Dialectic A

16
Quotation and Conceptions of LanguagePaul Saka Abstract This paper discusses empty quotation (‘’ is an empty string) and lexical quotation (his praise was, quote, fulsome, unquote), it challenges the minimal theory of quotation (‘ “x” ’ quotes ‘x’) and it defends the identity theory of quotation. In the process it illuminates disciplinary differences between the science of language and the philosophy of language. First, most philosophers assume, without argument, that language includes writing, whereas linguists have reason to identify language with speech (plus sign language). Second, philosophers tend to think of languages as abstract objects whereas linguists tend to think of them as natural objects. These foundational differences help to explain disagreements in grammaticality judgments and conse- quent disagreements in semantic theory. Roy Sorensen (2008, 57) asks us to “subtract our way to the empty string”, the result being empty quotation: (1) ‘Io’ has two characters. (2) ‘I’ has one character. (3) ‘’ has no characters. Sorensen judges (3) to be grammatical and true, he sees in this a problem for most theories of quotation, and he advocates a version of the demonstrative theory as his own account of the phenomenon. I believe, however, that Sorensen’s arguments contain holes. First I question his basic premise, that (3) is grammatical and true, by recurring to the foundations of the study of language. This has ramifications for the evaluation of grammaticality judgments in general (section 1). Even if Soren- sen’s basic premise be accepted, I argue that it does not actually justify his theory of quotation, which holds that quotation marks are demonstrative referring terms. It is consistent with other theories of quotation as well (section 2). Taken together, the sections of my paper help to advance the recent scholarship on quotation (not only Sorensen’s, but also e.g. Brendel et al. 2011; Cappelen and Lepore 2007; De Brabanter 2005; Predelli 2003; Recanati 2001). 1. Empty quotation: null and void? Patterson (2005) and Sorensen (2008) regard (3), and all syntactically similar instances of empty quotation, as well-formed. They thus contradict Goodman (1974/1977, 44), Gomez-Torrente (2001, 130) and Saka (2006, 456), who regard the likes of (4) as ill-formed. Department of History & Philosophy, University of Texas Pan American, Edinburg, TX 78539, USA; Email: [email protected] dialectica dialectica Vol. 65, N° 2 (2011), pp. 205–220 DOI: 10.1111/j.1746-8361.2011.01266.x © 2011 The Author. dialectica © 2011 Editorial Board of dialectica. Published by Blackwell Publishing Ltd., 9600 Garsington Road, Oxford, OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA

description

Dialectic A

Transcript of Dialectic A

Page 1: Dialectic A

Quotation and Conceptions of Languagedltc_1266 205..220

Paul Saka†

Abstract

This paper discusses empty quotation (‘’ is an empty string) and lexical quotation (his praise was,quote, fulsome, unquote), it challenges the minimal theory of quotation (‘ “x” ’ quotes ‘x’) andit defends the identity theory of quotation. In the process it illuminates disciplinary differencesbetween the science of language and the philosophy of language. First, most philosophersassume, without argument, that language includes writing, whereas linguists have reason toidentify language with speech (plus sign language). Second, philosophers tend to think oflanguages as abstract objects whereas linguists tend to think of them as natural objects. Thesefoundational differences help to explain disagreements in grammaticality judgments and conse-quent disagreements in semantic theory.

Roy Sorensen (2008, 57) asks us to “subtract our way to the empty string”, theresult being empty quotation:

(1) ‘Io’ has two characters.(2) ‘I’ has one character.(3) ‘’ has no characters.

Sorensen judges (3) to be grammatical and true, he sees in this a problem for mosttheories of quotation, and he advocates a version of the demonstrative theory as hisown account of the phenomenon. I believe, however, that Sorensen’s argumentscontain holes. First I question his basic premise, that (3) is grammatical and true,by recurring to the foundations of the study of language. This has ramifications forthe evaluation of grammaticality judgments in general (section 1). Even if Soren-sen’s basic premise be accepted, I argue that it does not actually justify his theoryof quotation, which holds that quotation marks are demonstrative referring terms.It is consistent with other theories of quotation as well (section 2). Taken together,the sections of my paper help to advance the recent scholarship on quotation (notonly Sorensen’s, but also e.g. Brendel et al. 2011; Cappelen and Lepore 2007; DeBrabanter 2005; Predelli 2003; Recanati 2001).

1. Empty quotation: null and void?

Patterson (2005) and Sorensen (2008) regard (3), and all syntactically similarinstances of empty quotation, as well-formed. They thus contradict Goodman(1974/1977, 44), Gomez-Torrente (2001, 130) and Saka (2006, 456), who regardthe likes of (4) as ill-formed.

† Department of History & Philosophy, University of Texas Pan American, Edinburg,TX 78539, USA; Email: [email protected]

dialecticadialectica Vol. 65, N° 2 (2011), pp. 205–220

DOI: 10.1111/j.1746-8361.2011.01266.x

© 2011 The Author. dialectica © 2011 Editorial Board of dialectica.Published by Blackwell Publishing Ltd., 9600 Garsington Road, Oxford, OX4 2DQ, UK and 350Main Street, Malden, MA 02148, USA

Page 2: Dialectic A

(3) ‘’ has no characters.(4) ‘’ has eight letters.

The difference in opinion matters to the viability of the demonstrative theory ofquotation, which treats (4) as analogous to an utterance of ‘That has eight letters’produced in the absence of a demonstratum. The demonstrative theory is thuscommitted to regarding (4) as grammatically well-formed, even if semantically itlacked a subject referent and pragmatically it were infelicitous.

Whenever two grammaticality judgments or intuitions oppose one another,their evidential values cancel out unless one can be shown to be more reliable thanthe other. Sorensen is thus correct to attempt to diagnose the disagreement betweenthe pro-grammaticality and anti-grammaticality judgments: “I take the intuitionthat [4] is ungrammatical to be partly due to [4] being obviously false” (2008, 58).Sorensen’s explanation competes with another, however. Instead of saying that thefalsity of (4) creates an illusion of ill-formedness, I believe that (3) may appear tobe good because it communicates a truth. Yet the communication of truth does notrequire a complete grammatical sentence: witness pantomime, elliptical speech,and the solecisms of stutterers, stammerers, foreign-language flounderers andJoycean jokesters (cf. Barton 1990; Progovac et al. 2006; Stainton 2006; Cappelenand Lepore 2007, 112).

It is not incidental that (3), in isolation from the likes of (1) and (2), wouldtrouble most readers. (To isolate (3), remove it both physically and cognitively.Imagine interpreters who have not just seen the likes of (1) or (2), and do not havethem on their mind.) Tokens (1) and (2) induce a reading on (3) that (3) does notnecessarily express by itself.1 The phenomenon of context-induced reading can befound in other cases as well, for instance in the following fragment of a ditty byHughes Mearns:

(5) Yesterday upon the stair,I met a man who wasn’t there.He wasn’t there again today,Oh how I wish he’d go a-___.

Just as the rhyme scheme in (5) conspires with our lexical knowledge to evoke theabsent syllable ‘way’, the structure of (1-3) can transport the reader beyond whatis literally expressed; and just as we don’t react to (5) by thinking that its last lineis ungrammatical, we don’t stop to worry about the syntax of (3) when it trails after(1) and (2). In isolation, though, (5)’s last line would be ill-formed, and (3) wouldbe questionable.

1 It has been suggested that empty quotation can be recognized without the benefit ofcontext in ‘Io’ has two characters, ‘I’ has one character, and ‘’ has no characters. But eventhough the guiding cues in this case are in the same sentence, they are still part of the emptyquotation’s context.

Paul Saka206

© 2011 The Author. dialectica © 2011 Editorial Board of dialectica.

Page 3: Dialectic A

Beyond contextual filling-in, I believe that two additional factors account forthe divergent reactions to (3). Both stem from disciplinary differences between thescience of language and the philosophy of language regarding the exact object ofinquiry. First (section 1.1), linguists usually think of languages as natural objects– as belonging to the causal and empirical order, being for instance psychological,social or biological objects – whereas philosophers tend to think of languages asabstract objects. (The latter attitude can be found in mathematics and computerscience as well, but these disciplines do not purport to describe natural languagesas philosophy of language ostensibly does.) Second (section 1.2), linguists offi-cially identify language with speech whereas philosophers focus on writing. (Mytwo claims are formulated in terms of tendencies and emphases; they do notpretend to be universal and absolute.) Sections 1.1 and 1.2 contain general lessonsfor scholars of language in addition to supporting my main argument, whichresurfaces in section 1.3.

1.1 Natural versus abstract objects (and objection to the minimal theoryof quotation)Linguists usually think of languages as natural objects whereas philosophers tendto think of languages as abstract objects. The naturalistic approach is explicitlyrepresented by Chomsky (2000), while the opposing tendency is manifested byCappelen and Lepore and by Sorensen. (I mention Chomsky because his influencein linguistics has been epic, and I mention Cappelen, Lepore and Sorensen becausethey are leading philosophers of language who have worked on quotation, andbecause their work is part of my present focus. The attitudes of all, moreover,typify or represent the attitudes widely found in their respective disciplines.)

For example, Cappelen and Lepore (2007, 42), rebutting the claim that expres-sions can be mentioned (in the sense of use versus mention) without beingquotation-marked, hypothesize a counter-example language. Now if Cappelen andLepore intended to cite a nomological possibility for the class of natural humanlanguages, their argument would beg the question. It therefore seems that Cap-pelen and Lepore are citing a merely logical possibility. Another example occurswhen Cappelen and Lepore take issue with the widely embraced principle thatsyntactic structure is a guide to semantic structure.2 Since this principle threatenstheir project, Cappelen and Lepore try to find a counter-example to it, namely acase where recursive syntax does not match recursive semantics.

2 Cf. the homomorphism requirement of Montague (1970/1974), the GrammaticalityConstraint of Jackendoff (1983), the rule-to-rule hypothesis of Cann (1993), the default pre-sumption of Saka (1998), the Interface Rule of Stainton (1999), the ceteris paribus desideratumof Pietroski (1999), the fundamental tenet of Barker (2004) and the direct compositionality ofBaker and Jacobson (2007).

Quotation and Conceptions of Language 207

© 2011 The Author. dialectica © 2011 Editorial Board of dialectica.

Page 4: Dialectic A

Consider the recursive orthographic rule: ‘If a is a quotable item of English, then sois a concatenated with a.’ For example, since ‘Aristotle’ is a quotable item ofEnglish, then so is ‘AristotleAristotle’ . . . (1999, section 5).

But the reason I can quote ‘AristotleAristotle’ is not that I can quote ‘Aristotle’, itis that I can quote any sequence of syllables whatsoever. Hence Cappelen andLepore’s rule, though it helps to specify a possible language in the logicians’ senseof ‘language’, is not a psychologically real, causally efficacious rule in any naturallanguage.

Yet another example appears in Cappelen and Lepore (2007). According totheir minimal theory of quotation, the semantics of quotation is completely cap-tured by the following quotation schema:

(QS) “ ‘x’ ” quotes ‘x’.

But while QS may be an abstract analytic truth, it hardly counts as a naturalisticexplanation. To explain a phenomenon, a theory must explain it to someone whodoes not already understand it; yet no one can understand the explanans in QSwithout first understanding the explanandum. The minimal theory, in other words,is circular or vacuous (Saka forthcoming). The problem is compounded, actually.First, unless we already understand what quotation is, we do not understand thequotation marks in the explanans. Second, unless we already understand whatquotation is, we do not understand the verb ‘quotes’ in the explanans. As a result,positing a mental representation of QS would not explain how a speaker couldhave competence in using quotation.

Sorensen too implies that he regards language as an abstract object when hecompares natural languages to formal languages, which he does when he mentionsthe programming language Perl. For a linguist, however, natural and formallanguages are either disjoint, being as different as avians and aviators, or at leastas distinct as is the set of birds from the set of things that fly. (It’s true that manylinguists use formal languages to model natural languages, but this does not meanthat they identify the two; indeed, a great deal of work is dedicated to findingconstraints that limit the class of formal languages that can be so used.)

For those who believe that any syntacto-semantic system whatsoever counts asa language, and who want to express what (3) conveys, empty quotation will berecognized as grammatical; those who believe that languages are constrained byinnate universal grammar, by associationist psychology, by history, by pronounce-ability or by other empirical facts will wonder whether empty quotation is trulygrammatical, or even doubt that it is.

1.2 Speech and writing (and defense of Recanati)There is a second disciplinary difference that can fund disagreements in gram-maticality judgment. According to standard theoretical linguistics, spoken lan-

Paul Saka208

© 2011 The Author. dialectica © 2011 Editorial Board of dialectica.

Page 5: Dialectic A

guages constitute a natural kind, and written languages do not belong to this kind.(The status of manually signed languages is complicated, and can be sidelined forcurrent purposes; see note 4 and Saka 2007b.) Most philosophers, in contrast,assume that written language is a part of natural language. (Jacques Derridanotoriously claims that writing precedes speech and indeed that all language istext, but my discussion of philosophy is restricted to the analytic tradition.) Searle(2007, 17) makes the philosophers’ point of view explicit when he writes, “we canignore phonology because it is not essential to language that it be spoken”.Strawson (1970/2001, 111) goes so far as to say that utterances “need not bevocal”. Additional examples are easily found in the literature on quotation (e.g.Goodman 1974/1977, 55; Washington 1992, 589ff; Reimer 1996, 134; Saka 1998,119; Recanati 2001, 649; Cappelen and Lepore 2007, 147ff). Indeed, an almostexclusive focus on writing has deafened philosophers to speech, making CoryWashington’s work exceptionally worthy (cf. De Brabanter 2002, 73ff). The bestevidence for my claim, however – that philosophers typically assume that writingcounts as language – is found in referee comments and in the incredulity that Iencounter when I tell my philosopher friends that theoretical linguists assumeotherwise.

The linguistic point of view is that only speech counts as language. This view,henceforth the speech-only thesis, is made clear in the classic texts of modernlinguistics. Says Saussure: “Language and writing are two distinct systems ofsigns; the second exists for the sole purpose of representing the first” (1916/1959,23). Jespersen concurs: “the linguist in principle operates without regard forwritten forms. He takes them into consideration only insofar as they may onoccasion influence the form of vocal signs” (1924/1965, 17). Bloomfield too:“Writing is not language, but merely a way of recording language” (1933, 21).Even George Trager, who is known for emphasizing the importance and interest ofwriting, agrees that writing is distinct from language (1972, 18). The samemessage is chorused by Sapir (1921, 20),3 by Chomsky (2000, 56 and 78),4 and by

3 While Sapir writes of a “point-to-point equivalence” between written language andspeech, this equivalence is clearly meant to fall short of identity: “even those who read and writewithout the slightest use of sound imagery . . . are merely handling the circulating medium,money, of visual symbols as a convenient substitute for the economic goods and services of thefundamental auditory symbols”; “all voluntary communication of ideas, aside from normalspeech, is either a transfer, direct or indirect, from the typical symbolism of language as spokenand heard or, at the least, involves the intermediary of truly linguistic symbolism”. (Actually, toclaim even equivalence between writing and speech goes too far, as demonstrated by Nunberg1990.)

4 Writing that the language faculty “is not tied to specific sensory modalities”, Chomsky(2000, 121) evidently means only that language includes sign languages, given his repeatedaffirmation that languages have phonic systems (in linguistics ‘phonology’ encompasses thestudy of the elements of both speech and sign language; the study of written elements, in contrast,goes by a different term, ‘orthography’).

Quotation and Conceptions of Language 209

© 2011 The Author. dialectica © 2011 Editorial Board of dialectica.

Page 6: Dialectic A

every successful linguistics textbook that I am familiar with: Hockett (1958, 4),Martinet (1960/1966, 17), Fromkin (2000, 38 and 592), Akmajian et al. (2001, 8and 109), O’Grady et al. (2001, 11) and the Ohio State University Department ofLinguistics (2004, 8). The latter, for instance, states:

One of the basic assumptions of modern linguistics (as opposed to linguistics beforethe beginning of the twentieth century) . . . is that speech is primary and writing issecondary . . . Spoken language encodes thought . . . while writing, in turn, encodesspoken language (italics added).

The recurring idea is that writing is not language in part because it does notrepresent the same sort of thing that clear, central cases of language do: speechrepresents the world, while writing represents speech.5

To be more precise, when I speak of speech I am speaking loosely, for as a ruletheoretical linguists are not interested in actual spoken utterances. Theoreticallinguists oriented to society, following Saussure, are interested in langue ratherthan parole, langue being a social system that is abstracted from the vagaries ofparole; and those in the psychologistic tradition, following Chomsky, are inter-ested in competence rather than performance, competence being embodied in amental faculty that is one factor in the production of behavioral performance. Now– just as speech differs from writing – social systems of speech differ from socialsystems of writing, and mental systems of speech (representations and faculties)differ from mental systems of writing. Whenever I say that linguists identifyspeech with language, therefore, I mean that they regard speech-directed systems,be they social or mental, as the genuine object of their inquiry.

My characterization of theoretical linguistics does not extend to those appliedlinguists who study second-language learning. The study of how to improveforeign-language pedagogy, in any society where literacy is as important as it is inours, rightly emphasizes reading and writing at least as much as oral comprehen-sion and production. Still less does my characterization extend to those appliedlinguists who are not scholars at all, but practitioners who have MAs from clinicalprograms, which are separate from theoretical programs. Because writing can beused to record speech, it is sometimes convenient to treat writing as virtual speech,and indeed the difference between the two is sometimes forgotten even by theo-retical linguists, myself included. Nonetheless, the canonical view in theoreticallinguistics is that language and writing are as distinct from each other as a circuitis from its diagram.

The idea that writing primarily represents speech rather than the world willstrike some as inherently implausible: when one of the Nuremberg prosecutorsrecently wrote that George W. Bush should be tried as a war criminal, surely hereferred to Bush rather than the spoken word ‘Bush’. But this particular response

5 Olson (2001) observes that in representing speech, writing resembles quotation.

Paul Saka210

© 2011 The Author. dialectica © 2011 Editorial Board of dialectica.

Page 7: Dialectic A

to the speech-only thesis ignores the difference between linguistic reference andspeaker reference. That a person may refer to Bush by writing ‘Bush’ does not byitself demonstrate that the written mark ‘Bush’ denotes Bush, for denotation is aspecific kind of reference that is peculiar to natural language. By analogy, supposethat in a game of charades I point at my eye and you correctly guess ‘I’. This doesnot mean that a token of eye-pointing denotes the first-person singular pronoun ofEnglish or that it denotes a person. The speech-only thesis, therefore, cannot bedismissed on the grounds of naked denial or brute intuition; judging it requiresreasoned evidence.

In the only sustained critique of the speech-only thesis that I have seen, aself-professed piece of iconoclasm, Harris (1986) correctly argues that letters can’trepresent phones for the usual reasons that are familiar to linguists. He then argues,however, that letters can’t represent phonemes either, as the invention of thealphabet took place prior to modern phonemic theory. This is like saying thatvarious color words can’t represent colors, as they were invented prior to moderntheories of light, reflectance and vision. Harris adds that letters can’t representphonemes because the relationship is not one-to-one, which is like saying that‘Roma’ can’t represent Rome because it represents other entities too and becausethere exist other terms that represent Rome. Not all of Harris’s discussion is sospurious, but even if he established that writing does not represent speech, it wouldnot follow that writing is language.

Among opponents of the speech-only thesis, many of whom indicate thatwriting is a better example of language than speech is, I suspect some buy Plato’saxiom that the abiding is ‘more real’ than the transitory. For others, because somuch human neurological activity is devoted to visual processing, visual experi-ences may seem more real than the auditory. Then too, because so much scholar-ship is conducted on paper, it is unsurprising that students of words would fixateon writing rather than speech unless they have specific training to do otherwise.Durability, salience and convenience, however, are not reasons for thinking thatwriting is language.

The speech-only thesis is motivated by the fact that speech is natural in waysthat writing is not. It is learned automatically and effortlessly: human infants arewired to begin emitting vowels and consonants before they even begin to walk. Asa result speech is culturally universal, found in every human society; it is sociallyuniversal, found in every class and in every functioning member of society; and itsacquisition is a prerequisite for the acquisition of writing. In contrast, writing is acultural peculiarity, existing for but a tiny slice of humanity’s time on earth; it isdifficult to transmit, normally requiring deliberate, formalized instruction formodest returns (the college essays you see are written by an elite minority of thetotal population); and its meager punctuation marks hardly represent prosodicfeatures – intonation, stress and tempo – which are rich and essential in speech.

Quotation and Conceptions of Language 211

© 2011 The Author. dialectica © 2011 Editorial Board of dialectica.

Page 8: Dialectic A

Spoken language thus appears to be a natural kind that excludes writing, andmoreover it seems reasonable for this natural kind to be designated ‘naturallanguage’.6

It’s possible that proponents of the speech-only thesis mean for it to be takenfiguratively, for instance as hyperbole, but I think we should take the speech-onlythesis seriously. First, as a methodological matter: scholars deserve to be taken attheir word, absent very good countervailing reason. What would count as a goodreason to suppose that an author meant something other than what he or she said?Falsehood alone would not scotch the legitimacy of a literal interpretation (Saka2007c), and inherent implausibility does not apply to the speech-only thesis (asobserved four paragraphs back). Second, as a rhetorical matter: it seems unlikelyto me that so many linguists, across so many generations, would make the samemistake of formulating literal versions of the speech-only thesis without clarifyingtheir true intents if they all meant something figurative. Third, as a substantivematter: if formulations of the speech-only thesis were to be taken as hyperbole, thequestion would arise as to what it is hyperbolic of; how would a literal formulationgo? Instead of ‘writing is not language’, should we say that writing is not entirelylike language (which seems to entail the literal truth of ‘writing is not language’)?Or should we say that some writing is not language? If the latter, which elementsor instances of writing are linguistic and which are not? If some writing islinguistic and some is not, an obvious proposal would be that all graphic marksthat correspond to speech in some admissible way, and only those, are linguistic.Since the quotation marks in empty quotation do not correspond to pronounceablematerial, even a scaled-back, unexaggerated speech-only thesis would countempty quotation as non-linguistic.

To hold that writing is no part of natural language is not to deny that speech andwriting may jointly comprise some larger interesting and important natural kind, aputative natural kind being a category recognized by some scientific theory oranother. Natural kinds are multidimensional, but for the sake of simplicity considerbiological taxa: canines and felines belong to both the kind carnivora and awhole hierarchy of larger kinds including mammal, vertrebrate and animal; incontrast spiders and spider monkeys do not jointly belong to any biological kindexcept at a high level that also includes fish and silverfish. Now the question athand might be formulated in terms of the taxonomic relatedness of speech andwriting. Is there, or is there not, a taxonomic node that dominates speech andwriting without also dominating systems that we would rather not classify aslanguage?

6 Non-realists can accept natural kinds, as I use the term. One can hold that science isinterest-driven, carving categories according to parochial human contingencies, yet still use thelaws and theories of successful science to identify kinds that are natural in comparison to thosedefined by legislative fiat, religious tradition, commercial practice or personal whim.

Paul Saka212

© 2011 The Author. dialectica © 2011 Editorial Board of dialectica.

Page 9: Dialectic A

An umbrella kind, in order to count as language as the term is used not only bylinguists but by analytic philosophers of language as well, would have to excludeinfrahuman animal communication systems, non-verbal social conventions andmost of the other codes that semioticians study. Although such a natural kind mayvery well exist, I know of no argument for it. Speech and writing both servecommunicative functions, but so do many other signaling systems that are gener-ally deemed non-linguistic. Speech and writing are also thought to share theproperty of having recursively compositional semantics, but the same could besaid of computer languages and of honey-bee language. Other shared traits as doexist can be traced to the thesis that writing is but a representation of speech.

Philosophers and linguists are not merely talking past each other, one using abroader conception of language than the other. The two disciplines genuinelydisagree. First, philosophers assume that speech and writing constitute and exhaustone natural kind while linguists do not. Second, philosophers presumably believethat their assumption is justified, and linguists do not. On the first issue I find thephilosophers’ position somewhat plausible, but on the second issue I must sidewith the linguists. If language is regarded as a natural object in the sense thatlanguage scientists rather than philosophers get to decide what counts as language,then written sentences are not part of language; if language is a natural object inthe sense that languages belong to a natural kind, then there is no evidence thatwritten sentences belong to language while, say, capybara calls and Cobol do not;and if language is a non-natural kind then its study would be completely uncon-strained, and for that reason uninteresting, unless some rationale for delineating itin some particular way were actually provided. Such rationale, however, has notbeen forthcoming.

Recognizing the difference between phonocentric and graphocentricapproaches to the study of language helps to clear up a number of confusions. Inthe quotation literature, for example, Recanati (2001, 661) claims that (6) and (7)are the same sentence.

(6) Alice said that life is difficult to understand.(7) Alice said that life is ‘difficult to understand’.

In response, Predelli (2003, 19) calls Recanati’s claim “bizarre”. But Recanatidoes not mean that (6) and (7) share the same written form; he means that (6) and(7) represent the same linguistic sentence because quotation marks are paralin-guistic. (Recanati’s text is admittedly unclear. It says that quotation is a non-linguistic phenomenon yet accepts the role of prosody, which is widely regardedas a linguistic feature; and it says that quotation can be marked prosodically whilesaying that speech carries no marks of quotation proper.)

To summarize, the speech-only thesis identifies language with ‘speech’ (asocial abstraction of speech or a mental representation of speech), denying that

Quotation and Conceptions of Language 213

© 2011 The Author. dialectica © 2011 Editorial Board of dialectica.

Page 10: Dialectic A

written language is language in the operating sense of the word. I have observedthat philosophers tend to reject the speech-only thesis and I have given evidencethat it is standard for linguists to accept it; I have countered arguments for rejectingthe speech-only thesis and I have reported traditional arguments for accepting it;and I have used the foregoing ideas to make sense of a certain disagreement in thequotation literature. My next step is to apply the speech-only thesis to resolve thedisagreement regarding the grammaticality of empty quotation.

1.3 Lexical and punctal quotationAlthough written form (3) cannot blindly be accepted as being a strict instance oflanguage, might it represent an instance of language?

(3) ‘’ has no characters.

In other words, might (3), like non-empty quotation, be read aloud so as to yieldan instance of language? The correct answer, I believe, rests on a distinctionbetween describing and pronouncing. To specify (3) orally you might describe it:

(8) It begins with an open quote, then a close quote, . . . then it ends with afull stop.

Alternatively you might use a mixture of description and straight reading:

(9) Quote, unquote has no characters.

In (8) we have a grammatical sentence, and possibly in (9) as well; more on thatanon. But both cases give descriptions of, not a pronunciation of, quotation marks.Compare: the word ‘comma’ is a description, not a pronunciation, of ‘,’. Thecomma can represent, signal or reflect prosodic effects on surrounding words, andquotation marks sometimes do the same. But in the absence of words separated bya comma, the comma has no pronunciation; and in the absence of words inside ofquotation marks, quotation marks have no pronunciation. A straight reading aloudof (3), therefore, would be linguistically indistinguishable from a straight readingaloud of (10).

(10) has no characters.

And (10), although grammatical as a predicate, is not grammatical in the relevantway, in the way that (3) allegedly is (as a sentence). Therefore empty quotations,such as (3), do not belong to the set of grammatical natural-language sentences.

Recanati (2001, 649) asserts that erstwhile non-linguistic elements canundergo “linguistic recruitment”. For instance, following Larry Horn, Recanaticites musical notes (more pedantically, the playing thereof) that seem to functionas singular terms when a piano teacher says and does the following:

Paul Saka214

© 2011 The Author. dialectica © 2011 Editorial Board of dialectica.

Page 11: Dialectic A

(11) It’s not [plays C sharp], it’s [plays C flat].

If Recanati is right then perhaps the inverted commas in (3) can similarly berecruited to function as genuine linguistic elements.

Unfortunately, Recanati gives no argument for his claim that the playing ofmusical notes actually becomes part of language in token (11), and it strikes me asimplausible that it does. First, (11) seems ungrammatical. I can easily imagineuttering it, but only if e.g. I subvocalized demonstratives so as to mentally token:

(12) It’s not this [plays C sharp], it’s this [plays C flat].

Second, if the playing of C sharp has been linguistically recruited then it hasentered language, in this case presumably English.Yet the playing of musical notesdoesn’t seem to be part of English or any other natural language. Indeed, ifRecanati seriously meant that all communicative elements are linguistic elementsthen he would have to say that (6) and (7) above are linguistically distinct, but aswe’ve seen he denies that. What’s more, elsewhere he rightly follows Reimer indistinguishing between demonstrations and demonstratives, but regarding (11) hetreats the playing of a piano key as both demonstration and demonstrative.Because Recanati’s position is counter-intuitive, unargued for and arguably incon-sistent, I reject it as a basis for holding that empty quotation marks have beenrecruited into natural language.

I’ve contended that (3) does not represent a grammatical spoken sentence, andnow I am doubtful of (9) as well. For to be a grammatical spoken sentence, it isplausible to suppose that (9) might require a grammatical subject; and to have agrammatical subject, one suspects there must be a noun phrase; yet it is question-able whether the words ‘quote, unquote’ serve as a noun phrase. To my ear, thefollowing two utterances are equally effective at dramatically, suspensefully orplayfully reporting that Grace stood silent, and I conclude they are equallyungrammatical:

(13) Grace said –.(14) Grace said, quote, unquote.

What’s more, (9) does not strike me as neologistic, enjargonated or pedantic somuch as just plain artificial – ‘quote, unquote’ does not fill a gap in the expressivepower of English, it compensates for it. ‘Quote, unquote’ is an unwieldy device,one which native speakers have trouble with, as evinced by the fact that speakersare more wont to say (15) or (16) than (17):

(15) Her boss said she is a quote-unquote asset.(16) Her boss said she is an asset, quote-unquote.(17) Her boss said she is a, quote, asset, unquote.

Quotation and Conceptions of Language 215

© 2011 The Author. dialectica © 2011 Editorial Board of dialectica.

Page 12: Dialectic A

Finally, the fact that ‘quote, unquote’ can float from position to position reinforcesmy idea that it is not a noun phrase. It functions like an adverbial parenthetical,which means that (13) is no more grammatical than ‘Grace said loudly’ or ‘Gracesaid fortunately’.

We can thus distinguish between lexical marks of quotation and punctal marksof quotation. Lexical quotation can be either written, as in (17) itself, or spoken, asin what (17) represents; and punctal quotation can be either written or spoken, asin (1) itself or (1) pronounced with special prosody (the quoted word is lengthened,it is amplified, it gets special intonation and it precedes a pause).7 Empty punctalquotation can be written, witness (3), but it cannot be pronounced so as to yield asentence.

2. Empty quotation: no problem!

I have argued that empty quotation, as a phenomenon of grammatical English, mayvery well not even exist. For this reason alone it poses no immediate threat totheories of quotation that do not recognize it. But suppose I am wrong; supposethat (3) is grammatical and furthermore true, or that (9) is. What would theconsequence be? Would empty quotation refute those theories ‘that count thequotation marks as part of the quotation’, such as the name theory of Tarski andQuine? Under such theories, according to Sorensen, the likes of (3) would be false“because they undercount the number of characters” (2008, 58). But this commitsa use-mention fallacy. Just because quotation marks are part of the quotation – theexpression that a speaker uses when quoting – that does not make the quotationmarks part of the quotatum, i.e. the quotation’s referent (cf. Gomez-Torrente 2010,440). It’s as fallacious to hold (18) as (19):

(18) Under the theory that “ ‘Io’ ” is a name, (1) is false because its gram-matical subject actually has four characters.

(19) Under the theory that ‘Mars’ is a name, ‘Mars is red’ is false because itssubject term is actually just black ink.

Is it possible that Sorensen, in objecting to theories ‘that count the quotation marksas part of the quotation’, actually meant to address theories that count quotationmarks as part of the quotatum? If so, his criticism would be valid but without force:it would not actually apply to his targets, Tarski and Quine, nor to any otheraccount ever proposed (cf. Saka 2011). Indeed, if Sorensen’s reasoning weresound, his targets could be refuted by (1) and (2), without invoking emptyquotation.

7 If (1) were read without special prosody, the spoken form would not be an instance ofpunctal quotation; having no quotation-marking at all, it would be plain-mentioned.

Paul Saka216

© 2011 The Author. dialectica © 2011 Editorial Board of dialectica.

Page 13: Dialectic A

And what of the identity theory, the view that “the speaker uses the expressionbetween quotation marks to refer to itself” (Sorensen 2008, 57)? Is it incompatiblewith the phenomenon of empty quotation, the premise that (3) or (9) is grammati-cal? I do not see why it would be. Sorensen compares the empty string to suchabsences as holes, pauses and shadows, and he states that such can be perceptiblebecause of their surroundings. I think he is right, and further agree with him thatabsences ontologically depend on relevant surroundings, or frames: the donut holecan be seen, and indeed exists, because of the donut; the shadow can be seen, andexists, because of the light that frames it; pauses can be observed, and exist,because of the activities that frame them. Relative nothingness thus exists and canbe perceived in the right circumstance, and moreover it can be pointed at, used andreferred to. Now if a shadow can be used for lurking in, and an oral cavity forlodging delectables in, and if the empty string is analogous to shadows andcavities, then surely empty strings can be used, for instance by speakers and for thepurpose of reference. Writers can use an empty string for the purpose of referencebecause they are not limited to using only an empty string – they have quotationmarks at their disposal too. It is by using quotation marks as context, with nothingin them to speak of, that writers can create an empty string token so as to refer toan empty string (type or token).

To see more precisely how my particular version of the identity theory relatesto empty quotation, consider the following principles adapted from Saka (2011).

(Q) When speaker S uses an initial double apostrophe, then S signals that S isgoing to mention what comes next; and when S uses the next doubleapostrophe, S signals that S has finished mentioning something.

(M) S mentions x if and only if S produces a token of x, intending to refer tosome linguistic feature that is closely related to x (e.g. x’s pronunciation-type or x’s word-type).

If Sorensen is right then written empty strings are linguistic tokens, in which casethey are capable of being mentioned according to principle (M), and hence capableof being quoted according to principle (Q). On the other hand, if the speech-onlythesis is right then (Q) ought to be dropped in favor of a principle that is indepen-dently needed anyway:

(Q′) When S uses spoken signals to indicate that S is mentioning x, be theylexical or prosodic, then S quotes x.

In this event, grammatical quotation of the empty string, by prosodic means, doesnot look possible. Either way, empty strings pose no threat to (M) and (Q/Q′);either they have already been accommodated by our quotation conventions or theystill are not, contra Gomez-Torrente (2010, 442).

Quotation and Conceptions of Language 217

© 2011 The Author. dialectica © 2011 Editorial Board of dialectica.

Page 14: Dialectic A

So far I have argued that the identity theory is safe. What is in danger, I mightadd, is Sorensen’s account. Although Sorensen’s example suggests an appealinganalogy – a donut is to a donut hole as a quotation mark is to an empty quotation– evidently it is not what Sorensen has in mind:

We perceive absences by virtue of how they contrast with their surroundings . . . Thisexplains why it is easier to see the empty string in the context of other sentences(2008, 59).

Instead of saying that the empty string in (3) is framed by quotation marks,Sorensen seems to be saying that it’s framed by sentences elsewhere on the page,namely (1) and (2):

(1) ‘Io’ has two characters.(2) ‘I’ has one character.(3) ‘’ has no characters.

What’s more, for Sorensen the empty string “is a substring of every string” (Ib.,58). This would entail that every text contains infinitely many empty strings, whichwould render Sorensen’s account seemingly inconsistent, as it’s unclear how itsquares with comparing empty strings to donut holes. Positing an infinity of emptystrings, unexceptionable for the abstract formalist, is also dubious from the natu-ralistic point of view.

My own thinking borrows from Sorensen’s core idea that absences implicatesurroundings, but otherwise differs. For one thing, my account (Saka 1998, 2006,2007a, 2007b, 2007c, 2011) does not treat quotation marks as demonstrativereferring terms, as does Sorensen’s. For another, my account suggests that emptystrings exist only where there is the framing of emptiness (normally, only wherethere are empty quotation marks); their number is not infinite. Finally, on myaccount it does not necessarily follow that speakers can use empty punctal quo-tation as writers can. On the face of it, punctal quotation marks are marks ofpunctuation, and punctuation is syncategorematic. Punctuation can signifychanges in cadence and tempo, but only when there are syllables to carry them.When there is no vocal material, as in the case of empty quotation, then punctua-tion, as a speech phenomenon, cannot exist. Granted, speakers can compensate forthis expressive lack by making up new conventions, as by use of finger quotes orlexical quotation. But if these conventions do not yield natural grammatical con-structions, they do shift the topic.

I believe that speech is the primary form of language, contrary to most phi-losophers. I suspect that writing, though secondary, belongs to an umbrella naturalkind that is interesting, important and aptly called language, contrary to standardlinguistics (earlier I observed that philosophers haven’t justified this umbrellaassumption; I never denied its possible truth). I have documented how differences

Paul Saka218

© 2011 The Author. dialectica © 2011 Editorial Board of dialectica.

Page 15: Dialectic A

between philosophers and linguists can ground differences in grammaticalityjudgment. I believe that Cappelen and Lepore’s minimal theory of quotation doesnot explain how speakers acquire, deploy or comprehend quotation marks; Ibelieve it does not explain any natural fact whatsoever. I have defended Recanatiagainst Predelli. I believe that punctal empty quotation is not grammatical, con-trary to Sorensen’s claim (as understood by linguists). I believe, however, thatlexical empty quotation just might be grammatical, thanks to Sorensen’s insight.Finally, against Sorensen, I have argued that even if punctal empty quotation begrammatical, it would not undermine the identity theory of quotation.*

References

Akmajian, A., Demers, R. and Harnish, R. M. 2001, Linguistics, fifth edition, Cambridge, MA: MITPress.

Baker, C. and Jacobson, P. 2007, Direct Compositionality, Oxford: Oxford University Press.Barker, S. 2004, Renewing Meaning, Oxford: Oxford University Press.Barton, E. 1990, Nonsentential Constituents, Amsterdam: John Benjamins.Bloomfield, L. 1933, Language, New York: Holt, Rinehart & Winston.Brendel, E., Meibauer, J. and Steinbach, M. (eds) 2011, Understanding Quotation, Berlin: De

Gruyter Mouton.Cann, R. 1993, Formal Semantics, Cambridge: Cambridge University Press.Cappelen, H. and Lepore, E. 1999, ‘Reply to Saka’, Mind 108, pp. 741–750.Cappelen, H. and Lepore, E. 2007, Language Turned on Itself, Oxford: Oxford University Press.Chomsky, N. 2000, New Horizons in the Study of Language and Mind, Cambridge: Cambridge

University Press.De Brabanter, P. 2002, Making Sense of Mention, Quotation, and Autonymy, doctoral dissertation,

University of Brussels, URL=<http://tel.archives-ouvertes.fr/docs/00/15/52/94/PDF/BIGdOC.pdf>.

De Brabanter, P. (ed.) 2005, Hybrid Quotations, Amsterdam: John Benjamins.Fromkin, V. (ed.) 2000, Linguistics, Oxford: Blackwell.Gomez-Torrente, M. 2001, ‘Quotation Revisited’, Philosophical Studies 102, pp. 123–153.Gomez-Torrente, M. 2010, ‘On Quoting the Empty Expression’, Philosophical Studies 148, pp.

439–443.Goodman, N. 1974/1977, ‘Some Questions Concerning Quotation’, repr. in his Ways of Worldmaking,

Indianapolis: Hackett, pp. 41–56.Harris, R. 1986, The Origin of Writing, London: Duckworth.Hockett, C. 1958, Course in Modern Linguistics, New York: Macmillan.Jackendoff, R. 1983, Semantics and Cognition, Cambridge, MA: MIT Press.Jespersen, O. 1924/1965, The Philosophy of Grammar, repr. New York: Norton.Martinet, A. 1960/1966, Elements of General Linguistics, trans. by E. Palmer, Chicago: University

of Chicago Press.Montague, R. 1970/1974, ‘Universal Grammar’, repr. in R. Thomason, ed., Formal Philosophy, New

Haven: Yale University Press.Nunberg, G. 1990, The Linguistics of Punctuation, Stanford: CSLI.O’Grady, W., Archibald, J., Aronoff, M. and Rees-Miller, J. 2001, Contemporary Linguistics,

fourth edition, New York: St Martin’s.

* I gratefully acknowledge assistance from David Lebeaux, James Myers, RoySorensen and anonymous dialectica referees.

Quotation and Conceptions of Language 219

© 2011 The Author. dialectica © 2011 Editorial Board of dialectica.

Page 16: Dialectic A

Ohio State University Department of Linguistics 2004, Language Files, eighth edition, Colum-bus, OH: OSU Press.

Olson, D. 2001, ‘What Writing Is’, Pragmatics & Cognition 9, pp. 239–258.Patterson, D. 2005, Commentary for Paul Saka’s ‘Quotation Matters’, Central division meeting,

American Philosophical Association.Pietroski, P. 1999, ‘Compositional Quotation’, in: K. Murasugi and R. Stainton, eds, Philosophy and

Linguistics, Boulder, CO: Westview, pp. 245–258.Predelli, S. 2003, ‘Scare Quotes and their Relation to other Semantic Issues’, Linguistics and

Philosophy 26, pp. 1–28.Progovac, L., Paesani, K., Casielles, E. and Barton, E. (eds) 2006, The Syntax of Nonsententials,

Amsterdam: John Benjamins.Recanati, F. 2001, ‘Open Quotation’, Mind 110, pp. 637–687.Reimer, M. 1996, ‘Quotation Marks: Demonstratives or Demonstration?’, Analysis 56, pp. 131–141.Saka, P. 1998, ‘Quotation and the Use-Mention Distinction’, Mind 107, pp. 113–115.Saka, P. 2006, ‘The Demonstrative and Identity Theories of Quotation’, Journal of Philosophy 103, pp.

452–471.Saka, P. 2007a, How to Think about Meaning, Dordrecht: Springer.Saka, P. 2007b, ‘Language’, The Language of Science, Monza: Polimetrica.Saka, P. 2007c, ‘Spurning Charity’, Axiomathes 17, pp. 197–208.Saka, P. 2011, ‘The Act of Quotation’, in: E. Brendel, J. Meibauer and M. Steinbach (eds), Under-

standing Quotation, Berlin: De Gruyter Mouton, pp. 303–322.Saka, P. forthcoming, ‘Does “ ‘Quotation’ ” Quote “Quotation”?’, Protosociology.Sapir, E. 1921, Language, New York: Harcourt, Brace, Jovanovich.Saussure, F. de. 1916/1959, Course in General Linguistics, trans. by W. Baskin, New York: The

Philosophical Library.Searle, J. 2007, ‘What is Language’, in: S. Tsohatzidis, ed., John Searle’s Philosophy of Language,

Cambridge: Cambridge University Press, pp. 15–45.Sorensen, R. 2008, ‘Empty Quotation’, Analysis 68, pp. 57–61.Stainton, R. 1999, ‘Remarks on the Syntax and Semantics of Mixed Quotation’, in: K. Murasugi and

R. Stainton, eds, Philosophy and Linguistics, Boulder, CO: Westview, pp. 263–284.Stainton, R. 2006, Words and Thoughts, Oxford: Oxford University Press.Strawson, P. F. 1970/2001, Meaning and Truth, repr. in: A. P. Martinich, ed., Philosophy of Language,

fourth edition, Oxford: Oxford University Press, pp. 110–120.Trager, G. 1972, Language and Languages, London: Chandler Publishing.Washington, C. 1992, ‘Quotation’, Journal of Philosophy 89, pp. 582–605.

Paul Saka220

© 2011 The Author. dialectica © 2011 Editorial Board of dialectica.