SOME COMMENTS REGARDING FREGE’S CRITERION OF CORRECT INDIRECT SPEECH REPORT IN THE INDEXICAL POINT OF VIEW

Abstract

Bozickovic’s The Indexical Point of View is a richly informative and solid philosophical work about the problem of cognitive significance involving indexical thoughts and expressions. Although I tend to agree with most of what is said in the book, here I will make some comments on two minor correlated points regarding Bozickovic’s Fregean account of indirect speech reports (or ISRs). After presenting some of the author's ideas about reports, I will claim that the tracking and updating involved in ISRs is of a complex kind that requires attention to the role played by speech reporters and their audiences. Following that, I will hold that a general theory of correct ISR strictly in Fregean terms (as Bozickovic’s) fails to account for that - at least directly.

Keywords:
Indirect Speech Report; Semantic Content; Context; Thought; Speaker's Role

0. Introduction

Bozickovic’s The Indexical Point of View is a richly informative and solid philosophical work about the problem of cognitive significance involving indexical expressions and thoughts that includes discussions of the most relevant contributions to the topic since Frege. One of its motivations is the fact that indexicals and demonstratives pose special challenges for accounts of cognitive significance because of their perspectival and context-anchored semantics, which, among other things, demands an account of the role of tracking in cognitive dynamics: i.e., in diachronic thinking as well as in belief retention about individuals. In that respect, one of the problems tackled by Bozickovic has to do with the kind of tracking involved in the indirect speech reports (ISR, for short) of utterances containing indexicals.

Although I tend to agree with most of what he says in that regard, I will take this opportunity to make two minor correlated points: firstly, that the tracking and updating involved in ISRs is of a complex kind that requires attention to the role played by speech reporters and their audiences and, that a general theory of correct ISR strictly in neo-Fregean terms (as Bozickovic’s) fails to account for that - at least directly. I will begin with an overview of what contemporary philosophers of language take ISRs to be about, focusing on what is commonly affirmed about their theoretical utility. Next, a brief discussion of Bozickovic’s view on ISR is presented and, in the final section, I make some general comments about why I think the Fregean criterion of correct ISR, to which Bozickovic subscribes, should be broadened in the context of a more general theory of ISRs.

1. Looking for a guide to correct ISRs

An ISR is often taken to be a good way to identify what was expressed by the utterance of a well-formed sentence because it (supposedly) captures the stable, shareable, transferable information that the utterance conveys. That is one of the reasons why ISRs appear to be reliable tests to reveal context sensitivity, literality - to differentiate literal from non-literal meaning¹ - and to be good tools in separating what was said from what was implied, implicated etc. by an utterance.

In English, indirectly reported sentences appear embedded in the attitude predicate ‘said that’ which, in the case of expressions whose interpretations are anchored in their contexts of production, such as indexicals and demonstratives, will mandatorily require shifting. To use Bozickovic’s example: when Jack reports Jill’s utterance of (1) the day after its occurrence, he will have to shift the indexical ‘today’ for ‘yesterday’ to preserve the reported word’s reference²,

(1) Today is beautiful.

This is taken by semantic minimalists, for example, as evidence that ‘today’ is a genuinely context-sensitive expression as opposed to words such ‘blue’ in (2), which can be correctly reported by (3) with no need to have its linguistic type shifted³.

(2) Some fish are blue.

(3) Jill said that some fish are blue.

Also, because (3) purportedly captures the “stable, shareable, cross-contextually transferable” meaning of (2), it can be used to help identify the literal meanings of ‘fish’ and ‘blue’ as opposed to their non-literal or modulated⁴ senses, exemplified in (4) and (5), respectively,

(4) The boss is a big fish.

(5) Our planet is blue.

Finally, (3) serves to indicate what content the reported speaker of (2) asserted rather than what she implied or implicated. Compare (3) to (3’), which contains content that is implied by (2),

(3’) Jill said that not all fishes are blue.

or to (3”), which contains a possible conversational implicature⁵ generated by (2),

(3”) Jill did not want to comment on what Jack said and randomly replied that some fish are blue.

While (3) counts as a correct ISR of what Jill said, (3’) and (3”) would count as correct ISRs of either what the sentence used by Jill entails or of what Jill conversationally implicated. As we can see, then, ISRs serve numerous theoretical purposes.

More fundamentally though the act of reporting someone else’s utterances in one’s own words seems to serve the rather commonsensical purpose of transferring a piece of information (conveyed by an original utterer) from its context of production to a different one in the correct way. As Bozickovic (2020) points out: ‘Frege has claimed that a speech report can be correct just in case the thought referred to in the report is the thought expressed by the embedded sentence uttered on its own” (p.122). So, according to the Fregean perspective, a speaker who wishes to report an utterance correctly must be guided by the thought that the speaker intended to express, that is, by the content of the speaker’s belief. For instance, consider (1) again and now take (6) to be its report.

(6) Jill said that yesterday was beautiful.

The embedded sentence in (6), according to Frege, is a correct ISR of (1) because it contains the same thought or propositional content expressed by (1), which is an objective property that is interpersonally sharable and serves to play the role of bearer of cognitive significance. Hence, the Fregean claim that the criterion of correctness of an ISR should be the thought expressed by the original utterance is equivalent to the claim that ISRs are about tracking and preserving semantic content (ibidem, p. 123-124).

Following that line, Borg (2019), a prominent minimalist, affirms that the explanatory role played by the semantic content of an utterance is identifiable via linguistic liability or the speaker’s responsibility.

What we find […] when thinking about the responsibility a speaker assumes for a given content in virtue of uttering the sentence she does in a given context, and the conditions under which retraction of that utterance is required (on pain of a charge of linguistic incompetence) (BORG, 2019, p. 10)

Seemingly, then, linguistic liability is what Borg takes as a useful guide to correct ISR. And rightly so, since liability will be determinant in tracking semantic content/what was said in scenarios involving, for example, legal interpretation or the application of sanctions to harmful linguistic practices, such as defamation (ibidem, pp. 14-16). But it seems hard to see the direct relevance of linguistic liability to all the other scenarios in which the speaker’s responsibility and the application of sanctions are not at issue, such as in the example of (6). Apparently, for all the other cases, thought remains as a better, more plausible candidate to play the role of guide to correct ISR. And as I will discuss in the following section, the theory of belief update developed in Bozickovic’s book is compatible with this “very Fregean” criterion.

2. Thought, indexicals and ISRs

Bozickovic’s interest in ISRs derives from his concern with the function played by tracking in how indexical beliefs are updated, as in the case of ‘today’ in (1) and ‘yesterday’ in (7),

(7) Yesterday was beautiful.

From an intrapersonal perspective, according to Bozickovic, what guarantees that (7) is a correct update of the belief expressed by (1) is the fact that the way Jill thinks about, say, Tuesday (the day of the utterance of (1)) on Wednesday is the same as the way she thought about Tuesday on Tuesday. Or, alternatively, if Jill thinks about Tuesday on Tuesday and on Wednesday under the same mode of presentation, (1) and (7) will express the same thought and, consequently, both ‘today’ and ‘yesterday’ will have the same cognitive significance. Finally, this means that the cognitive attitudes that Jill has towards (1) will be the same as the cognitive attitudes she has towards (7) ⁶.

Bozickovic assumes that reports work similarly to intrapersonal belief updates. So, for (6) to count as a correct ISR of (1) Jack, the reporter, needs to capture and report the way Jill thought about Tuesday. And this will be achieved if what Jack reports is (a) an objective and communicable semantic property conveyed by (1) that is also (b) stable and sharable⁷. In other words, (6) will count as a correct indirect speech report of (1) if, on the one hand, the information originally conveyed by Jill - which is being transferred from its context of production to the context of production of the report - is kept objective, communicable, stable, and sharable. And, on the other, if the cognitive significance of (1) is preserved throughout the process, i.e., through embedment.

As Bozickovic recognizes himself, Frege and Fregeans alike typically faced difficulties in finding a conception of semantic content that could play both parts at once. The difficulties arise from two aspects: first, that shifted indexicals mandatorily belong to different type-expression than the original ones - ‘today’ and ‘yesterday’ differ in meanings or Kaplanian character, for example⁸. Secondly, that properties of the type-expressions are too fine-grained to account for sameness of cognitive significance between the original and the reported indexical.

To better understand both aspects, take the famous Dr. Lauben case and the utterances of (8) and (9) below, which are also discussed in the book⁹.

(8) I was wounded.

(9) You said that you were wounded.

If we account for (9), that is, Frau Lauben’s (or Leo Peter’s) report of what Dr. Lauben’s said with (8), under the assumption that the differences between the properties of the type-expression ‘I’ and ‘you’ determine differences in cognitive significance, (9) will not be a correct ISR of (8), according to Frege’s criterion. As a matter of fact, as acknowledged by Frege (1918) himself, the sense of ‘I’, as uttered by Dr. Lauben, can never be accessed by a putative same-sayer, since it is individuated by the non-sharable way Dr. Lauben thinks about himself.

Bozickovic’s solution to this difficulty appeals to Mark Sainsbury’s notion of token meaning. For Sainsbury (2002), non-synonymous type-expressions, such as ‘today’/‘yesterday’ and ‘I’/‘you’ in the above-mentioned cases, are not simply relatable via sameness of reference¹⁰, they correspond to synonymous tokens. His working hypothesis is that the token of ‘I’ and the token of ‘you’ in (8) and (9) respectively share a meaning, which is not given by the semantic properties of their corresponding types but by the semantic properties of their occurrences. Meaning here is to be understood as a kind of sense or mode of presentation that other words can have¹¹, which is evidenced through speech reports: “If meaning is the least that must necessarily be accessed in understanding, then meaning is specified when speech is reported” (SAINSBURY, 2002, p. 137). Finally, meaning is what explains why the token of ‘you’ in the embedded sentence of (9) could not be replaced by a co-referring expression such as ‘the friend of Peter and Lingens’ without generating an intuition of inadequacy related to the report¹².

Token meanings then seemingly satisfy the requirements of objectivity, communicability, and stability through embedment favored by Bozickovic, allowing for an explanation of how cognitive significance is preserved when indexicals are shifted¹³. In resorting to Sainsbury’s notion of token meaning, Bozickovic has all he needs to preserve the Fregean criterion for correct ISR, namely, to admit that two indexicals that belong to different type-expressions may have the same cognitive significance. His conclusion is that there is no need to abandon Frege.

3. On the reporter’s role

My proposal in this last section will be to go in the opposite direction of Bozickovic’s conclusion, though, and try to answer the following questions: what would a theory of ISRs look like if we abandoned the concern with sameness of cognitive significance between original utterance and embedded sentence? Could such a theory account for correctness of speech report?

Before risking answers, however, I begin with two general remarks. First, that the assumption of parallelism between belief update and ISR on which Bozickovic bases his discussion of ISRs seems to be built on an oversimplification of how reports work. Speech reporters are not just in the business of updating the beliefs they transfer from one context to another, they are in two businesses at the same time: firstly, of capturing what they take the original speaker to be communicating, and, secondly, of allowing their audiences to understand just that. As a result, they update beliefs in terms not only of their own roles as interpreters, but in terms of what information is available for their interlocutors and of what amount (and kind) of information they need to provide in order to accomplish their goal of reporting. Hence, a theory of ISR must consider parallels and asymmetries between matters of intrapersonal and interpersonal belief update, as Bozickovic’s does, but it must also take into account that, in an ISR, an original utterance is reported by someone to someone else. So, its interpretation undergoes a more complex kind of process of update.

At this point, we can briefly return to Borg (2019) and her seemingly correct claim that semantic content/what is said - the objective, sharable, communicable etc. semantic property that is captured by correct reports - is a composite and sociolinguistic notion whose explanatory role in each context depends on social norms and cultural expectations (BORG, 2019, pp.8-9). Behind this idea, lies the rather intuitive view that what counts as semantic content in each circumstance will depend on many variables, such as the speaker’s practical and theoretical goals. That is why sameness of cognitive significance may give room, in some contexts of report, to concerns with speaker’s intentions, implicit content etc. For example, (3”), an ISRs of (2), in a context in which the reportee tries to avoid an embarrassing question, may be a correct ISR of what was said, even though the ISR captures intentions rather than the content of the belief expressed by (2).

My second remark has to do with acknowledging that stablishing a criterion of correct ISR seems to depend on goals that are ultimately determined by the reporter’s reconstruction of the reportee’s cognitive states and of the audience’s informational statuses and expectations. Thought will most certainly count as the best guide in many circumstances, but not always. In some contexts, aspects like, for example, linguistic liability - as we saw in section 1 - might be the most relevant choice of guide¹⁴.

My answer to the two questions is then that a theory of ISR that is more permissive about guides to correctness as a result of taking the reporter’s and the reporter’s audience’s roles into consideration would probably look a bit different from the one Bozickovic is advancing in his book, because it would be goal-oriented instead of strictly Fregean. Correctness of ISR would be determined according to a less straight-forward criterion and in a more context-dependent fashion. Nevertheless, I do not believe that these remarks bring about any serious problems for Bozickovic’s argument, specially bearing in mind that his main concern is with ISRs involving indexicals and not with providing a general theory of ISRs. My comments are meant merely as complementary reflections on his claims about the kind of interpersonal belief update involved in reporting and, ultimately, on Frege’s criterion of correct indirect speech report.

References

• BORG, E. “Explanatory roles for minimal contente”. Noûs, v. 53, n. 3, p. 513-539, 2019.
• BOZICKOVIC, V. The Indexical Point of View: On Cognitive Significance and Cognitive Dynamics Routledge, 2020.
• CAPPELEN, H.; LEPORE, E. “On an alleged connection between indirect speech and the theory of meaning”. Mind & Language, v. 12, n. 3‐4, p. 278-296, 1997.
• _____.; _____. Insensitive semantics: A defense of semantic minimalism and speech act pluralism John Wiley & Sons, 2008.
• CARSTON, R. ‘Truth-conditional content and conversational implicature”. The semantics/pragmatics distinction, p. 65-100, 2004.
• CHAPMAN, S. Paul Grice: philosopher and linguist Springer, 2005.
• FREGE, G. 1918/1977. Thoughts. In Geach, P. T. (ed.), Logical Investigations Oxford: Blackwell, 1977, pp. 1-30.
• KAPLAN. D. “Demonstratives”. Em J. Among, J. Perry, and H. Wettstein (eds.), Themes from Kaplan Oxford: Oxford University Press, p. 481-563, 1989.
• PERRY, J. "Frege on demonstratives." Philosophical Review 86, 47497; reprinted in Perry (1997) p. 325 (page numbers refer to this printing), 1977.
• _____. The Problem of the Essential Indexical and Other essays Oxford: Oxford University Press, 1997.
• _____. Reference and reflexivity Stanford: CSLI, 2001.
• RECANATI, F. Literal Meaning Cambridge: Cambridge University Press, 2004a.
• _____. ‘What is said’ and the Semantics/Pragmatics Distinction. Em C. Bianchi (ed.), pp. 45-64, 2004b.
• _____. Literalism and Contextualism: Some Varieties. Em Gerhard Preyer & Georg Peter (eds.), Contextualism in Philosophy: Knowledge, Meaning and Truth Oxford: Oxford University Press, pp.171-96, 2005.
• _____. On truth-conditional Pragmatics Oxford: Clarendon Press, 2010.
• SAINSBURY, R. M. Departing from Frege: Essays in the Philosophy of Language Londres: Routlegde, 2002.
• STRAWSON, Peter F. “On referring”. Mind, v. 59, n. 235, p. 320-344, 1950.

• 1
  See Cappelen & Lepore (1997) and Recanati (2003, 2004).
• 2
  Cf. BOZICKOVIC, 2020, p. 122.
• 3
  Minimalists argue that context-dependency affects a limited set of expressions composed mainly by indexicals and demonstratives. In their view, the task of truth-conditional semantics is to generate minimal propositions, determined strictly by what is linguistic encoded by sentences. According to some of them, like Cappelen & Lepore (2008), other communicative aspects are dealt with in terms of speech act pluralism.
• 4
  See Recanati (2010).
• 5
  See Chapman (2005) and Carston (2004).
• 6
  Regardless of the fact that Jill associates different information to Tuesday on each day or if she is acquainted with Tuesday on Tuesday and only remembers Tuesday on Wednesday. Such differences are not relevant to the individuation of modes of presentation, as Bozickovic discusses them in his book.
• 7
  Cf. BOZICKOVIC, 2020, p.130.
• 8
  Meanings should be understood here in terms of Strawson’s idea of orientations of use (Strawson, 1950). In the case of Kaplanian characters, Bozickovic takes them to be inadequate in explaining how different indexicals can have the same cognitive significance because they predict different attitudes for different words.
• 9
  Originally from Frege (1918).
• 10
  As is assumed by Kaplan (1989) and Perry (1997, 2001), for instance.
• 11
  Or, in Sainsbury’s words: “a mode of presentation of a subject which others can have, where this mode constitutes the sense of that subject’s tokens of the first person pronoun, and is shared by appropriate tokens of the second and third person pronouns in the mouth of others” (SAINSBURY, 2002, p. 156). Moreover, meanings are Fregean senses that are individuated in terms of the needs of the semantics: a combination of the demands of reported speech and rational cotenability (see SAINSBURY, 2002, p. 155)
• 12
  Sainsbury’s theory predicts that only a de dicto report - and not a de re report - would preserve the rational cotenability of the original utterance in this case.
• 13
  To use Bozickovic’s favored terminology throughout the book: a token meaning is fine-grained enough not to reduce the criterion of correct ISR to sameness of reference. At the same time, it is coarse-grained enough to guarantee that difference in type-expression does not dictate difference in cognitive significance.
• 14
  It could be the case then that a de re report of (8) as ‘The friend of Peter and Lingens was wounded’ counted as correct according to certain contextually determined goals and depending on the audience’s informational status and expectations. If, for example, it was a manifestly known fact to all conversational participants that Dr. Lauben is a friend of Peter and Lingens and if that was taken as relevant to the reconstruction of what was said by Dr. Lauben.

• 15
  CDD: 193.

Publication Dates

• Publication in this collection
  02 Sept 2022
• Date of issue
  Jul-Sep 2022

History

• Received
  02 Dec 2021
• Accepted
  06 Jan 2022