Abstract
In “Demonstratives”, David Kaplan introduced a simple and remarkably robust semantics for indexicals. Unfortunately, Kaplan’s semantics is open to a number of apparent counterexamples, many of which involve recording devices. The classic case is the sentence “I am not here now” as recorded and played back on an answering machine. In this essay, I argue that the best way to accommodate these data is to conceive of recording technologies as introducing special, non-basic sorts of contexts, accompanied by non-basic conventions governing the use of indexicals in those contexts. The idea is that recording devices allow us to use indexicals in new and innovative ways to coordinate on objects. And, given sufficient regularity in the use of indexicals on such devices, linguistic conventions will, over time, come to reflect this innovation. I consider several alternatives to this ‘character-shifting’ theory, but none is able to account for the data as well as the present proposal. Many face additional theoretical difficulties as well. I conclude by explaining how the character-shifting theory not only retains many of the virtues of Kaplan’s original semantics, but also coheres with a plausible view on the nature of semantic theorizing more generally.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
Kaplan himself speaks not of tokens, but rather of evaluating ‘sentences in contexts’. Kaplan focuses primarily on utterances of sentences, but also mentions both written sentences and sentences recorded on tape. Since our focus here will be on voice recordings and playback of such recordings, it will help to have a more flexible term than ‘utterance’ to indicate of the sort of event that associates a sentence with a context and thereby makes it ripe for semantic evaluation at that context. I opt for the term ‘token’ (and the associated event ‘tokening’), which I mean to cover utterances, inscriptions, and also playback of recordings—all of which, I take it, can appropriately be assigned a meaning and truth-value relative to a context.
Perry (2006, 2009) and Sherman (2013) go one step further in clarifying Kaplan’s apparatus by associating each indexical with a particular contextual role—such as the speaker or the time. Such roles, as I understand them, aren’t meant to supersede character-rules. Rather, they help explain how indexicals are used by speakers to coordinate with listeners on particular objects: by guiding their attention to whatever object satisfies a particular role at the context. While I am quite sympathetic to this amendment to (or, perhaps, clarification of) Kaplan’s semantics, it introduces further complications that are best avoided in the main text. If, however, the reader finds it useful to think in terms of contextual roles and satisfiers rather than in terms of functions from contexts to features of those contexts, I harbor no objection.
Note that Kaplan intends for logical truth to come apart from necessary truth. In contrast to (1), “I might not have been here now” looks to be true on most, if not all, occasions of use (Kaplan 1989b, p. 509). Kaplan accommodates this observation by distinguishing between contexts—relative to which indexical-reference is determined, and which are required to be proper—and circumstances—relative to which modal claims are evaluated, and which are allowed to be improper. Since indexical-reference is determined relative to contexts only, not circumstances, Kaplan’s semantics is able to allow simultaneously that: (i) unembedded tokens of (1) will always be false and unembedded tokens of (3) will always be true, and (ii) tokens of (3) embedded under modals may, once the reference of their indexicals is fixed relative to the context, turn out to be false relative to a circumstance of evaluation (or, more precisely, a proposition derived from a token of (3), with the reference of each of its referring terms fixed relative to the context, may turn out to be false relative to one or more circumstance).
If the reader prefers, she is welcome to think of the proposal purely in terms of the question of what constitutes the features of agency, location, and temporality for contexts. The claim is that we find both variation in such features across contexts, but also invariance relative to particular sorts of context-types—regimented according to the type of recording technology employed at that context. In the main text, I will employ the notion of ‘character’ in the broader, informal sense—as referring to rules of reference rather than functions from features of contexts to objects—unless otherwise specified.
‘Ownership’ might not be quite the right notion here. As Jessica Pepp has pointed out to me, we don’t want to say that a token of ‘I’ on a work-related voicemail system refers to the company at which one is employed. ‘Being the original speaker’ is another obvious option, as is ‘being the primary user of the line’. Both of these options face their own sets of difficulties, however. I make no claim here to have found a perfect characterization of the character-rule associated with answering machine tokens of ‘I’, just to had developed a theoretical picture that allows us to productively consider this question. That said, I believe something in the vicinity of ‘ownership’—along the lines of ‘being the recognized possessor of’—might do the job.
Context-typing can be implemented formally in several different ways. I tend to favor modifying our representation of contexts from agent, place, time, world four-tuples to agent, place, time, world, type five-tuples (with the contextual roles above specifying what it means to be the ‘agent’ or ‘time’ relative to a particular context-type). My character-shifting theory would thus claim that, quantifying over all contexts of a particular type, what constitutes an agent, time, or place remains constant.
Note, however, that when the context of production and interpretation do come apart, it is clear that spoken indexicals refer to the place and time of production rather than interpretation (consider, for instance, the truth-conditions of (3) shouted over a long distance, or from a location where the speaker is not visible to her auditor). One might account for this observation either by making spoken indexicals sensitive to the context of utterance and preserving Kaplan’s original thought that ‘here’ and ‘now’ refer to the place and time of the context, or by making indexicals sensitive to the context of interpretation and specifying that ‘here’ and ‘now’ refer instead to the place and time of production. I opt for the latter route.
The case for such intention-sensitivity is two-fold: first, not all uses of ‘here’ uttered in front of a map, even with a pointing gesture accompanying them, refer to a place represented by a point on that map. Rather, speakers can both intend to and succeed in referring to the place at which they are uttering a sentence containing ‘here’ even when this utterance is accompanied by a pointing that seems to be directed at a nearby map (consider the case of a history lecturer with a nervous tick that makes her point erratically, and who is also standing in front of a large map). It is not clear how intention-insensitive accounts, like that of Corazza et al. (2002), will be able to account for this phenomenon. Second, two tokens of ‘here’ uttered while pointing at the same place on the same map can, in fact, refer to distinct places. Consider, for instance, a flight instructor who taps a single point on a map twice and says “We’ll fly here, enter a holding pattern, and then touch down here.” Plausibly, the first ‘here’, but not the second, refers to a region some distance above the ground. Appealing to the flight instructor’s intentions offers an obvious way to fix each of these referents. In contrast, fixing reference here via an appeal to intention-insensitive conventions, as Corazza et al. propose, looks to be a difficult (if not impossible) task. Unless our conventions on map-use include appeals to discourse-context, it is highly unclear how two tokens of ‘here’ accompanied by two qualitatively identical pointings and tappings might end up referring to two very different regions of space.
This lack of a well-defined set of cues that indicate when ‘now’ refers to a past time looks particularly problematic for Corazza et al. (2002), who do in fact propose that the (intention-insensitive) conventions governing ‘now’ specify that it sometimes refers to times in the past. The worry is that, except via an appeal to intentions, there seems to be no promising response to the question: when exactly?
Both of these examples are from Sherman (2013, p. 21). Sherman takes (5) and (6) to illustrate a phenomenon he calls ‘indexical choice’—one interesting upshot of which is that, subsequent to an initial choice-point, the speaker must remain faithful to the ‘frame’, or way of conceiving of the objects in the context, that she has adopted. For Sherman, the semantics of indexicals is best thought of in terms of such framing effects, not in terms of rules of reference. While I am generally sympathetic to Sherman’s approach, I worry that his theory leaves unexplained why the indexicals are, by default, interpreted in only certain ways on many sorts of recordings—like answering machines and postcards—even though alternative interpretations might also be possible. There may, however, be ways of combining our theories so as to account for both of these effects.
I am told that John Perry once had an answering machine message very much like this.
Note that Kaplan was already doing something similar to this when he claimed that what it is to be the agent is to be either the speaker or the author (1989b, p. 505).
Cohen (2013) aptly demonstrates why positing ambiguity within a single utterance ought to be avoided in his discussion of Smith (1989), who posits that indexicals are straightforwardly ambiguous. Smith’s theory predicts, for instance, that there should be a true reading, in a single context and without Alyosius moving, of “Alyosius is here but Alyosius is not here”—namely, when this message is recorded in Alyosius’ absence but played back in his presence (this example is from Cohen 2013, pp. 5–6). I concur with Cohen that a true reading is impossible here. My own theory avoids positing a true reading here by fixing a single character-rule for ‘here’ relative to a context-type.
Romdenh-Romluc (2006, pp. 262–63) offers a variant on this case where these truth-conditions plausibly fail to match our intuitions on what is communicated by the message. The case hinges on the owner of the line having asked someone else to record her message for her. Such ‘friend cases’ are interesting, but I am not convinced they should be dealt with semantically. I therefore leave this objection to the side for the purpose of evaluating Cohen’s proposal. Two things about such cases are worth briefly noting: first, my particular variant of character-shifting theory can deal with such friend cases semantically. However, this doesn’t strike me as a central consideration in favor of the theory. Second, on Cohen’s theory, Romdenh-Romluc (2006)’s friend case actually turns out to be a Gettier case—since the listener will come to believe that the owner of the line isn’t at home on the basis of coming to believe that the recording of (1) is true, and (1) is in fact true iff the owner’s friend isn’t at the owner’s home at the time of playback. Assuming that neither the owner of the line nor the owner’s friend are at the owner’s home when the answering machine is triggered, the listener will have justifiably formed a true belief on the basis of antecedently accepting a true, but irrelevant, content. In spite of such justification, I take it that the listener does not know that the owner is not at home.
In defending his view against an earlier version of this objection, Cohen seems to concur (Cohen 2013, p. 24 fn. 32).
Cohen might claim at this point that our ordinary intuitions actually track pragmatic, rather than semantic, content in this case. That is, Tarek’s use of ‘here’ really refers to whatever he intends it to refer to; we just tend to confuse ourselves about this. However, such a move would be inconsistent with Cohen’s own (convincing, I think) objections to Predelli (2002)’s invocation of the semantics/pragmatics distinction in an analogous context, in response to a very similar set of challenges (Cohen 2013, pp. 11–12).
Another option for preserving Cohen’s view, undiscussed by Cohen himself, would be to claim that indexicals are actually ambiguous. However, Cohen explicitly denies this (Cohen 2013, pp. 8–9). What’s more, he advertises the fact that he can preserve indexical-univocality as one of the primary virtues of his theory.
The basic form of this objection is derived from Corazza et al. (2002, p. 9).
Cohen (2013, pp. 13–15) levels a related objection against Predelli’s appeal to the semantics/pragmatics distinction.
In light of these issues, one might be tempted to instead classify the content conveyed by (1) as recorded on an answering machine as conventionally, rather than conversationally, implicated. This suggestion faces two problems: first, it is not at all clear that conventional implicature is actually a species of pragmatic content. Grice calls it a type of ‘conventional meaning’, seemingly in contrast to conversational implicature (Grice 1989c, p. 121). Since Grice never uses the term ‘semantic content’, it is difficult to say how exactly we ought to map that notion onto his various sorts of meanings. Semantics, however, is often conceived of as tied to conventions as opposed to use—making conventional meaning more likely a species of semantic, as opposed to pragmatic, content. Second, conventional implicatures generally ‘add to’ the total content communicated by an utterance, rather than supplanting the semantic content associated with that utterance (cf. Neale 2001). Here, however, it seems odd to think that answering machines convey both something absurd about the speaker’s state at the time and place where she made the recording and something potentially true about the speaker’s state at the time and place where the answering machine is triggered.
This is essentially a variant of Devitt (1997a, b, 2004) and Reimer (1998a, b)’s ‘Argument from Convention’ against unitary Russellian theories of definite descriptions: pragmatic explanations are well-suited for explaining why we might interpret such-and-such word or phrase in a particular way on one or another particular occasion. They are, on the other hand, significantly less well-suited for explaining why people regularly and consistently interpret a word or phrase in a particular sort of way across a wide range of contexts—and why they are, in fact, justified in doing so. Even if such an interpretive regularity were not originally the result of a convention, it would seem that a convention would be expected to arise over time, supposing that this interpretive regularity were sustained for a sufficient period.
There is an additional worry about Stevens’ proposal as well: the proposal relies on a sharp distinction between saying and asserting according to which, when (1) is played back on an answering machine, nothing is asserted (Stevens 2009, pp. 217–19). Kaplan’s theory, by Stevens’ lights, provides an account of asserted content, something which is only properly associated with utterances, not recordings. Presumably though, something is still said by (1) when tokened on an answering machine. Unfortunately, Stevens fails to specify what exactly he takes this content to be. Even more problematically, the rough gloss he does provide looks to be at odds with Kaplan’s original semantics for indexicals (Stevens 2009, p. 217). This leaves one to wonder both just what Stevens’ theory of the said content of indexicals is, as well as how that said content is supposed to help explain the content asserted by uses of sentences containing indexicals.
References
Bach, K. (1992). Intentions and demonstrations. Logic and Analysis, 52, 140–146.
Cohen, J. (2013). Indexicality and the puzzle of the answering machine. The Journal of Philosophy (forthcoming).
Corazza, E., Fish, W., & Gorvett, J. (2002). Who is I? Philosophical Studies, 107, 1–21.
Devitt, M. (1997a). Meaning and psychology: A response to Richard. Noûs, 31, 115–131.
Devitt, M. (1997b). Responses to the Maribor papers. In D. Jutronic (Ed.), The Maribor papers in naturalized semantics (pp. 353–411). Slovenia: Pedagoska Fakulteta.
Devitt, M. (2004). The case for referential descriptions. In M. Reimer & A. Bezuidenhout (Eds.), Descriptions and beyond (pp. 280–305). Oxford: Oxford University Press.
Donnellan, K. (1966). Reference and definite descriptions. Philosophical Review, 75, 281–304.
Grice, H. P. (1989a). Further notes on logic and conversation. In Studies in the way of words (pp. 41–57). Cambridge: Harvard University Press.
Grice, H. P. (1989b). Logic and conversation. In Studies in the way of words (pp. 22–40). Cambridge: Harvard University Press.
Grice, H. P. (1989c). Utterer’s meaning, sentence-meaning, and word-meaning. In Studies in the way of words (pp. 117–143). Cambridge: Harvard University Press.
Kaplan, D. (1989a). Afterthoughts. In J. Almog, J. Perry, & H. Wettstein (Eds.), Themes from Kaplan (pp. 565–614). Oxford: Oxford University Press.
Kaplan, D. (1989b). Demonstratives. In J. Almog, J. Perry, & H. Wettstein (Eds.), Themes from Kaplan (pp. 481–563). Oxford: Oxford University Press.
Kripke, S. (1977). Speaker’s reference and semantic reference. In P. A. French, T. Uehling, Jr., & H. K. Wettstein (Eds.), Midwest studies in philosophy (Vol. 2, pp. 255–276). Minneapolis: University of Minnesota Press.
Lewis, D. (1969). Convention. Cambridge: Harvard University Press.
Lewis, D. K. (1979). Scorekeeping in a language game. Journal of Philosophical Logic, 8(3), 339–359.
Neale, S. (2001). Implicature and colouring. In G. Consenza (Ed.), Paul grice’s heritage (pp. 135–180). Rome: Brepols.
Perry, J. (2006). Using indexicals. In M. Devitt & R. Hanley (Eds.), Blackwell guide to the philosophy of language (Chap. 17, pp. 314–334). Oxford: Blackwell.
Perry, J. (2009). Reference and reflexivity, 2nd ed. Stanford: Center for the Study of Language and Information.
Predelli, S. (1996). Never put off until tomorrow what you can do today. Logic and Analysis, 56(2), 85–91.
Predelli, S. (1998a). I am not here now. Logic and Analysis, 58(2), 107–115.
Predelli, S. (1998b). Utterance, interpretation and the logic of indexicals. Mind and Language, 13(3), 400–414.
Predelli, S. (2002). Intentions, indexicals, and communication. Logic and Analysis, 62(4), 310–316.
Predelli, S. (2005). Contexts. Oxford: Oxford University Press.
Quine, W. V. O. (1968). Ontological relativity: The Dewey lectures 1969. Journal of Philosophy, 65(7), 185–212.
Reimer, M. (1998a). Donnellan’s distinction/Kripke’s test. Logic and Analysis, 58(2), 89–100.
Reimer, M. (1998b). Quantification and context. Linguistics and Philosophy, 21, 95–115.
Romdenh-Romluc, K. (2006). I. Philosophical Studies, 128, 257–283.
Sherman, B. (2013). Constructing contexts. University of Rochester.
Sidelle, A. (1991). The answering machine paradox. Canadian Journal of Philosophy, 21(4), 525–539.
Siegel, S. (2002). The role of perception in demonstrative reference. Philosopher’s Imprint, 2(1), 1–21.
Smith, Q. (1989). The multiple uses of indexicals. Synthese, 78(2), 167–191.
Stevens, G. (2009). Utterance at a distance. Philosophical Studies, 143, 213–221.
Vision, G. (1985). I am here now. Logic and Analysis, 45(4), 198–199.
Wilson, M. (2006). Wandering significance: An essay on conceptual behavior. Oxford: Oxford University Press.
Acknowledgments
Thanks first to my advisers on this project: David Kaplan and Sam Cumming. Particular thanks are also due to Jonathan Cohen, whose work on indexicals has significantly informed my views. For extensive discussion and comments, thanks to: Zed Adams, Josh Armstrong, Matthew Babb, Kent Bach, Mel Bervoets, Alexis Burgess, Arudra Burra, Alejandro Pérez Carballo, Jorah Dannenberg, Shamik Dasgupta, Kenny Easwaran, Andy Egan, Ashley Feinsinger, Marina Folescu, David Friedell, Nathaniel Hansen, Daniel Harris, Brent Kious, Karen Lewis, Kathryn Lindeman, Dustin Locke, Chauncey Maher, Alex Morgan, Matthew Moss, Sarah Murray, Tyke Nunez, Jessica Pepp, David Plunkett, Alex Radulescu, Andrew Reisner, Indrek Reiland, Brett Sherman, Will Small, and Will Starr. Thanks as well to my audiences at the UCLA Albritton Society, the Rutgers philosophy of language working group, the New York Philosophy of Language Workshop, and the CLAP.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Michaelson, E. Shifty characters. Philos Stud 167, 519–540 (2014). https://doi.org/10.1007/s11098-013-0109-7
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11098-013-0109-7