalling, beauty is in the eye of thebeholder is sufficiently well-known to be widely identified as an echoic hybrid
quotation. Whether the citation will be rightly attributed to its originator (in this case,
conventional received wisdom) is immaterial. In (15), the sequence these poor, yearning, tired huddled masses is used to refer to a group of Haitians who had tried to
enter U.S. territory clandestinely. It evidently conjures up Lazarus’s poem, with
alterations. The original wording of the relevant passage is:
(16) Give me your tired, your poor / Your huddled masses yearning to breathe free
These alterations do not, I contend, prevent these well-known lines of the poem from
Other unmarked cases of (what are perhaps) hybrid quotations were offered longago by Roman Jakobson (1985) in his sketch of the ‘metalingual function’ of language.
Jakobson claimed that definitional examples like (17) (and (18)) “impart information
about the meaning assigned to the word hermaphrodite [...] but [...] say nothing about
the ontological status of the individuals named” (1985: 119):
At least Maier (2012: 133) does, about English (though not Ancient Greek). Cf. footnote 13.
J. Rey-Debove called such instances ‘crypto-citations’ (1978: 261).17
(17) Hermaphrodites are individuals combining the sex organs of both male and female.
(18) A sophomore is a second-year student.
These examples are exactly like (10), were it not for the missing quotation marks, which
are easy to supply:
Q ) ‘Hermaphrodites’ are individuals combining the sex organs of both male andfemale.
) A ‘sophomore’ is a second-year student.
This suggests that (17)-(18) can possibly be regarded as unmarked hybrids.
The lesson to be drawn from these examples is this: in hybrid instances,quotation marks are not systematically necessary to make the addressee understand that
he is to ascribe a string of words to some echoed speaker. This highlights the fact that
the quoter can, given the right circumstances, rely solely on contextual clues, with no
need to signal the allusion with a dedicated linguistic marker.
None of the above conclusively shows that it would be misguided to adopt aone-dimensional semantics for quotation marks in hybrid quotations. Yet, it does make
that option much less attractive. For one thing, if, as I’ve claimed is desirable (and in the
spirit of TCP), the theorist seeks to devise an account of quotation that covers writing
and speech, then she will have to provide a pragmatic explanation of how quotation
works in the absence of any conventional marking. And this she will have to do not just
for non-hybrid but also for hybrid instances. This theory exists: it is the radical
pragmatic theory that Recanati defends in chapter 7.
This, to me, means that it should be much less tempting to endorse a polysemicaccount, even if only for written quotations. A pragmatic account is necessary for
written cases too, since we have seen that unmarked written instances exist both in the
non-hybrid and the hybrid variety. One consequence is that quotation marks cannot be
said to be a necessary ingredient of echoic quotation. If quotation marks are, as Geurts
& Maier propose, context-shifting operators, then they are sufficiently powerful to ‘do
the quoting’ by themselves. The role of the quoter is reduced to selecting the requisite
means to achieve her communicative purpose. But examples like (14) and (15) show us
that it’s the quoter who does the quoting (and the additional echoing in the relevant
cases), and that she does not need quotation marks to that end. Since we have a theory
that can deal with all varieties of quotation, there’s no need to take on the extra baggage
of a partial theory (one of marked written hybrids) that requires (what now turns out to
be an unnecessary) multiplication of senses. Here, the application of Modified Occam’s
Razor seems to be fully warranted.
In “Open quotation”, Recanati seemed to seize every opportunity to show whyone should embrace truth-conditional pragmatics. Most of the effects on truth-
conditions (except those resulting from recruitment as a singular term) were
convincingly explained in terms of independently justified pragmatic mechanisms. In
this respect, “Open quotation revisited” takes a step backward. Most earlier critics of
Recanati’s theory of quotation found fault with him for being too pragmatic about
too keenly to the tug of semantics.
5. Conclusion and prospects
I take François Recanati’s pragmatic theory of quotation to among the best. Though I
have not refrained from criticising what I took to be less compelling aspects, only in a
few cases have I made suggestions as to how to remedy the purported flaws. The critic’s
‘destructive’ task is always easier than the author’s painstaking construction.
I suppose I should stress that Recanati somehow makes his critic’s life easy,precisely because he is at pains to provide as detailed and complete a depiction as
possible of the many mechanisms involved in the interpretation of quotation. It is this
determination to leave no stone unturned that inevitably provides his critic with
opportunities for disagreement.
A typical criticism against Recanati is voiced by Gómez-Torrente: “Withinspiration from Herbert Clark, François Recanati has claimed that quotation marks
quite generally “conventionally indicate the fact that the speaker is demonstrating the
enclosed words” (2001: 680) [...]. It’s hard to make precise sense of this vague claim so
that it can seem true of all uses of quotation but not of any other use of
expressions” (2005: 131-132; my italics). I understand Gómez-Torrente’s concern:
pragmatic theories of quotation seem to inherently exhibit some degree of vagueness
(and intricacy). Recanati’s is no exception. This makes understandable the semanticist’s
eagerness to offer more definite characterisations of the meaning of quotation, via a
specification of the meaning(s) of quotation marks, conceived as necessary to quotation.
This way the semanticist gets a good grip on the phenomenon and eschews vagueness.
There is, however, a price to pay: semantic theories tend to provide18
descriptions that are much more limited in scope. Thus, some semanticists dismiss one
or other quotational phenomenon from the domain of quotation. Cappelen & Lepore’s
treatment of scare quoting is a case in point. But there were precedents. Peter Geach
judged that the mention of sheer nonsense (i.e. strings that do not correspond to actually
existing elements of any language) could not qualify as quotation, and therefore
excluded such mentions from the data that the theory was accountable for (1957: 85).
Nowadays, although most theorists grant that ‘just about anything’ can be quoted, be it
linguistic or non-linguistic material, few — with the exception of a couple of
pragmaticists-about-quotation like Recanati or, especially, Clark — attempt to account
for those ‘quotations’ where they are really challenging, i.e. in open cases, outside of
inertia-inducing linguistic recruitment. Now, it is possible that further empirical
research will show that certain phenomena which some now wish to include within the
ambit of quotation do not belong there. My worry is that the semanticist, because of the
tools at her disposal, will occasionally rule these out a priori.
Pragmaticists-about-quotation may seem to keep away from rigorous definitions
and the kind of formalisations that enable clear predictions. They may also seem to
As we saw in section 4, one can be more or less semantic-about-quotation. In the remaining paragraphs,
the term ‘semantic theories’ will designate the radical accounts that regard quotation marks as necessaryto quotation, and a theory of quotation marks as the right kind of theory of quotation.
engage in vaguer, less easily evaluable, language than semanticists. However, their best
representatives — and Recanati, together with Clark & Gerrig (1990), certainly ranks
amongst the finest — truly engage with the complexity and variety of quoting in a way
that the semanticist (so far, at least) seems to me incapable of doing. Only pragmaticists,
so far, have carefully attended to the fundamental pictoriality of quotation. Naturally,
their attempts at describing, not to mention explaining, what goes on in the various
kinds of quoting do not yield the sorts of strong predictions that semanticists are looking
for. But it is in the nature of pictorial meaning to be less definite than linguistic meaning
(especially linguistic meaning as approached by the formal semanticist). Therefore, the
relative vagueness of the analyses and predictions matches the actual vagueness of the
interpretation of pictorial signalling. What the pragmaticists lose in precision, they gain
There is a major virtue to the semantic theories: they force the pragmaticist tobecome more and more precise. While the pragmaticist forces the semanticist to
develop frameworks capable of accounting for an ever increasing variety of data. It is a
fact that recent developments in formal semantics have shown an ability to deal with a
very broad range of data. Emar Maier’s work stands out in this respect. So, we might
optimistically conclude: “let the interaction between semantics and pragmatics continue
this way, and the theories will keep improving”. I still have a worry, though. That worry
is very similar to the one Recanati voiced in the introduction to “Open quotation”: doing
what the semanticist does, which in essence means regimenting quotation, involves the
risk of turning the researcher’s attention away from the essential feature of quotation, its
pictoriality: at bottom, quotation is a non-linguistic communicative act. That is
something, I believe, that must escape the semanticist, simply because semantics is not
designed to deal with ‘non-symbolic’ communicative behaviours, notably ‘iconic’ ones.
What it can do is clarify certain important aspects of quotation, those in which linguistic
elements do play a significant role. Semantics alone, however, can never offer a viable
alternative to pragmatics in terms of empirical coverage. That is why pragmaticists-
about-quotation must be wary of ‘going semantic’. The point is not simply a matter of
defending one’s own turf, an understandable but ultimately irrational goal, but to
continue working towards the only sort of theory that can hope to describe and explain
quotation, because it starts from the recognition of its essential pictorial quality.
Benbaji, Yitzhak. 2005. Who needs semantics of quotation marks?. Belgian Journal of Linguistics 17.27-49.
Cappelen, Herman, and Ernie Lepore. 1997. Varieties of Quotation. Mind 106.429-50.
Cappelen, Herman, and Ernie Lepore. 2005 Varieties of quotation revisited. Belgian Journal of Linguistics 17.51-75.
Clark, Herb H., and Richard J. Gerrig. 1990. Quotations as demonstrations. Language
De Brabanter, Philippe (ed). 2005a. Hybrid Quotation. Amsterdam: John Benjamins.
(Belgian Journal of Linguistics 17)
De Brabanter, Philippe. In preparation. Quoteless quotations.
García-Carpintero, M. 2004. The Deferred Ostension theory of quotation. Noûs 38.674–
García-Carpintero, Manuel. 2005. Double-duty quotation: the deferred ostension
account. Belgian Journal of Linguistics 17.89-108.
García-Carpintero, Manuel. 2011. Double-duty quotation, conventional implicatures
and What Is Said. In Elke
Brendel, Jörg Meibauer, Markus Steinbach
Understanding Quotation. Berlin/New York: de Gruyter, 107-138.
Geach, Peter T. 1957. Mental Acts. Their Content and Their Objects. London:
Routledge & Kegan Paul.
Geurts, Bart, and Emar Maier. 2005. Quotation in context. Belgian Journal ofLinguistics 17.109-28.
Gómez-Torrente, Mario. 2005. Remarks on impure quotation. Belgian Journal ofLinguistics 17.129-51.
Gómez-Torrente, Mario. 2011. What quotations refer to. In Elke
Meibauer, Markus Steinbach
(eds), Understanding Quotation. Berlin/New York: de
Kaplan, David. 1989. Demonstratives. In Joseph Almog, John Perry and Howard
Wettstein (eds), Themes from Kaplan. New York/Oxford: Oxford University Press,
Kasimir, Elke. 2008. Prosodic correlates of subclausal quotation marks. ZAS 49.67-78.
Klewitz, Gabriele & Couper-Kuhlen, Elizabeth. 1999. Quote-unquote. The role of
prosody in the contextualization of reported speech sequences. Pragmatics
McCullagh, Mark. 2007. Understanding mixed quotation. Mind 116.927–46.
Maier, Emar. 2007. Quotation marks as monsters, or the other way around? In M. Aloni,
P. Dekker, F. Roelofsen (eds), Proceedings of the Sixteenth Amsterdam Colloquium,
Amsterdam, ILLC, 145-150.
Maier, Emar. 2012. Switches between direct and indirect speech in Ancient Greek.Journal of Greek Linguistics 12.118-139.
Potts, Chris. 2007. The dimensions of quotation. In Chris Barker and Pauline Jacobson
(eds), Direct compositionality. Oxford: Oxford University Press, 405-31.
Predelli, Stefano. 2003. Scare quotes and their relation to other semantic issues.Linguistics and Philosophy 26.1-28.
Recanati, François. 2000. Oratio Obliqua, Oratio Recta: an essay onmetarepresentation. Cambridge, Mass.: MIT Press, Bradford Books.
Recanati, François. 2001. Open quotation. Mind 110.637-87.
Recanati, François. 2004. Literal Meaning. Cambridge: Cambridge University Press.
Recanati, François. 2008. Open quotation revisited. Philosophical Perspectives
Recanati, François. 2010. Truth-Conditional Pragmatics. Oxford: Clarendon Press.
Rey-Debove, Josette (1978) Le Métalangage. Etude linguistique du discours sur le langage. Paris: Le Robert.
Sperber, Dan & Wilson, Deirdre. 2005. Pragmatics. In Frank Jackson and Michael
Smith (eds), Oxford Handbook of Contemporary Philosophy. Oxford: Oxford
University Press, 468-501.
Do'stlaringiz bilan baham:
©2018 Учебные документы
Рады что Вы стали частью нашего образовательного сообщества.