Skip to navigation – Site map

HomeAnglophonia28The Interpretation of Indexicals ...

The Interpretation of Indexicals in Hybrid Quotation: A Pragmatic Account

Philippe De Brabanter


In this paper, I offer an analysis of a type of hybrid (including ‘mixed’, or ‘double-duty’) quotation that contains expressions indexed to a deictic centre distinct from the speaker of the host sentence, as in Trump insisted there was “zero chance I’ll quit”. Most writers on quotation offer a semantic account of hybrid quotation, in terms of the conventional meaning of some linguistic unit that occurs as part of the construction, usually quotation marks. Against these, and with Recanati (2000, 2001), I provide an analysis in terms of context-shifts. Understood in a broad sense, context can shift along several parameters: language, situation of utterance, circumstance of evaluation. I show that what is required in the case at hand is a situation-shift, and suggest how that shift can be articulated with other shifts. Relying on Clark & Gerrig (1990) and Clark (1996), I also show that shifts are selective and can therefore be partial. This further allows me to account for those less common hybrid quotations in which indexicals are indexed to the reporting context. All in all, my account points towards the viability of a pragmatic theory that views quotation as being a non-conventional, depictive act of communication, typically different from ordinary linguistic acts, the nature of which is essentially conventional.

Top of page

Author's notes

I would like to thank Laura Devlesschouwer, two anonymous reviewers, and the audience at the workshop on Sens conventionnel et implicite, held in Perpignan on May 25, 2018, for useful questions and comments. I also gratefully acknowledge the financial support of the F.R.S.-FNRS research project T.0184.16, 2016-2020.

Full text


1 Quotation is a phenomenon that has long attracted the attention of those linguists, chiefly (formal) semanticists and pragmaticists, who draw most of their questions and concepts from the analytic philosophy of language. As a rule, semanticists will tend to explain meanings, explicit or implicit, in terms of constraints or conditions that are carried by words or morphemes, working bottom-up, as it were. Pragmaticists, in contrast, will tend to account for implicit meanings (and some explicit meanings as well) in terms of general communicative or conversational principles, working top-down instead. Disagreements, however, are only partial. Nowadays, even diehard semanticists concede that the utterance of a sentence communicates not only semantic meanings, i.e. meanings that are carried by the linguistic structure itself and by the lexical items that constitute it, but also meanings that are pragmatic in nature. Conversational implicatures are a case in point. The majority of semanticists, however, still hold that any aspect of truth-conditional content is semantic in nature, i.e. can be traced back to the conventional meaning of some lexical or grammatical unit.

2 In this paper, I am concerned with a particular way in which what I call hybrid quotations — including ‘mixed’ (Cappelen & Lepore 1997) and ‘double-duty’ and quotations (García-Carpintero 2005) — impact the truth-conditions of the sentences in which they occur. I show that this truth-conditional effect can be accounted for via pragmatic mechanisms. I also outline why a pragmatic account is attractive.

3 My focus is on examples of hybrid quotation in which the quoted segment contains an indexical expression, as in (1).

  • 1 I use boldface to facilitate identification of the main string under discussion.

(1) Trump insisted there was “zero chance I’ll quit”,1 telling The Wall Street Journal: “I never, ever, give up.” (The Sunday Times online, 2016/10/09)

4The I that occurs between quotation marks refers not to the reporter but to Trump. And that is the case despite the quotation not being an instance of Direct Discourse (DD), as is made clear in Section 3.1. This means that the quotation affects the truth-conditions of (1): without the quotation, the I would refer to the reporting journalist and the ’ll to the future of the reporting context, rather than to Trump and to the future with respect to his utterance.

5 As suggested above, semanticists will seek to account for this by appealing to the conventional meaning of some element of linguistic structure, namely quotation marks (e.g. Cappelen & Lepore 1997; Benbaji 2005; Geurts & Maier 2005; Gómez-Torrente 2005, Maier 2014). By way of illustration, Benbaji (2005), Geurts & Maier (2005) and Maier (2014) would render the meaning of “zero chance I’ll quit” as something along the lines of “what Trump means by zero chance I’ll quit”. This is assumed to capture the impact of the quotation marks on the semantic value of zero chance I’ll quit without quotation marks. What Trump means by zero chance I’ll quit is that there’s no chance (at a time t-1 before the utterance time) that Trump will quit. When that meaning is plugged into the truth-conditions of (1), we get the following, appropriate, result:

Informal representation of the truth-conditions of (1): “Trumpi insisted that there was no chance that hei would quit”

6 As we see, those semantic accounts have no trouble accounting for the meaning of examples like (1). In a sense, this is not surprising at all. Consider what difference the removal of the quotation marks appears to make:

(2) Trump insisted there was zero chance I’ll quit.
Informal representation of the truth-conditions of (2): “Trump insisted that there was no chance that the reporting speaker (≠ Trump) would quit”

7It seems straightforward to ascribe the truth-conditional difference between (1) and (2) to the presence/absence of quotation marks. From this, it is an easy step to the conclusion that theories like Benbaji’s and Geurts & Maier’s are vindicated.

8 I will argue, however, that this temptation should be resisted. The full reasons for doing so cannot be given here, because a full-fledged critique of semantic theories of hybrid quotation (and of quotation, more generally) is too complex a business to be promptly dispatched in a single paper (see Recanati 2001; Reimer 2005; De Brabanter 2013, 2017; for arguments against semantic accounts). More modestly, what I will do here is sketch a competing pragmatic approach which, initial appearances notwithstanding, can explain the truth-conditional effects highlighted above.

A pragmatic theory of quotation

  • 2 See also Ducrot (1984: 199) for an early statement of this type of theory.

9 The rationale behind the pragmatic theory of quotation endorsed in this paper is that a quotation is at its core an iconic act of communication, i.e. an act that is quite different from typical linguistic acts, which rely on conventional form-meaning pairings. Clark & Gerrig (1990) and Clark (1996: chapter 8)2 differentiate between three modes of communication: demonstrating, describing-as, and indicating. Roughly speaking, each mode relies on the use of one of the three different types of signs distinguished by C. S. Peirce (1984: 56) (according to one of the trichotomies he made): icon (for demonstrations), symbol (for descriptions), index (for indications). This trichotomy is based on the relation between the sign and the ‘object’ it denotes. Icons resemble their objects by exhibiting one or more property that the object possesses; indices have an existential or physical connection to their objects; symbols are linked to them via a rule, or convention (see Atkin 2010).

  • 3 For clarity, demonstratives are associated with ‘indications’ (acts using indices). This means t (...)

10 Language use as commonly understood is a matter of conventional signs and falls squarely under describing-as. Language scholars widely grant some role to indications: they are often needed to support the use of certain categories of words, notably demonstratives.3 Iconic signs, by contrast, are usually accorded only a peripheral role, with onomatopoeic words – which are composite symbolic-iconic signs — receiving some attention. As for iconic acts, i.e. demonstrations, up until recently they were not normally thought to be relevant to the study of language.

11 According to Clark & Gerrig (1990) and Clark (1996), the two central properties of demonstrations are their being nonserious and selective. Since quotations are argued to form a special type of demonstrations, they too must be nonserious and selective. Consider Brenda’s uttering (3) aloud:

(3) And then Greta Garbo said, ‘I just want to be alone!’. (cf. Clark 1996: 175; Recanati 2001: 642)

12Unlike Garbo’s character’s utterance of I just want to be alone!, the token Brenda produces will not be understood as a request of hers to be left alone. That is the sense in which it is nonserious. Besides, this token will exhibit a lot of properties, only some of which are ‘depictive’, i.e. are intended by Brenda to be recognised as being properties of the utterance originally produced by Garbo in Grand Hotel. For example, Brenda’s weary voice, her painful facial expression, the tilt of her head, her use of English and of the very words uttered by Garbo will be depictive. By contrast, her open eyes, her Yorkshire accent, her sitting on a chair will be non-depictive aspects of the whole demonstration. That is the sense in which the demonstration is selective.

13 In producing the utterance-token in (3) the utterer seeks to enable her addressee to experience what it is like to hear and see Garbo utter a certain sentence (Clark & Gerrig 1990: 765). Addressees or other witnesses of a quotation as in (3) may sometimes wish to reject it as inappropriate and perhaps even untrue based not only on the choice of the words used by the quoter but also on other aspects, say voice quality or intonation. In this respect, the purely linguistic aspects of the demonstration, in particular the quoter’s choice of words, are not more important than the non-linguistic ones in making the quotation felicitous. To press that point further, consider (4).

(4) After a moment a man’s voice said in Italian: ‘Is it you?’ (British National Corpus)

14This example shows that even the language of the quotation need not be depictive. The adjunct in Italian makes it clear that Italian words were initially uttered. But there is no problem with using English words to stand for them. Since (4) is taken from a novel, it also illustrates how written occurrences of hybrid quotation can be selective.

15 Semantic theories of quotation trace the meanings that quotations convey back to the conventional meaning of some element of structure (chiefly quotation marks), thereby making these elements of structure necessary to the generation of quotations — quotations do not exist in the absence of these linguistic triggers. In other words, quotation is viewed as just another convention-based linguistic phenomenon, and semantic theories must therefore deny that the key mechanism involved in the generation of quotations is iconic. This to me is an inexcusable fault.

16 One might argue that pragmatic theories too fail in a major respect: they do not at first sight endow quotation marks with any role at all. Appearances are deceptive, though. Pragmaticists do not deny that quotation marks are important and have a semantics: quotation marks may be useful indicators of the occurrence of a quotation, thus helping readers to process it. But quotation marks are not always present — think of the gazillions of quotations produced orally, the marking of which is varied and unsystematic (Klewitz & Couper-Kuhlen 1999: 473; Kasimir 2008). It would be unwise to make entire theories of quotation dependent on their presence, unless one thinks that spoken quotations and written quotations are distinct phenomena requiring different explanations. Pragmaticists generally agree that one and the same theory must apply across mediums, consistently with their assumption that a quotation is a communicative act, not a phenomenon whose existence is conditional on linguistic markers that may only exist in one medium. The relative importance of quotation marks will depend on contextual factors. There will be genres (e.g. academic or news writing) in which quotation marks are expected to occur with every quotation. In other genres, however, as in informal writing or in conversation, markers of quotation will be used much less systematically.

17 Not anymore than I was able to offer a full-blown critique of semantic theories can I provide a more extensive justification of the pragmatic approaches, in particular of the demonstration theory. (For more arguments, see Clark & Gerrig 1990; Clark 1996, 2016; De Brabanter 2017; Recanati 2001.) I think, however, that the outline just provided shows that the demonstration theory is a serious contender in the debates about quotation. To further test its viability, the present paper investigates if it can (i) coherently deal with challenging examples like (1), and (ii) do so in a way that suitably accords with the general picture of quotation as a demonstration. But before this can be done, I need to make the very concept of a hybrid quotation more tangible.

Hybrid quotation

  • 4 My use of the label ‘hybrid quotation’ draws inspiration from Recanati (2001), although Recanati (...)

18 Hybrid quotation is especially common in certain written genres in English, news writing paramount among them.4 The examples below illustrate what are often regarded as two separate varieties of hybrid quotation, ‘mixed quotation’ (e.g. Cappelen & Lepore 1997) and ‘scare quoting’ (e.g. Predelli 2003). I have argued elsewhere that the distinction between mixed and scare quoting rests on shaky foundations (De Brabanter 2017: 238) and will therefore exclusively use the label ‘hybrid quotation’. Note, however, that many examples I’ll be concerned with in the present paper show similarities to Cappelen & Lepore’s mixed quotation.

(5)Gerald said that he would “consider running for the Presidency”. (Searle 1983)
(6) Sometimes [the director] decides he needs to change a shot, or insert a new one, but it’s striking how seldom he has to do this. He’s already
“seen” the entire show in his head, shot by shot. (Lodge 1996: 76)

19 Examples (1) and (5) exemplify mixed quotation as originally defined by Cappelen & Lepore, i.e. a hybrid quotation under the scope of a reporting verb (told, said), while (6) illustrates scare quoting, by and large the use of a quotation as a way of providing a metalinguistic comment, e.g. that “the enclosed expression is used non-standardly, or is unsuitable for the tone of the discourse; [or, very differently] that the enclosed material is borrowed from an allegedly identifiable source” (Predelli 2003: 4).

Why ‘hybrid’?

20 It is revealing to contrast hybrid quotation with more widely studied varieties of quotation, such as Direct Discourse (DD) and Metalinguistic Citation (MC). Prototypically, these quotations function as NPs in clause structure — they are replaceable by, e.g. that — as illustrated in (7) and (8).

(7) DD: And then Kim said, “Quite red, that one!”.
(8) MC:
‘In a minute’ is not an adverb.

21 Importantly, the internal syntactic structure of the quotations is segregated from their surroundings: never mind that the quoted string in (7) is a sequence of an AdjP and an NP. The complement of said is not generated via a phrase-structure rule like VP V AdjP NP. Semantically, the quotation refers to words uttered by Kim, not to some object that is quite red. Likewise with (8): syntactically, it is not generated by a rule that states that PPs can be subjects; semantically, it is about a certain grammatical structure, not a particular time interval. Note further that strings instantiating just any syntactic structure (or no structure at all, e.g. She the with me six) can be placed between quotation marks in DD and MC, without altering the grammaticality of the host sentence, witness (9) and (10):

(9) DD: And then Kim said, “She the with me six”. At least that’s what I thought I heard.
(10) MC:
‘She the with me six’ is not a well-formed English sentence.

  • 5 In the sense of Huddleston & Pullum (2002: 329). This is similar to an N-bar in generative synta (...)

22 Just the opposite can be observed with hybrid quotations, as they involve strings of words whose internal structure is part and parcel of the host structure. Consider example (11), in which the nominal5 billion Muslims who reject their ideology is quoted at the same time as it functions as the head of the NP the billion Muslims who reject their ideology: you could not substitute an NP or AdvP or a VP or a finite clause for it without making the host structure ungrammatical.

(11) The president differentiated militant groups from the “billion Muslims who reject their ideology”. (The Guardian online, 19/02/2015)

23Likewise with (5) above, in which the clause consider running for the Presidency is at once quoted and the complement of the auxiliary would: again, no other structure than a bare infinitival clause fits into this slot.

24 A significant assumption that I make, contra semantic accounts and some pragmatic ones as well, is that the label hybrid quotation does not identify a separate ‘variety of quotation’ requiring its own separate explanation. Instead, I take it that hybrid quotation is just a quotation of a string of words already in ordinary use. Consider (11) again. One can say there are two parts to this utterance. The first, which ends with the second the, is descriptive, in Clark’s sense. The second, starting with the left quotation mark, is bi-modal, combining description with demonstration. There are also two communicative acts performed. These acts, however, do not map onto the parts just mentioned. The first act, a descriptive one, extends over the whole duration of the utterance. The second, a demonstration, concerns the part between quotation marks. In other words, there is no bi-modal act. A hybrid is not, to use Clark’s (1996) term, a composite signal (an act combining description and demonstration). Instead, it is a composite part of a signal. Hence, on the view defended here, hybrid quotation is not a sui generis variety of quotation with its own syntax and semantics. This places a significant constraint on the account I will propose: it must use the same concepts and mechanisms that are used to explain just any quotation.

Two main approaches to hybrid quotation

25 Unsurprisingly, there are two main families of approaches to hybrid quotation, semantic and pragmatic. When it comes to hybrid quotations containing indexicals, we have seen above how some semantic accounts accurately render their meaning, and especially the truth-conditional impact of the hybrid quotation (see Section 1). These accounts also deal successfully with hybrid quotations without indexicals, like (5).

26 Pragmatic theories, which are not typically concerned with truth-conditions, are faced with a major challenge. Some pragmaticists have dismissed any truth-conditional effect as merely apparent (e.g. Stainton 1999), suggesting instead that hybrid quotation adds felicity conditions to an utterance, not truth-conditions. On this view, the whole issue of the truth-conditional contribution of hybrid quotations dissolves. Other pragmaticists have granted those truth-conditional effects (Recanati 2001; Reimer 2005), and have taken up the serious challenge of accounting for them.

27 For a variety of reasons (see notably Reimer 2005: 171ff), I side with the pragmaticists that acknowledge the truth-conditional effects. Note that with respect to examples like (1), those are simply not deniable, making Stainton’s position untenable. In Section 5, I’ll introduce the mechanism needed to meet the above challenge. Before that, in the next section, I take a closer look at the data that this paper is chiefly concerned with.

Indexicals in hybrid quotations: empirical data

  • 6 This paragraph is a response to an anonymous reviewer’s criticism that my analyses are inadequat (...)

28 I have been gathering the examples that make up my informal corpus6 for several years, in a way similar to the way old-school grammarians used to collect examples on cards. The corpus currently contains 90 relevant examples in five different languages, 72 from English sources, and a few from Dutch, French, German and Italian ones. My goal in collecting this corpus is to confirm the existence of the phenomenon examined in the present paper, and to establish that it is not vanishingly rare, thus warranting its being taken seriously.

29 My informal observations suggest that hybrid quotation containing indexicals is more frequent in English than in close languages like the four mentioned above. The phenomenon mainly comes up in news writing (1) but is also occasionally encountered in other genres, such as autobiographical accounts, for instance (12):

(1) Trump insisted there was “zero chance I’ll quit”, telling The Wall Street Journal: “I never, ever, give up.” [repeated]

(12) By the autumn of 1937 the ‘Socialist’ Negrin was declaring in public speeches that we respect private property,’ […]. (Orwell 1937: 56)

30What is striking about the indexicals in (1) and (12) is that they seem to occur as part of Indirect Discourse (ID) — as evinced by the fact that they are part of a clause that is licensed by a reporting verb that takes a complement in the form of a declarative content clause — but also as part of DD, as they are enclosed in quotation marks. One might hesitate between two ways of characterising these indexicals, in terms of deixis or anaphora. If the examples were simple cases of ID, they would not contain the 1st-person pronouns I and we, and we’d have no hesitation talking about anaphora. If instead of (1) we had Trump insisted there was no chance he would quit, the he would be anaphoric to Trump. If instead of (12) we had the ‘Socialist’ Negrin was declaring in public speeches that he and his comrades respected private property, he and his would be anaphoric to the ‘Socialist’ Negrin. One might therefore be tempted to say that I and we in (1) and (12) are similarly anaphorically linked to the same antecedents, even though they are unusual in (i) exhibiting a change in person, and (ii) being 1st-person pronouns. But, precisely, it seems convoluted to insist that the indexicals in (1) and (12) are anaphors, when treating them as deictics is both possible and more straightforward. We’ll soon see how the context-shift mechanism to be introduced in the next section makes good sense of these indexicals being deictics.

31 Hybrids as in (1) and (12) also sometimes crop up in speech. Here is an example from the BBC panel show Have I Got News For You (2017/12/08, 18’05”, Paul Merton speaking):

(13) Yeah, this could be kinda ironic that this man who tweets his innermost thoughts may have accidentally shot himself in the foot because he said that er “the reason why I had to sack Flynn was because he lied to the FBI”. (Transcription and punctuation mine)

32Examples like (1), (12), (13) have received some attention (Recanati 2001, 679-80; Cumming 2005; Cappelen & Lepore 2005: 61-65; De Brabanter 2017: 238-39, McCullagh 2017). They have also been noticed in studies on other languages, notably French (e.g. Komur 2004; Biardzka 2010). I provide examples for French and German, in (14) and (15) respectively:

(14) Au passage, en prenant […] le contre-pied de son colistier Mike Pence, qui avait précédemment affirmé que « nous respecterons absolument le résultat des élections ». (Le Soir, 2016/10/02: 14)
“… who had asserted before that ‘we will absolutely respect the outcome of the election’”
(15) „Komplott gegen den Brexit, titelde die Daily Mail und zitierte aus einem „Geheimpapier“. Dem zufolge haben die Brüssel-Freunde „für ihre Sabotage“ zehn Millionen Euro zusammengetragen. Das Geld kommt vor allem von „ausländischen Milliardär“ George Soros und dessen „Elite-Freunden, diesen gewissenlosen Geldsäcken, gegen die sich unsere Demokratie nun zur Wehr setzen muss“. (Die Welt, 2018/06/02: 7)
“… George Soros and his ‘elite friends, these unscrupulous moneybags against whom our democracy must now defend itself’”

33 The only studies that discuss those kinds of examples at length are Recanati (2000, 2001) and McCullagh (2017). But some semanticists-about-quotation have used them to argue against pragmatic accounts. Thus, Cappelen & Lepore (2005: 63f) take them to demonstrate that Recanati’s pragmatic account of hybrid quotation is a non-starter. Consider (1) again. Remove the quote marks and you get (2), i.e. what Cappelen & Lepore (2005: 62) call a ‘mistaken rendering’, i.e. a sentence whose truth-conditions are quite different. In Cappelen & Lepore’s view, this means that the quotation marks have a semantic import (because truth-conditional effects must be semantic). Hence, pragmatic theories must be false.

34 I have argued elsewhere that Cappelen & Lepore’s argument is flawed (De Brabanter 2010: 116) and actually undermines their theory. In the next three sections, I further show that an account in the spirit of Recanati (2000, 2001) is viable.

Demonstrations cause context-shifts

  • 7 Henceforth I will use the terms ‘source context’ and ‘reporting context’ for these two contexts; (...)

35 Building on ideas in Clark & Gerrig (1990) and Clark (1996), Recanati (2000: 172; 2001: 679) proposed that quotations involve a context-shift, namely a shift in (some of) the parameters that govern the interpretation of an utterance. That such a shift takes place is most obvious in DD: any indexical that occurs within a DD report is interpreted with respect to the situation of utterance (or ‘deictic centre’) of the reported speaker, not that of the reporter.7 Consider the deictics in example (16):

(16) He listened, then said, “I told you what I’ll do and I mean every word of it. No retrievers. No followers. Just leave us alone.” (COCA, Mark Davies 2008-)

36The occurrences of I, you and us refer not to the reporter, her addressee, and the reporter plus her addressee, respectively, but to the source speaker (Gary), the source addressee (his mother), and Gary together with someone else. In other words, the indexicals (and the tense form) are ascribed a referent from a context that is not the reporting context. That is an analysis that no one would dispute.

37 Context-shifts, Recanati argues, are pragmatic operations that can be labelled ‘presemantic’. This notion is to be understood within a framework for the interpretation of utterances that is popular amongst formal semanticists and philosophers of language, a framework whose general structure was fixed by David Kaplan (1989). Very roughly, on this model, a well-formed expression (usually a sentence) has a ‘character’ (essentially a compositional meaning built from the conventional meanings of its constituent words), and this character is paired with a context to yield a ‘content’. For example, the character associated with the first-person pronoun I, namely “the speaker or writer of the relevant occurrence of the word I” (Kaplan 1989: 505), when associated with a context, will yield a particular individual that then forms part of the content of the sentence. That content, in turn, must be evaluated at a ‘circumstance’, and that is because the same content (e.g. that Gordon Brown was Prime Minister of the United Kingdom from 2007 to 2010) can turn out to be true in the ‘actual’ world but not in a counterfactual world like the one conjured up by uttering If Gordon Brown had never been PM, ….

38 The above paragraph sketches what Recanati and Predelli, amongst others, have called the ‘semantic machinery’. This is usually articulated with a model of grammar that is inherited from generative syntax, i.e. a model on which syntactic derivation takes place prior to and independent from semantic and phonological interpretation. Since I do not want to be committed to such a model, my arguments below will not appeal to the notion that the output of syntactic derivation is the input to semantic interpretation. A further caveat is that one should not view the semantic model just outlined as capturing how actual utterance processing unfolds. Clearly, the various types of meaning are computed simultaneously by the addressee. What Kaplan’s model does capture, however, are constraints on the completion of processes: thus, in order to complete the identification of the content, one needs to have fully completed the identification of the character. In order to decide if the sentence uttered is true or false, one must evaluate a complete content.

39 Of special interest here is Kaplan’s observation that an utterance(-token) must undergo some processing before it becomes suitable input to the semantic machinery, i.e. before it is identified as being that sentence:

[…] given an utterance, semantics cannot tell us what expression was uttered or what language it was uttered in. This is a presemantic task. When I utter a particular vocable, for example, the one characteristic of the first person pronoun of English, you must decide what word I have spoken or indeed, if I have spoken any word at all (it may have been a cry of anguish). (Kaplan 1989: 559)

40The typical presemantic tasks are (i) the recognition that some linguistic utterance has been produced, (ii) the determination of which language it was produced in, (iii) the disambiguation of any ambiguous string the utterance may contain. The identification of the context with respect to which any indexicals in the utterance are to be given a semantic value is an additional presemantic operation. What transpires from examples like (16) is that there may be more than one context involved: a first context may be relevant for the interpretation of part of the sentence, and a second one for the interpretation of the rest of the sentence. In other words, the context can shift and, when it does, the shift takes place presemantically, because nothing in the semantics of the sentence tells us in what context(s) the sentence was uttered (e.g. who the speaker is).

41 Point (ii) in the previous paragraph suggests that it is not just the context in the narrow sense that can shift; the language according to the norms of which the words of the utterance are to be interpreted can shift too. This is what happens in (17), where the host structure is in English while the DD report is in French.

(17) But the hostess was totally cool, soothingly repeating, “C’est pas grave.” (Lamar 2003: 243)

42In his 2001 paper, Recanati endorses a broad conception of the context as a triple <L, s, c>, where L is a language, s a situation (i.e. a context in the narrow sense of that which is needed to provide indexicals with a referent), and c is a circumstance of evaluation. All three can shift in DD. And, as Recanati shows, that is true, too, in the case of hybrid quotation. Whereas examples (1), (12), (13) illustrated a situation-shift coincidental with the onset of a hybrid quotation, (18) illustrates a language-shift in a similar environment. (I do not illustrate circumstance-shifts; they have no important role to play in the present paper. See Recanati (2001: 677f) for an excellent discussion)

(18) But M. Dumont says that as discontent increases the low efficiency of investment “pourrait encore diminuer”. (Times Literary Supplement, 1970/05/24: 532)
“… the low efficiency of investment ‘could further diminish’”

43 Examples like (18) are very similar to cases of ‘intra-sentential code-switching’ (see Poplack 1980), which are relevant here, because they confirm that identification of the context lato sensu is required in order to yield the proper input to semantic interpretation. The semantic machinery must be ‘informed’ of the language for semantic interpretation and of any changes in it. Failing that — if, e.g. the semantic machinery assumes that mi hermano y yo in (19) is an English constituent — it will simply be unable to provide any semantic interpretation.

(19) This morning mi hermano y yo fuimos a comprar some milk. (Belazi, Rubin & Toribio 1994: 532)
“This morning my brother and I went to buy some milk”

44Example (19), however, remains different from (18) in that it involves no demonstration of the segment in the other language. And that is a good thing, because it shows that the notion of a language-shift necessary to explain (18) is independently motivated.

Context-shifts are selective

45 If context-shifts are really concomitant with demonstrations (and hybrid quotations in particular), then they must be selective; hence there should be at least some cases where they are clearly partial.

46 The examples of hybrid quotation examined so far invite the inference that only one component of the triple <L, s, c> can shift at a time, and also that there isn’t always a shift. Thus, examples like (5) or (11) seem to involve no shift at all; whereas examples like (1), (12)-(15), or (20) below seem to involve only a shift in the situation of utterance, and (18) or (21) below seems to involve only a language-shift.

(5) Gerald said that he would “consider running for the Presidency”. [repeated]
(11) The president differentiated militant groups from the “billion Muslims who reject their ideology”. (The Guardian online, 19/02/2015) [repeated]
(20) For his part, Mr. Powell said the other day in a television broadcast that Mr. Heath “fears me”. (Times Literary Supplement, 1970/11/27: 1375)
(21) He was, he always claimed,
“un individualiste — un farouche individualiste”, with a firm belief in “les valeurs essentielles de la France libérale”. (Times Literary Supplement, 1995/01/27: 7)

47This is the picture that emerges from Recanati (2001) as well. In the absence of a shift, the hybrid quotation is said to be ‘cumulative’ (2001: 669ff) because it entails its disquoted counterpart, as (5) entails (22).

(22) Gerald said that he would consider running for the Presidency.

48 Whenever there is a context-shift, the hybrid quotation is ‘non-cumulative’ (2001: 670), as the sentence that contains it does not entail its disquoted counterpart. Thus, the hybrid quotations in (20) and (21) are said to be non-cumulative. Because of a situation-shift and a language-shift, respectively, they do not entail their disquoted counterparts (23) and (24):

(23) For his part, Mr. Powell said the other day in a television broadcast that Mr. Heath fears me.
(24) He was, he always claimed, un individualiste — un farouche individualiste, with a firm belief in les valeurs essentielles de la France libérale.

49Whereas me in (20) refers to Powell, in (23) it refers to the TLS journalist who reports Powell’s words. The alleged reason why there is no entailment from (22) to (24) is rather different: (24) is judged to be ungrammatical because it contains non-English words (Cappelen & Lepore 1997: 437; Recanati 2000: 289). If it is ungrammatical, the argument goes, it is also semantically ill-formed: its truth-conditions cannot be determined. That, however, is a parochial judgment that fails to acknowledge the existence of intra-sentential code-switching, as in (19). There is every reason to regard (24) as well-formed.

50 There are further problems with the picture sketched so far. Consider (25), which involves overt shifts in both the language and the situation – toi refers to Flaubert’s friend Maxime Du Camp. By contrast, the toi in the disquoted counterpart refers to whoever the TLS journalist was addressing in 1985:

(25) Nor did he flinch when Flaubert, who was collecting asinine items for Bouvard et Pécuchet, told him he had “une quinzaine de phrases de toi qui sont d’une belle niaiserie”. (Times Literary Supplement, 1985/02/01: 126)

51And there is more. If, as Clark & Gerrig put it, “when we hear an event quoted, it is as if we directly experience the depicted aspects of the original event” (1990: 793), then we should expect most of the components of the context to shift with every quotation. At the very least, there cannot be quotation without some context-shift. Recanati himself later amended the analysis proposed in 2001, suggesting that some context-shifts are inconspicuous because they are ‘benign’, and such is the case “whenever the content of the words at issue is the same in the current context and the source context” (Recanati 2008: 456).

52 At that point, Recanati was close to granting that a context-shift takes place with every hybrid quotation, including examples like (5) or (11), but realised that this brought him dangerously close to the semantic accounts that Benbaji and Geurts & Maier had provided for hybrid quotation (2008: 458). We have seen that these writers assume that the quotation marks change the meaning (the character) of the enclosed words to something like “(do) what the source speaker means by [enclosed words]”. As previously pointed out, that appears to neatly capture the meaning of all the hybrids in the examples above, witness the following informal representation of the truth-conditions of (20):

“Mr. Powell said the other day in a television broadcast that Mr. Heath does what Mr. Powell means by “fears me”.”

53 I conceded earlier that I couldn’t articulate a genuine critique of semantic accounts within the space of this article. Here I will simply highlight one problem they face. As predicted by the selectivity principle, when encountering a quotation, the language parameter of the context is not systematically reset to the value it had in the source context. This can be aptly shown in (26), where the TLS reviewer produces two hybrid quotations, only the first one of which treats language as depictive.

(26) But his [Céline’s] certitude as to the stature of his book [Voyage au bout de la nuit] never wavers. It is “une oeuvre sans pareille” and “the great fresco of lyric populism” beyond anything in Zola. (Times Literary Supplement, 2010/02/12: 3)

54 The problem for Benbaji, Geurts & Maier, or Maier is that they should provide a paraphrase as below:

“But Céline’s certitude as to the stature of Voyage au bout de la nuit never wavers. It is what Céline means by “une oeuvre sans pareille” and what Céline means by “the great fresco of lyric populism” beyond anything in Zola.”

  • 8 In a letter to Gaston Gallimard, he wrote: “C’est de la grande fresque du populisme lyrique”, us (...)

55One can certainly accept that Céline meant something by une oeuvre sans pareille. But what could he have meant by the great fresco of lyric populism? Since Céline wrote the source words in French,8 it is unclear that he meant anything by the great fresco of lyric populism, since he never wrote those words. Let me add that it is immaterial that Céline, by his own account, spoke English fluently. The validity of the theory cannot be made to depend on contingent facts like whether source speakers do speak the language in which their words are reproduced.

56 Note further that Benbaji, Geurts & Maier, or Maier could perhaps retort that the quotation in their gloss above need not reproduce the source language… because a quotation is selective! Such a move, however, would only strengthen the view of quotations as demonstrations. And then there is no justification for a semantic account whose reliance on a convention-based mechanism for the generation of quotations is incompatible with the view of quotations as demonstrations.

  • 9 The TLS reviewer’s motivation for this is anyone’s guess.
  • 10 I write likely because the demonstration theory predicts that that shift is optional too. For ex (...)

57 The foregoing discussion shows that the key consequence of Recanati’s amended proposal – that every demonstration (hence every quotation, including hybrid quotations) shifts the context – can be embraced without there being a concomitant need to make concessions to a semantic account of hybrid quotation (cf. De Brabanter 2013). My proposal goes as follows: with every context-shift the values of the three components of the context lato sensu are reset. But such a reset does not automatically result in a perfect fit to the source context. Some aspects may remain tied to the reporting context; in other words they do not shift in the narrow sense of that word. So much follows from the assumption that a demonstration is selective: the quoter may choose not to change the value of every component. We have just seen how, in (26), the reporter once chose to adjust the language parameter to the source context, and once chose not to.9 My description of what happens with the great fresco of lyric populism would go like this: since that string of words is demonstrated, it must be interpreted with respect to a context whose parameters have been reset. This reset, however, does not involve any manifest shift. As said above, the language parameter is the same after and before reset. That there is reset is shown, I believe, by the possibility of modifying the value of the parameter (a possibility that is exploited in the previous hybrid quotation). Similarly, there is no outward symptom of a situation-shift. Yet, again, had a more extensive hybrid quotation included an indexical, that indexical would likely10 have been shifted, as in the great fresco of lyric populism that I penned. As before, I will not discuss the circumstance of evaluation. But as with the other two components of the context lato sensu, there is nothing in (26) that manifests that it has been reset.

58 Perhaps then I should use the term ‘context-reset’ instead of ‘context-shift’. I have not done this here, because I think that shift can be understood in a loose sense, as in Recanati (2008). But I would not object to adjusting the terminology, should other writers consider ‘context-reset’ to be a better term.

59 In this section, I have shown that the demonstration theory, which takes context to shift selectively with every quotation, can account for a category of challenging examples just as well as (and in one case arguably better than) the best semantic theories. I now turn to a final challenge that the demonstration theory must be able to meet.

Avoidance of indexicals within hybrid quotation

60 The last issue I need to address can be formulated as a couple of related questions: why do some writers, or some languages, avoid indexicals in hybrids? Why do some writers, when they use an indexical in a hybrid quotation, choose not to shift it? I indicated earlier in the paper that, based on the data available to me, hybrid quotations containing indexicals appeared to be rather less frequent in languages other than English. While going through the French-speaking press, I have often come across examples like the following.

(27) C’est à ce moment-là qu’il précise aussi qu’il devait se rendre au Stade de France, sans ticket, « pour [s]e faire exploser ». (Le Monde online, 2016/03/25)
“… he was to go to the Stade de France, without a ticket, ‘to blow himself up’”
(28) Le candidat Vert à l’Elysée, Alain Lipietz, a déclaré (Le Figaro 6/8)
qu’« en tant que président de la République il n’appuierait jamais le premier sur le bouton nucléaire ». (Le Canard enchaîné, 03/08/01, quoted in Komur 2004: 65)
“… declared that ‘as President of the Republic, he would never be the first to press the red button’”

  • 11 Both Shan (2011) and Maier (2014) would treat (27) and (28) as exhibiting unquotation. They make (...)

61In my corpus, these kinds of examples are more numerous than the ones with shifted indexicals. And that certainly does not seem the case in the English-speaking press (though that should be ascertained through actual corpus study). (27) and (28) are interestingly different from each other. The former has an explicit correction of the 1st-person reflexive me to 3rd-person se, by means of what typographers call an ‘interpolation’ (and what Shan 2011 and Maier 2014 name ‘unquotation’). That interpolation makes it manifest that the choice of the indexical is not depictive. In the latter example, by contrast, interpolation remains covert.11 Indeed, it is quite unlikely that Mr. Lipietz used the third person to refer to himself. (It is further possible that he used a future tense rather than a conditional form in the source context.)

62 Examples like (27) and (28) suggest a reluctance to reset the situation parameter of the context, and may therefore appear to challenge an account predicated on context-shifts. It is tempting to defuse the threat they pose by appealing, as I did in Section 6, to the selectivity that is characteristic of context-shifts. Though selectivity does play a role, more is needed in the case at hand, because not all quotations behave like hybrid quotations in this respect. Thus, I agree with Authier-Revuz (1993: 12) that in DD all deictics are indexed to the source situation, in other words, that a complete situation-shift takes place. Now, although that is usually the case as well in hybrid quotations in English, this adjustment to the source situation is not systematic. There are cases like (27) and (28) in English as well. Consider (29) and (30).

(29) As a schoolboy, Michael Caine, the son of a charlady and a fish-market porter, was repeatedly taught to respect “his betters”. It took the social revolution of the 1960s, he says, to make it clear that such a hierarchy did not exist. “I’ve met lots of my equals since. But no betters,” Caine told the Observer […]. (The Guardian online, 2017/10/07)
(30) But short of confessing, Chongo did little to disguise his residence. He viewed Yosemite as his rightful home. […] He even bragged to rangers that
“if one simply writes, Chongo, Yosemite, CA on a letter, then he will receive it.” (The New York Times online, 2008/09/30)

63In (29), it is most likely that the quoted words are attributed to Michael Caine. If that is so, Caine must have said your betters or my betters in the source context (cf. You were/I was taught to respect your/my betters). In (30), the quoted words are unquestionably attributed to Chongo. The latter is most likely to have used I, not he, in the source context. (Caution is in order, because there remains a minute possibility that the quotations are actually verbatim. For all we know, Chongo, as an alleged braggart, might have referred to himself in the third person.) The question raised by these examples is this: if, as I claim, hybrid quotation is not a separate variety of quotation, if it is ‘just quotation’, how come it behaves differently from DD?

64 I believe a satisfactory answer can be provided. The vast majority of hybrid quotations that contain indexicals occur in the scope of a reporting verb taking a clausal complement. This means that, when the situation parameter is reset, the speaker/writer is faced with a choice between choosing the source situation as deictic centre — a choice encouraged by the quotation — or the reporting situation — a choice supported by the ID. In DD, by contrast, there is no competition with ID, and the internal structure of the quoted string is segregated from the host sentence. Hence, the only choice as deictic centre is the source situation. That goes some way towards explaining why hybrid quoters are more often led to produce unshifted indexicals aligned with the reporting context.

65 There is a further factor, which might be termed ‘stylistic’, that may help us understand why (some genres in) some languages are more hospitable to shifted indexicals in hybrid quotations. In the English-speaking press, nowadays, it is quite common to produce them; hence, there is little pressure on writers to steer clear of them. In other environments, where that use is not common (or not common yet), writers may be more sensitive to the tension between the demands of ID and those of quotation, so that (i) they tend to avoid producing hybrid quotations containing indexicals, and (ii), when they do produce them, they are more often impelled to use unshifted indexicals, as those are more in line with ‘safe’ ID.

66 It nonetheless remains that unshifted indexicals do not always occur as part of ID. I have found one example so far which diverges from the regular pattern – from an article in Le Monde. In (35), it is highly likely that the quoted words are attributed to Youssef Benayad Daoudi. Yet, the sa in the quotation is indexed to the reporting situation.

(31) Youssef Benayad Daoudi a été l’un des premiers à arriver sur les lieux de l’explosion dans le métro de Bruxelles, mardi matin. Ces images « tournent en boucle dans sa tête depuis hier ». (Le Monde online, 2016/03/23)

67Even in the absence of ID, I would argue, a shifted indexical ma would be more of an infringement of the normal course of things than sa. Every hybrid quotation has to be adjusted to the syntax of its host sentence. Since this host sentence represents the perspective of the reporter, not that of the source, it is likely that there is some pressure on maintaining that perspective throughout. That may be enough to prime the writer to avoid a shifted indexical. Still, I readily acknowledge that more research needs to be done into these questions before an absolutely compelling explanation can be offered.


68 Because hybrid quotations, especially those that contain shifted indexicals, exhibit truth-conditional effects, the majority of writers on quotation have defended semantic accounts of hybrid quotation, tracing truth-conditional effects to the interaction of the conventional meaning of quotation marks with their syntactic environment. While acknowledging the merits of these analyses, I have argued that a completely different kind of theory, a pragmatic theory that takes quotations to be iconic communicative acts, can account for the meaning of sentences containing hybrid quotation just as well.

69 The key mechanism I have appealed to in order to explain the abovementioned truth-conditional effects – context-shift – is a pragmatic one, as it is not ‘controlled’ by any element of linguistic structure, but instead depends on the utterer’s intentions. That pragmatic mechanism affects truth-conditions at a presemantic level. This builds on the idea that, in order to be semantically interpretable, a linguistic expression must have been associated with a language, a situation and a world beforehand. With every demonstration, hence every quotation, the utterer resets the features of the context (language, situation, and world). Not every such feature is systematically shifted to the source context, however, as befits a process that is fundamentally selective. In short, the proposal in this paper reconciles requirements that are typically those of semantics with mechanisms whose non-conventional character ensures that they are pragmatic.

70 In this paper, I have worked on the premise that the demonstration theory offers an adequate account of quotation. I have not demonstrated that it was superior to other theories. In fact, I have pointed out that some competing semantic accounts did a good job of addressing the data I was concerned with. My main purpose lay elsewhere: showing that what may appear as the exclusive preserve of semantic theories — truth-conditional effects — can be handled by a radically pragmatic theory. The other strengths of that theory, which have been argued for elsewhere but naturally require still further support, are the following: first, it does justice to the depictive nature of quotation; it shows how quotation belongs together with a more general communicative phenomenon, demonstration. Second, it assumes that hybrid quotations are not sui generis, and can therefore be explained using the same concepts and mechanisms as enter into the explanation of other quotations, thereby ensuring parsimony. Third, the mechanisms appealed to are independently motivated: situation-shifts are uncontroversially required in the explanation of direct discourse; language-shifts are required in the explanation of code-switching. This again ensures parsimony. Finally, it connects a core property of demonstrations, selectivity, with context-shifts, preventing appeal to the latter from being ad hoc, and succeeding in making correct predictions.

71 No doubt the demonstration theory advocated in this paper needs to be further tested against ‘unfriendly’ data, but this paper shows that it holds a lot of promise.

Top of page


Atkin, A. “Peirce’s theory of signs”. The Stanford Encyclopedia of Philosophy (Summer 2013 Edition), Edward N. Zalta (ed.), URL:, 2010.

Authier-Revuz, J. « Repères dans le champ du discours rapporté (suite). » L’information grammaticale 56 (1993): 10-15.

Belazi, H. M., Rubin, E. J. & Toribio, A. J. “Code switching and X-bar theory: The functional head constraint.” Linguistic Inquiry 25 (1994): 221-237.

Benbaji, Y. “Who needs semantics of quotation marks?” Belgian Journal of Linguistics 17 (2005): 27-49.

Biardzka, E. « Différentes facettes de la mixité des discours rapportés. » In F. Neveu et al. (éds.) Congrès Mondial de Linguistique Française - CMLF 2010. Paris: Institut de Linguistique Française, 2010.

Cappelen, H., & Lepore, E. “Varieties of quotation.” Mind 106 (1997): 429-450.

Cappelen, H., & Lepore, E. “Varieties of quotation revisited.” Belgian Journal of Linguistics 17 (2005): 51-75.

Clark, H. Using language. Cambridge: Cambridge UP, 1996.

Clark, H. “Depicting as a method of communication.” Psychological Review 123 (2016): 324-347.

Clark, H., & Gerrig, R. “Quotations as demonstrations.” Language 66 (1990): 764-805.

Cumming, S. “Two accounts of indexicals in mixed quotation.” Belgian Journal of Linguistics 17 (2005): 77-88.

Davies, M. The Corpus of Contemporary American English: 450 million words, 1990-present. Available online at, 2008-.

De Brabanter, P. “The semantics and pragmatics of hybrid quotations.” Language and Linguistics Compass 4 (2010): 107-120.

De Brabanter, P. “A pragmaticist feels the tug of semantics: Recanati’s ‘Open Quotation Revisited’.” Teorema 32 (2013): 129-147.

De Brabanter, P. “Why quotation is not a semantic phenomenon, and why it calls for a pragmatic theory.” In I. Depraetere & R. Salkie (eds.), Semantics and Pragmatics: Drawing a Line. Springer, 2017, 227-254.

Ducrot, O. « Esquisse d’une théorie polyphonique de l’énonciation. » In Le dire et le dit. Paris: Minuit, 1984, 171-233.

García-Carpintero, M. “Double-duty quotation: The deferred ostension account.” Belgian Journal of Linguistics 17 (2005): 89-108.

Geurts, B., & Maier, E. “Quotation in context.” Belgian Journal of Linguistics 17 (2005): 109-128.

Gómez-Torrente, M. “Remarks on impure quotation.” Belgian Journal of Linguistics 17 (2005): 129-151.

Kaplan, D. “Demonstratives.” In J. Almog, J. Perry & H. Wettstein (eds.), Themes from Kaplan, New York: Oxford UP, 1989, 481-563.

Kasimir, E. “Prosodic correlates of subclausal quotation marks”. ZAS papers 49 (2008): 67-78.

Klewitz, G., & Couper-Kuhlen, E. “Quote-Unquote. The role of prosody in the contextualization of reported speech sequences.” Pragmatics 9 (1999): 459-485.

Komur, G. « Les modes du discours rapporté dans la presse et leurs enjeux polyphoniques. » Pratiques : linguistique, littérature, didactique 123-124 (2004): 57-74.

Lamar, J. Rendezvous Eighteenth. New York: St Martin’s Minotaur, 2003.

Lodge, D. Therapy. London: Penguin Books, 1996.

Maier, E. “Mixed quotation: The grammar of apparently transparent opacity.” Semantics & Pragmatics 7, Article 7 (2014): 1-67.

McCullagh, M. “Scare-Quoting and Incorporation.” In P. Saka & M. Johnson (eds.), The Semantics and Pragmatics of Quotation. Dordrecht: Springer, 2017, 3-34.

Orwell, G. Homage to Catalonia. Boston: Harcourt, 1937.

Poplack, S. “Sometimes I’ll start a sentence in English Y TERMINO EN ESPANOL: Toward a typology of code-switching.” Linguistics 18 (1980): 581-618.

Peirce, C. S. The Writings of Charles S. Peirce: A Chronological Edition. Volume 2: 1867-1871. Edited by the Peirce Edition Project. Bloomington: Indiana University Press, 1984.

Predelli, S. “Scare Quotes and Their Relation to Other Semantic Issues.” Linguistics and Philosophy 26 (2003): 1-28.

Recanati, F. Oratio Obliqua, Oratio Recta: An Essay on Metarepresentation. Cambridge, MA: MIT Press, Bradford Books, 2000.

Recanati, F. “Open quotation.” Mind 110 (2001): 637-687.

Recanati, F. “Open quotation revisited.” Philosophical Perspectives 22 (2008): 443-471.

Reimer, M. “Too counter-intuitive to believe? Pragmatic accounts of mixed quotation.” Belgian Journal of Linguistics 17 (2005): 167-186.

Searle, J. Intentionality. An Essay in the Philosophy of Mind. Cambridge: Cambridge UP, 1983.

Shan, C.-c. “The character of quotation.” Linguistics and Philosophy 33 (2011): 417-443.

Stainton, R. “Remarks on the syntax and semantics of mixed quotation.” In K. Murasugi & R. Stainton (eds.), Philosophy and Linguistics. Boulder: Westview Press, 1999, 259-278.

Top of page


1 I use boldface to facilitate identification of the main string under discussion.

2 See also Ducrot (1984: 199) for an early statement of this type of theory.

3 For clarity, demonstratives are associated with ‘indications’ (acts using indices). This means they are unrelated to demonstrations, which are iconic.

4 My use of the label ‘hybrid quotation’ draws inspiration from Recanati (2001), although Recanati himself spoke of hybrid ‘uses’ of ‘open quotation’.

5 In the sense of Huddleston & Pullum (2002: 329). This is similar to an N-bar in generative syntax.

6 This paragraph is a response to an anonymous reviewer’s criticism that my analyses are inadequately supported by too sparse a corpus. The reviewer suggested that I build a larger and better described corpus, so as to be able to demonstrate, rather than merely illustrate, my hypotheses. I thank the reviewer for encouraging me to be more explicit about the data that I collected. At the same time, I cannot help pointing out that the present paper is not designed as a corpus study, and that its validity in no way depends on comprehensive corpus analyses. No amount of additional data would make a difference to the theoretical thrust of the paper: showing the ability of the demonstration theory of quotation to account for a phenomenon that (i) can be successfully handled by competing, incompatible theories and (ii) seems to pose a particular challenge to the family of theories I endorse. Note also that at no point am I seeking to give a quantitative account of (certain) varieties of hybrid quotations. Finally, in spite of the modest size of my informal corpus, I address a broader range of cases than have been addressed by advocates of competing theories. I hope this goes some way towards assuaging the reviewer’s concerns.

Now, even if I sought to compile an exhaustive corpus of hybrid quotations in, say, the BNC or COCA, the task would be daunting, because no query terms can be devised that will achieve any reasonable degree of precision. The only way to ensure sufficient recall is to look for every quotation mark in the corpus (double inverted commas and single inverted commas). This yields a huge number of irrelevant DD reports and metalinguistic citations, and of apostrophes as well (cf. single inverted commas)! A further problem is that hybrid quotations are not systematically signalled by quotation marks. Hence even looking for all the quotation marks in the corpus does not guarantee perfect recall. I can only conclude that that work is simply not worth doing.

7 Henceforth I will use the terms ‘source context’ and ‘reporting context’ for these two contexts; and will similarly modify the terms ‘situation’, ‘speaker’, etc.

8 In a letter to Gaston Gallimard, he wrote: “C’est de la grande fresque du populisme lyrique”, using fresque as a noncount noun, something else that the TLS journalist did not choose to depict.

9 The TLS reviewer’s motivation for this is anyone’s guess.

10 I write likely because the demonstration theory predicts that that shift is optional too. For examples with unshifted indexicals, see Section 7.

11 Both Shan (2011) and Maier (2014) would treat (27) and (28) as exhibiting unquotation. They make very little of the difference between the explicit and implicit cases.

Top of page


Electronic reference

Philippe De Brabanter, The Interpretation of Indexicals in Hybrid Quotation: A Pragmatic AccountAnglophonia [Online], 28 | 2019, Online since 20 December 2019, connection on 20 April 2024. URL:; DOI:

Top of page

About the author

Philippe De Brabanter

Université Libre de Bruxelles, ULB

Top of page



The text only may be used under licence CC BY-NC-ND 4.0. All other elements (illustrations, imported files) are “All rights reserved”, unless otherwise stated.

Top of page
Search OpenEdition Search

You will be redirected to OpenEdition Search