A Davidsonian version of Dissemination and Abandonment

This essay has two parts:1 The first part argues that Davidson and Derrida agree on a central point. The very structure of the naturalistic Davidsonian account of language means that any account of the idiolect of an utterer or inscriber at a time is subject to modification by that agent’s own application of that idiolect in producing utterances.2 Davidson’s account of how truth-definitions are grounded entails that accounts of the meaning even of particular utterances are unstable in virtue of the very process by which we ascribe meaning to utterings and inscribings. Davidson is (correctly) committed to the intrinsic necessity of dissemination and “play.”

The second part discusses utterances, iterations, and the way in which an uttering or ascribing is “abandoned” as soon as it happens.3 The product of an inscribing or uttering—namely, an inscription or utterance—can be iterated in another uttering with different force and truth-conditions from what was historically present in the original event. This iteration can still be a copy and so an iteration of the original uttering. While there can be exact reports of the force and truth-conditions of any utterance or inscription, no iteration of the sequence of words of a tensed sentence can do so. The essay adapts the concept of “copy” developed in an earlier work on textual identity to account for these phenomena.4 Being a copy of an uttering or inscription is iterating that very utterance or inscription. An iteration is a copy that says something close enough to an original to be saying the same thing.

I. Davidson’s account of meaning and its disseminative consequences

Decades ago, I argued that indeterminacy of interpretation, construed in the terminology of Husserl’s phenomenology and Saussure’s conception of language, was very close to Derrida’s “deconstruction.”5 I have come to see that the very structure of Davidson’s account of language commits him to a version of “dissemination,” the fluidity and “play” of ascriptions of meaning. The core of Davidson’s account of language and interpretation guarantees that natural languages will be disseminative.

Husserl construed meanings as entities that in principle could be common to different languages. Frege’s “senses,” the Analytic tradition’s version of Platonic forms, were similar metaphysical equipment. Languages were ways of expressing such meanings.

Derrida and Davidson attacked trans-linguistic meanings. Derrida treated metaphysical trans-linguistic meanings historically, as an aspect of philosophy’s obsession with Presence, which had shaped the history of metaphysics. Davidson had a research project focused on what a systematic account of understanding a language must be, connecting interpretation with action and intention. Both Derrida and Davidson doubted that there could be a complete account of a language.6

Both Derrida and Davidson argued that force and intention could not be expressed in truth-conditions (Davidson) or included in a “code” for meanings (Derrida).7 Davidson’s argument was that, if there were a word or stress-pattern that signaled sincerity, it could be used by an actor on the stage.8 Derrida puts the point differently, by the metaphor that an utterance, having been made, is out of the speaker’s control. The audience can take an utterance with different force than intended, and nothing about the utterance can guard it from such misinterpretation.

Davidson adapts Tarski’s account of a truth-definition.9 He proposes that a truth-definition is an adequate account of meaning.10 A truth-definition takes elementary parts of a language, particular expressions, and elementary ways of combining those expressions, and defines an iterated procedure by which an infinity of sentences is assigned meanings. Assigning a meaning to a sentence is giving a sentence in the interpreting language that would have the same truth-value. A theorem of a truth-definition is a sentence in which an utterance-type of the speaker is mentioned and what the world would be like if it were true is given. “‘Grass is green’ would be true if and only if grass were green.”11 Note that the second occurrence of “grass is green” is being used by the interpreting clause, not mentioned. An expression of the speaker being interpreted is mentioned (referred to) on the left side, and the interpreter uses one of their sentences to say how the world would be if the speaker’s sentence were true.

To understand a speaker’s language is to have an account such that any possible utterance of the speaker (at a time) is paired with a sentence of the interpreter’s language which would have the same truth-value. The empirical theory of the speaker, couched in the interpreter’s language, gives the meanings of the interpreted individual’s sentences. A speaker has an unlimited repertoire of possible utterances. A truth-definition provides the meaning of each of them, and so must be compositional. The contributions of expressions to truth-conditions must be tracked as expressions become more complex, ending with truth-conditions of complete sentences, the theorems of the theory.12

The meanings assigned are sentences of the interpreter’s language. So, of course, different interpreters will give different accounts. Those accounts can be accurate and equally right. If there are no trans-linguistic meanings, this is what “giving the meaning of a person’s utterance” amounts to.

Tarski showed how to construct such an algorithm for artificial languages whose elementary parts and constructions are given artifacts. Davidson’s idea was to construct such an algorithm for actual natural languages. So, the elements, both words and the constructions that built sentences with truth-conditions, had to be derived from what speakers say in various situations. Davidsonian truth-definition, interpreting agents speaking for reasons, has as its basis speech-actions of the individual being interpreted. Understanding a language is a special case of understanding intentional actions.

According to the Davidsonian truth-definitional theory of meaning, expressions of the interpreter’s idiolect at the moment fill the role of Fregean senses, Platonic Forms, and other versions of notion that there are trans-linguistic meanings which expressions of natural languages express. Davidson thinks that an interpreter’s idiolect at a time, however fluid and constantly changing, is the only ground for communication and understanding another. It works pretty well, but there is no real possibility of a language or interpretation scheme which does better. I and my resident squirrels share a world in which we recognize the presence of acorns. We have different degrees of interest in them. But we are able to interpret squirrel behavior as intentional action, without supposing that squirrels have a “concept” of “acorn” or thoughts about the coming winter. We can undertake much more detailed attempts to understand humans, especially when they are producing speech for reasons. That’s what truth-definitions do. Squirrels seek acorns for comprehensible reasons. People say things for comprehensible, obscure reasons. The next sections discuss limits to the degree of detail and reliability of our understanding of those reasons.13

Davidson’s account of language learning and interpretation uses the notion of “triangulation” among a speaker, an interpreter, and components of their shared world.14 The common world of the interpreter (language-learner) and the speaker or inscriber (trainer) is central.15 The interpreter sees speech actions, presentations of sentences, happening in a shared environment.16 The interpreter assigns meaning to parts of the speaker’s sentences in the interpreter’s own language. In effect, the language of the interpreter replaces the traditional trans-linguistic meanings as what the speaker’s utterings express.

The learner or interpreter conjectures forces and intentions of speech actions in order to construct a theory of truth-conditions. The learner-interpreter must also assign referents to demonstratives and indexicals. Forces and intentions are not encoded in the linguistic objects being interpreted, as we discuss at length. Referents of demonstratives and some indexicals (“we” for instance) are linguistically restricted, but not determined linguistically. Tenses are indexicals, so ascribing a referent to an indexical is always a part of interpretation. An interpreter likewise has to guess which Tom the speaker is referring to with “Tom.”17

Referents of names, demonstratives, and indexicals are parts of inscribings and utterings that are not available in their linguistic products, utterances and inscriptions. However, unlike the force with which an utterance is presented, the referents of demonstratives and indexicals are part of the truth-conditions of linguistic actions. “That man has four children” is true or false depending on who the speaker has in mind.18

On the basis of interpreting speech-actions, an interpreter conjectures simultaneously truth-conditions, sometimes guessing referents of demonstratives and indexicals, and the forces with which those truth-conditions were presented. The interpreter is guided by the consideration that the speaker is an agent with mostly true beliefs and values that are comprehensible. The theory is constrained and guided by the fact that, if the event is meaningful, the speaker or inscriber is an agent producing expressions on purpose. Interpreting an agent as doing things on purpose in the world requires treating the agent as having comprehensible desires and an understanding of the world in which action can achieve desires. Speech actions can bring about intended results. Applying the principles of the intentional system of concepts, the interpreter constructs a theory of truth-conditions for the speaker at the time.19

Truth-conditions of sentences make them equipment for speech actions with various forces. “The door will be closed at five” has truth-conditions which determine what is commanded or predicted. Which action the speaker is doing cannot be not encoded in truth-conditions. But the content of what is commanded or predicted depends on the truth-conditions of the utterance. The utterance may be either a command that the door will be shut or a prediction that the door will be shut. The truth-conditions deliver what the command would be and what the prediction would be.20 The theory generating those truth-conditions is a truth-definition for a speaker at a time. That theory is based on past speech-actions of the speaker, taken in context, etc.21 Ideally, the theory of the truth-conditions of the speaker’s utterances, actual and possible at the moment, is accurate.

But when the speaker produces new utterances, the basis for assigning those truth-conditions has changed. The speaker is using those truth-conditions for novel actions. Even if a truth-definition correctly applies to a speaker at a time, the speaker is at liberty to use those truth-conditions for whatever purposes, including non-literal figures, extensions beyond previous applications, and so on. But the occurrence of these new actions changes the data on the basis of which truth-conditions were assigned. The speech action basis for interpretation, based on the earlier data of interpreted actions, has altered. The truth-definition for the speaker at the moment does not thoroughly survive the first utterance which it interprets. This is the Davidsonian expression of “différance,” the fact that given an ideal truth-definition, that truth-definition is not quite right as soon as the interpreted speaker begins to add another speech-act into the database from which a truth-definition must be constructed.22 A truth-definition is always already aufgehoben.

Thus, even given an ideal truth-definition that optimally takes into account the speaker’s speech actions up to the moment, the speaker can use those truth-conditions for speech-actions which extend the application of a predicate in novel ways or apply a term as a figure.23 The speaker thus adds new data to the base and renders the truth-definition possibly not accurate. An account of meaning, a truth-definition, can only be completely optimal, accounting for every speech-action data-point, before a speaker speaks. When there is a best truth-definition for a speaker at a time, any new uttering or inscribing will add novel data making that truth-definition not quite adequate to the data. The truth-definition must in principle be re-calculated for the new database during every new inscribing.24

Truth-conditions are tools that speakers and inscribers use to make meaningful speech actions and inscribings. The action, a speaking for a reason, depends on truth-conditions, but the intent cannot be part of truth-conditions. The actor trying to tell us that there is a fire in the theatre by shouting “There is a fire in the theatre” is not aided by “really” or “believe me,” which can also be part of the script. Derrida and Davidson agree that a central part of meaning, the force with which an expression with given truth-conditions is presented, is part of the uttering or inscribing but cannot be part of the utterance, the linguistic product, the text.

Davidson’s account of language use and understanding thus builds dissemination into its foundation. Intentional actions determine truth-conditions, and truth-conditions are applied in novel actions, determining possibly different truth-conditions applied to yet newer speech actions, and so on. Truth-definitions are derived from speech actions, which depend for their nature on prior truth-definitions. This circle in Davidson between actions as data for truth-definition construction and the freedom of speakers to use expressions for a variety of purposes means that language-meaning is guaranteed to be fluid. Davidson refers to truth-definitions as “passing theories.”25 I think he was well aware of the circle and its disseminative consequences.

Language change is not words coming to express different meanings, as it would be for Frege. Fluidity of meaning is entailed by any account that is naturalistic and recognizes that force cannot be encoded and that every account of meaning is couched in a language whose interpretation rests on the same sort of data as does the interpretation of the speaker being interpreted. It is transparently a consequence of Davidson’s basic picture. There is a sort of ground, namely the shared objective world in which we learn to speak by observing what people say when, but that ground does not provide what senses and noemata do. A language of thought into which natural language utterances are paraphrased would give determinate results. A domain of Platonic Forms as metaphysical under-pinning of trans-linguistic meanings would likewise give determinate results, given an account of the mind’s access to such entities.

Davidsonian dissemination, as we have described it so far, accounts for the non-surprising fact that languages evolve and that terms and expressions come to have different meanings.26 It does so without supposing a stable ground of meanings relative to which correct interpretations of expressions come to be different over time. Without trans-linguistic meanings, “language evolution” is the fact that, given a fluid metalanguage, the interpretation of an utterance in an object language will vary. On the account of which meanings are metaphysical entities expressions express, language evolution is just something that happens. Derrida and Davidson show that drift is not an accident but a consequence of there being no domain of meanings. If there is no domain of trans-linguistic meaning the interpreter’s language, with its drift, has to fill the role. It does, but it does not quite meet the expectations that trans-linguistic meanings would justify.

II. Intention, truth-conditions, and abandonment

Derrida’s notion of “iterability” is more than just that there can be multiple occurrences of the same word-sequence. We argue below that iterations are copies. When iterations can differ in force or truth-conditions from an uttering (or inscribing), that uttering is “abandoned” to reuse in ways that the agent cannot control. Those reuses can be iterations of that uttering or just duplicates. No iteration or duplicate of an uttering, even by the original speaker, exactly reproduces force and truth-conditions.27 Universal claims such as “Two is prime” or “Iron is a metal” are present tense. Still, there are other occurrences of the same or close to the same words which are iterations. They can differ from the original in many ways.

It is of course possible to utter or inscribe a sequence of words with exactly the same truth-conditions as an utterance; such utterings are reports, not iterations. They fail to be copies. Iterations are about words. And report-sequences differ too much. Suppose Fred says, “On my honor, I will do my best to do my duty to God and my country.” There are two ways you can say what Fred said. The first, you saying that “Fred said that on his honor he would do his best to do his duty to God and his country” does not iterate Fred’s uttering. It is not a copy. The actual words matter. The content of the “that”-clause has the very same truth-conditions as Fred’s uttering, but different pronouns and tenses. The other way is for you to say, “On my honor, I will do my best to do my duty to God and my country.” This utterance by you is a copy of that text Fred has recited, but the references of the indexicals “my” and the present tense differ from those of Fred’s uttering. Your uttering “On my honor …” and Fred’s uttering “On my honor …” have different truth-conditions but are iterations of some original inscribing or uttering. For such cases, both utterances are iterations, but the iterations are by chains to an original. Neither is an iteration of the other.

This section argues that one inscription (or utterance) x is an iteration of inscription y if and only if x is a copy of y.28 An inscribing may be a copy of another inscribing, and so an iteration, but not a word-for-word duplicate.29 An inscription may be a duplicate of another inscription but not an actual copy, and so not an iteration. A duplicate that is not a copy does not have the right relation to an original. Two electrons are duplicates but not copies of one another. Two simultaneous utterances of “It is five o’clock” may be duplicates while neither is an iteration of the other. The immediate mutated descendant of a virus is a copy of that virus but not a duplicate of it. We explicate Derrida’s notion of “iteration” in terms of the “copy of” relation.

Speech and writing actions differ from their products. We can call the actions utterings and inscribings. An uttering’s product is an utterance. An inscribing’s product is an inscription. An uttering or inscribing is “abandoned” because features of the uttering or inscribing are not linguistically available to the audience in the utterance or inscription. Iterations of what is linguistically available, whether by the speaker or others, differ in truth-conditions from the original simply because the present and past tenses are indexicals. The force of an original is not linguistically available, so every utterance or inscription is “abandoned.” Repetitions of utterances with indexicals generate expressions with different truth-conditions but may be copies, and so iterations. The referents of demonstratives are part of the uttering or inscribing, but not available in the linguistic object the speaker or inscriber presents. Someone’s “that” has a referent the speaker “has in mind” but the audience may mis-guess that referent.30

We argue that differences from an original in indexicals—when, where, or by whom an uttering or inscribing is made—do not always disqualify an iteration from being a perfect copy of an original. Likewise, differences in referents of terms and meanings of predicates do not always disqualify an utterance from being an iteration of the original uttering. How far an uttering by another can differ from the original in force or truth-conditions and still be a copy of that original is vague, as we discuss below. A re-occurrence of the same words with the same referents and same meanings of predicates may not be an iteration of the same uttering or inscribing if the utterance is not a copy of the uttering. We understand the “copy of” relation to be the iteration relation.

a. The “copy of” relation31

Whether a product that re-occurs in a use by a later speaker is an iteration, though intended differently or with different truth-conditions, depends in part on whether the re-occurrence is a copy or a duplicate.32 A word-for-word duplicate of an utterance or inscription need not be a copy, and so need not be an iteration. The numerous occurrences of “It is time for supper” are duplicates but not copies of some original speaker’s or inscriber’s utterance.

Tolhurst and Wheeler, whose 1979 view we adapt, held that a copy not only had to be connected by the right sort of chain of copyings of the words with their semantic form, but also had to have the same intention, the same force. They took for granted, as I do, that some rational agent’s intention is “behind” every meaningful uttering or inscribing.33 Their theory was in many respects a version of Knapp and Michaels’s.34 My view is that the “copy” relation requires a chain of the right sort between iterations, but that the copies may differ in force from the intention of the original producer. A copy can also differ in truth-conditions when the difference depends on indexicals.

“A is a copy of B” is not transitive. If A is a copy of B and B is a copy of C, then, pretty reliably, A is a copy of C.35 But a long enough chain of copies starting with Tristram Shandy, each having one typographical error, but each a copy of its predecessor, could start with Sterne and end with the Critique of Pure Reason. At some point, the copy ceases to be a copy of Tristram Shandy, even though each scribe is doing his best to transmit the assigned book itself.36

b. Iteration and differences

Since force, as both Derrida and Davidson argue, cannot be encoded in linguistic forms, matching the force of an uttering or inscribing cannot be a requirement for being an iteration. But the truth-conditions (the meaning) of an uttering or inscribing are also not completely given in the utterance or inscription. The available linguistic meaning of an utterance or inscription is the surface form of the utterance. That is what a potential iterator has to work with. A sequence of words may have differently referring expressions, determined by the covert intentions of the producer. The same sequence of words may have different truth-conditions in virtue of differences in semantic form or different meanings of homonyms. Referents of indexicals and demonstratives also yield different truth-conditions.

1. Different semantic form and homonyms

Utterances with the same words may have different truth-conditions by having different covert structure, or terms with multiple truth-definition clauses, different lexical meanings. An inscribing is an expression with its semantic form. “My daughter attends a pretty little girls’ school” has several different possible truth-conditions, readings, considering that there might be schools for pretty little girls, and so on.37 Copies of utterances of this sentence can get the reading wrong and still be copies and so iterations.38

“Pretty” in “My daughter attends a pretty little girls’ school” can mean “rather” rather than “attractive.” The structural ambiguities are still present, but different. Iterators getting structure or word-meaning wrong are still copying, but their copies may deviate too much to be an iteration.

2. Indexicals and demonstratives

Iterations of non-universal utterings and inscribings cannot have precisely the same truth-conditions as the events they iterate, even if the iteration is by the same speaker or inscriber, as noted above in reference to A Way of a Pilgrim. A person who recites the Scout Oath at one troop meeting utters a sentence with different truth-conditions when he recites it at the next meeting. Any iteration of an utterance has an indexical, its tense, which yields different truth-conditions when uttered at different times. Matching truth-conditions cannot be the criterion for a re-occurrence of an expression to be an iteration of an uttering, since that would make it impossible for anyone, including the original speaker, to iterate an utterance or inscription. Even you cannot utter an expression with the exactly same truth-conditions twice.

Indexicals that make claims beyond inserting different agents, times, or places can yield different utterances, with different iteration-conditions. When Fred says, “I am the best bridge-player in Connecticut,” and I repeat the same words, the repetition is not an iteration. “The best” can be true of only one individual. So, if Fred is the best, I cannot be. Both Fred and I can do our best.39

Different things are said and different iterations are produced when the utterances have demonstratives, rather than indexicals, with different references. When Becky says, “Tom is handsome,” speaking of Tom Snyder, and Eileen says “Tom is handsome,” speaking of Tom Smith, they are saying different things. “Tom” has a demonstrative component.40 Their utterances are word-for-word identical, but even if said in the same space in sequence, neither is a copy of the other. Their utterances are duplicate word sequences, but not copies and not iterations. Neither remark is a duplicate of what the other woman said.

Some re-utterings or re-inscribings are copies, and so are iterations even though the re-uttering has different truth-conditions. My recitation of “On my honor …” is a copy, a recitation of the first part of the Scout Oath. Following or simultaneous with Fred’s utterance, it is a copy and an iteration. But Becky’s and Eileen’s remarks are not. Demonstratives, “that,” “he,” and so forth allow different truth-conditions not depending just on who is speaking or inscribing but on covert referents available only to the speaker or inscriber. Indexicals, such as “I” and the tenses, are in principle overt. A speaker using “I,” “here,” or the present tense fixes referents that may be available to an audience from the circumstances of the inscribing or uttering. My iteration of what Fred said is inserting me for Fred and another present (since I repeat after him) in place of Fred’s indexicals. I have copied and produced an iteration, just as a newborn iterates a grimace when a facial expression is presented.41 Copying and imitation are “doing the same thing,” even though the agents, times, and places are different. Indexical differences yield copies, and so iterations.

Speech actions are actions. When I am demonstrating a yoga position, and you assume the same position, we have done the same thing. The events are different, and the participants are different, but the actions are the same. Indexicals depend for their reference of who, where, and when the speaker is.42 Copies of utterances and inscriptions with indexicals necessarily replace their referents with the producer, place, and time of the copier.43

c. What does it take to be a copy? Copying and naming

Tolhurst and Wheeler’s conception of “copy” was inspired by Kripke’s views about reference.44 Kripke held that the reference of a name depended on there being a chain of the right sort of connection between the referent and the name. Some chains of connection do not qualify. While there may be an historical source for Santa Claus, I was still mistaken until age four about whether Santa Claus is real. Kripke’s argument is compelling. There is reference. Reference does not depend on having any correct views about the entity referred to, no “analysis” of “refers to” in other terms will get the facts right. There are cases, such as whether Balboa was naming a bay or an ocean the “Pacific,” that are absolutely indeterminate. But some of my colleagues in the fire department have heard of Heinrich Biber because I mentioned once that I prefer Heinrich Biber to the Beatles. They may have forgotten where they heard the name. Their term “Biber” still refers to the seventeenth-century composer.45 Names have no descriptive content but refer to their bearers because there is the right connection.46

The notion of “x is an iteration of y” rests on an incomplete, irreducible notion, “is a copy of,” with puzzle cases and no sharp definition of the “copy of” relation.47 As with “name of,” “copy of” and so “is an iteration of” require pairs of word-sequence occurrences to be connected by the right sort of chain. “The right sort of chain” is an incomplete, irreducible notion of what it takes for someone’s use of a name to refer to an entity and what it takes for one inscription to be an iteration of the other.

There are puzzling cases where whether “is a copy of” applies is undecidable. In exactly the same way, Kripke’s account of the name relation resists definition. Puzzle cases are easy to construct, even though very many cases are clear in the absence of a theory.

This is not surprising. A term such as “table” cannot be defined in terms of configurations of molecules. There is no height in millimeters that is the shortest height a tall man can be.48 Referring and being an iteration are similar to being a table and being a tall man. Very few terms outside of mathematics and physics have the sharp criteria of application that would allow division of all possible cases into “yes” or “no.” Did Balboa name an ocean or a bay?

We conclude this section by presenting examples where one inscribing is clearly an iteration of an original but is an iteration intuitively. It is a copy, an iteration, but has a different force or truth-conditions from that intended by the original producer or with a different referent for a demonstrative. The producer’s act has “abandoned” its product in the sense that copies can differ in force and reference. That product can be copied, and the copy is an iteration and can itself be copied.

d. Examples of iterations of inscribings and utterances which have different force and reference

1. Genesis:

In Genesis 3:8–11, “The man and his wife heard the sound of the Lord God walking in the garden at the time of the evening breeze and hid from the Lord God among the trees of the garden. But the Lord God called to the man and said to him, ‘Where are you?’”49 “Walks” implies that God has a physical body and feet. God’s question indicates that God does not know where Adam is.

Genesis 1–2:4 precedes Genesis 3. Genesis 1–2:4 describes a God who creates by fiat. He commands that there be light, creates heaven and Earth, and makes man by fiat, not out of dust. Such a God does not plausibly have a physical body. Such a God would know where Adam was and what had happened. Interpreted as a single continuous inscription, Genesis 1–3 requires that “walking” is figurative and that the interrogative is a “rhetorical” question. God, being non-corporeal and omniscient, could not have walked and must have known where Adam was.

The original inscribing of Genesis 2:5–3, according to modern Biblical scholarship, supposed a God lacking omniscience and having a divine but physical existence. Genesis 2:5–3, as incorporated into Torah, describes God as God had come to be conceived by scribes some centuries after the original inscription. As part of Genesis as a whole, “walk” is figural and God’s question is rhetorical.50

On that understanding, given that both Genesis 1 and Genesis 3 were true, the real meaning of the iterations is not determined by the intentions of the original inscriber of Genesis 3 but by what Torah means. Torah as it now exists seems to have been in place since the time of Ezra, after the Babylonian Captivity. The evolving understandings of scribes between circa 1000 and 500 BCE became the understandings of Torah by Judaism, Christianity, and Islam.

Modern Biblical scholarship has uncovered the force of the original inscriber(s) of Genesis 3. The original force was literal assertion. God’s question was a sincere request for Adam to reveal his whereabouts. That is different from what the iteration, as part of a larger inscription, means. The original inscriber’s inscribing, as an inscription, has been iterated in an inscribing that now means something other than what the inscriber intended.

2. Copies of the New Testament with different reference

Changes of reference in interpreters’ understanding of demonstratives and indexicals can have consequences. In Matthew, becoming a Christian is becoming a Jew who follows Jesus.51 A Christian convert is joining a sect, not changing religions.52 So, in Matthew 27:25–26, the referent of “the people” and “us and our children” in “And with one voice the people cried, ‘His blood be on us and on our children’” is the people present on the occasion.

“Us” is an expression with an indexical component (it is a plurality including the speaker or inscriber) but also has a demonstrative component (the intended other members of the plurality). So, the reference of “us” in Matthew depends on a demonstrative element in Matthew’s inscribing. We can reasonably conjecture the extension of “us” Matthew was reporting. “Us” combines indexical and demonstrative elements—a determinate speaker and a plurality that depends on who the speaker has in mind. When Matthew was written, Christians regarded themselves as of course Jews. Most Christians were Jews. Almost certainly, Matthew took the Jews calling on themselves this curse to be those present. Matthew, Peter, and the other disciples were all Jews, so he could hardly have taken the curse to apply to all Jews.

But in the decades between the writing of the Gospel of Matthew and the Gospel of John, converts to Christianity, thanks to Paul, have come to be largely Greek.53 So, in John, even a Jew converting to Christianity is no longer a Jew. “The Jews” throughout John are the opponents of the Christians. With that conception of “Jew,” and reading the Gospels as a single document, it is possible to read the referent of “the people” and “us and our children” in the Matthew 27 passage as referring to all Jews and their descendants.54 In much of the history of Christianity the inclusive reading of “us” was assigned as the reference. Monks copying the Gospels were copying “us” with the “inclusive” reading. There can be little question that Medieval scribes produced iterations of Matthew’s original inscription.55 There can be little question also that their copies of Matthew, even though word for word identical and intended as copies, differed from Matthew’s intention. A demonstrative had been gotten wrong, under the influence of a very powerful work, the Gospel of John. This re-understanding of the intended plurality of “us” differs little from the re-assignment of force to Genesis 3.

3. Senior English, 1962

The intended referent of a demonstrative: Mrs. Ackley had assigned a Beat poem, one of Ginsberg’s formless kvetchings. She asked my opinion. I said, “I think it was an experiment which failed.” Mrs. Ackley took my “it” to refer to her choice of assignment. I intended my “it” to refer to the poem, not its assignment. But I ended up at the principal’s office, despite protestations. Mrs. Ackley had misunderstood my uttering, but she reasonably understood my utterance. My uttering had become an utterance whose effective interpretation was in the hands of authority.

4. The Declaration of Independence

This document declares that it is self-evident that “all men are created equal, that they are endowed by their Creator with certain unalienable Rights, that among these are Life, Liberty and the pursuit of Happiness.” The original writers and the endorsers can be taken as meaning by “men” “human being” and somehow forgetting about women, slaves, and children or, rather, taking the extension of “man” to be “adult free male.” When the phrase is quoted on the Fourth of July, it means “human being.” The founding fathers, like early writers of Biblical passages, necessarily abandoned their inscription, the product which, for similar reasons to those that motivated understandings of Biblical texts, later had to be true. If “all men are created equal,” as a founding document, has to be true, “men” means “human being.” Whether that was the intent of the signers’ inscription is another question. Whether a current citation of the Declaration of Independence is a duplicate or a copy is an interesting question, to which I have no answer.

5. The Pledge of Allegiance

Every monthly meeting of the Willington Hill Fire Department starts with a group recitation of the Pledge of Allegiance. Each member’s recitation is a copy of an original. The indexical “I” in each recitation refers to a distinct individual. The indexical determined by the present tense refers to the time of the beginning of this month’s meeting, which is different from last month’s meeting. Differences in force are also possible. My re-citation of “under God” and “with liberty and justice for all” may have different force from the recitations by my fellow fifty-year mates. I could have a covert “(I wish)” after “with liberty and justice for all,” altering its force. On other occasions when I feel more patriotic and hopeful, the force of my uttering may be a sincere assertion. I’m saying the Pledge of Allegiance, iterating an original by some chain. My recitation is a copy, even though the forces and referents of indexicals of many phrases of my uttering may differ from those of the composing committee and many of my fellow reciters.

6. The Haggadah and the Nicene Creed

There have been billions of recitations of Deuteronomy 28:5 (“My father was a wandering Aramaean …”) and of the Nicene Creed (“We believe in one God, the Father, the Almighty, maker of heaven and earth, of all that is, seen and unseen …”). The Haggadah and the Nicene Creed are paradigm iterable, recitable documents. The referent of “my” and “we” is different at different Seders and church services. The tense indexical components of “was” and “believe” are always different. Recitations of the Haggadah and the Nicene Creed are iterations of the Creed, despite differences in the referents of indexicals.


1. The title of this essay uses “Davidsonian” rather than “Davidson’s” because some of the conclusions, while consequences of things he wrote, do not correspond to any Davidson inscriptions. His inscribings were abandoned when inscribed.

2. Readers and auditors who iterate an utterance or inscription likewise account for drift of ascription of both force and reference.

3. The relation between an uttering and an iteration of it is not the “type-token” distinction. John Searle, “Reiterating the Differences: A Reply to Derrida,” Glyph 1, no. 2 (1977): 199, thought Derrida’s concept was just the type-token distinction. Derrida was rightly insulted and wrote Limited Inc (Evanston, IL: Northwestern University Press, 1988).

4. William Tolhurst and Samuel Wheeler, “On Textual Individuation,” Philosophical Studies 35 (1979): 187–97.

5. Samuel Wheeler, “Indeterminacy of French Interpretation,” in Truth and Interpretation, ed. Ernie LePore (London: Blackwell, 1989), 477–94.

6. See Derrida, Limited Inc, 69–70, 118–19, for instance. Jacques Derrida, “Différance,” in Margins of Philosophy (Chicago: University of Chicago Press, 1982); and Jacques Derrida, Dissemination (Chicago: University of Chicago Press, 1981) give more detailed arguments. Donald Davidson, “A Nice Derangement of Epitaphs,” in Truth and Interpretation, ed. Ernie LePore (Oxford: Blackwell, 1989): 433–46 characterizes truth-definitions as “passing theories.” His discussion on 442–43 questions whether there is anything “we would normally call a language” (443). He concludes that “there is no such thing as a language, not if a language is anything like what many philosophers and linguists have supposed” (446).

7. Ludwig Wittgenstein’s Tractatus Logico-Philosophicus (London: Routledge and Kegan Paul, 1961), 4.441, makes much the same point. “Frege’s ‘Judgement stroke’ ‘|-’ is logically [truth-conditionally] quite meaningless … it simply indicates that these authors hold the propositions marked with this sign to be true. Thus, ‘|-’ is no more a component of the proposition than is, for instance, the proposition’s number.” Wittgenstein thus recognizes what Donald Davidson in “Moods and Performances,” in Inquiries into Truth and Interpretation (Oxford: Oxford University Press, 1984): 109–22, calls “… the autonomy of linguistic meaning. Once a feature of language has been given conventional expression, it can be used to serve many linguistic ends” (113).

8. For instance, in his 1967 undergraduate course in Philosophy of Language. See Samuel Wheeler, “Remembering Donald Davidson: His 1967 Undergraduate Philosophy of Language Course,” in Donald Davidson: Life and Words, ed. Maria Baghramian (Oxford: Routledge, 2012): 65–70.

9. Donald Davidson, “Truth and Meaning,” in Inquiries into Truth and Interpretation (Oxford: Oxford University Press, 1984): 17–36; Alfred Tarski, “The Concept of Truth in Formalized Languages,” in Logic, Semantics, Metamathematics (Oxford: Oxford University Press, 1956).

10. In Donald Davidson, “Radical Interpretation,” in Inquiries into Truth and Interpretation (Oxford: Oxford University Press, 1984), Davidson calls his account “radical interpretation.” Radical interpretation is an account of what another person’s uttering or inscribings mean, which systematically generates expressions of the interpreter’s language with the same meaning. A mentioned sentence of the interpreted language is connected with a use of a sentence in the interpreter’s language. A theorem would be “‘Jean est vert’ would be true if and only if John were green.” Chapter 2 of Willard Van Orman Quine’s Word and Object (Cambridge, MA: MIT Press, 1960) developed the notion of “radical translation.” Davidson’s account of interpretation thus differs from Quine’s radical translation, which correlates “mentioned sentences.” Unless one knows Greek or Hebrew, being told that the Hebrew “ehyeh asher ehyeh” means the same as “ego eimi ho on,” as the Septuagint conjectures is not much help. A theory of meaning must use radical interpretation, not radical translation. The point of “radical” in both radical interpretation and radical translation is that there are no cognates or other linguistic clues to assist the interpreter.

11. In a footnote to Davidson’s “Truth and Meaning,” in its re-printed version: “This paragraph is confused. What it should say is that sentences of the theory are empirical generalizations about speakers, and so must not only be true but also lawlike. (S) [‘“snow is white” is true if and only if grass is green’] presumably is not a law, since it does not support counterfactuals. It’s also important that the evidence for accepting the (speaker and time relativized) truth-conditions for ‘That is snow’ is based on the causal connection between a speaker’s assent to the sentence and the demonstrative presence of snow.” So, a truth-definition must support counterfactuals about un-uttered sentences. Thus, a truth-definition entails the truth-conditions merely possible utterances would have had. We find it most accurate to phrase truth-definition clauses in the subjunctive to capture this law-like and empirical character. Donald Davidson, Inquiries into Truth and Interpretation (Oxford: Oxford University Press, 1984), 26n11.

12. The account must be compositional because no phrase-book pairing speaker sentences with interpreter sentences could be complete. Speakers are finite, but the sentences they can produce are in principle infinite. Hence a truth-definition for a language at a time must be compositional, an algorithm.

13. Much of the time, we know what is said and why. I am running a truck at a structure fire, and the guys on the hand line on the inside say, “More pressure.” In principle, their short message could mean “More pressure? What are you doing?” perhaps because another line has shut down. But they probably want the pressure in their line cranked up.

14. Donald Davidson, “The Second Person,” Midwest Studies in Philosophy 17, no. 1 (1992): 255–67. Davidson’s focus on language learning as requiring a common world would not have been rejected by Derrida, who told me that he was amused by people who thought he did not believe there were really rabbits. We share a world with terrorists, squirrels, and Cro-Magnons. If we did not share a common world, we would not know that we disagree. Disagreement requires a ground of agreement relative to which we can differ. But the “ground” need not have the sort of detail that the traditional account of natural language meanings as expressions of something trans-linguistic requires. Squirrels and I have a common world containing acorns.

15. “Training” is by and large unintentional. The child is present when people are talking and picks up the language. Actual recordings of what a child hears show that a very fragmentary database suffices for a four-year old to become a native speaker.

16. “Speech action” need not be “speech act” in the sense of J. L. Austin’s How to Do Things with Words (Oxford: Oxford University Press, 1962). A speech action need not be an utterance or inscription which brings about a state of affairs. Davidson thought that “I declare you man and wife” had a truth-value, even though the speaker is not presenting the expression to the audience for that purpose. Likewise, “I promise to pay you” may be true or may be being used as an example in an academic paper. In neither case is it produced in order to assert that a promising has occurred. See Mitchell Green, “Speech Acts,” Stanford Encyclopedia of Philosophy, ed. Edward N. Zalta (Stanford, CA: Metaphysics Research Lab, Stanford University, 2021), https://plato.stanford.edu/entries/speech-acts/. Davidson and Derrida both appreciated but argued against Austin’s idea that force could be encoded in linguistic forms; see Derrida, “Différance.” Wheeler, “Remembering Donald Davidson,” reports several sessions of Davidson’s 1967 class on J.L. Austin.

17. There are roughly two million people named “Tom” in the United States. Names have a demonstrative component.

18. Gestures and context assist the interpreter, just as context, facial expression, and knowledge of the speaker assist in assigning force. But interpreters can get a reference, and so the truth-conditions, wrong, just as they can get force wrong.

19. Donald Davidson, “Mental Events” in Essays on Actions and Events (Oxford: Oxford University Press, 1980), 207–27, distinguishes the mental (intentional) from the physical (causal) as two families of concepts, not distinct phenomena. Davidson, Truth and Interpretation cites our comprehension of Goodman Ace and Mrs. Malaprop as illustrations of how we routinely interpret by ascribing reasonable beliefs to speakers and writers. My one-time fire chief pronounced “format” as “floormat.” I understood by interpreting him as an agent. Everyone understands other dialects (collections of similar idiolects).

20. With perhaps further levels of intention—sarcasm, joking, humiliating the audience, and so on, as Austin elaborates.

21. Earlier utterings may be less reliable, given the drift we describe below.

22. Derrida, “Différance.”

23. Someone first used “crush” as a term for a thorough defeat. Someone first used “screw” for a sexual act. Someone else first used “screw” as a term for ill-treatment. Copiers caught on and metaphors became related senses of an expression.

24. The interpreter’s own language at a time is itself subject to the same drift. Our understanding of ourselves, as a truth-definition, is sometimes indeterminate. My utterance “The Patriots crushed the Cowboys” has two equally good accounts. It could be a metaphor, or there could be two related words in my idiolect. I cannot tell “from the inside.” Part of language-change over time is extended uses and metaphors becoming different senses of words. “Berth” used to be the space between passing ships. By an iterated series of metaphors and extended applications, it came to apply to sleeping-car accommodations. The earlier sense, now extinct, is fossilized in the idiom “give wide berth to.” See Samuel Wheeler, “Derrida and Davidson on Metaphor,” in Beyond the Analytic/Continental Divide, ed. Jeffrey A. Bell, Andrew Cutrofello, and Paul M. Livingston (London: Routledge, 2016), 172–90.

25. Davidson, Truth and Interpretation, 442.

26. This paragraph owes much to Mitchell Green’s remark on an earlier draft.

27. The millions of repetitions of “God have mercy on me, a miserable sinner” by the protagonist of The Way of a Pilgrim (Boulder, CO: Shambhala, 2019) differ in the referent of the tense of “have.” Uses of the historical present—“A priest, a rabbi, and a minister go into a bar. The priest says …”—may likewise lack tense that changes truth-conditions. Questions such as “What do Winnie the Pooh and Alexander the Great have in common?” as a joke (same middle name) and “What do Julius Caesar and Alexander the Great have in common?” as an essay question likewise may lack tense.

28. So, if Fred in Atlanta says, “It’s dinner time,” and I say, “It’s dinner time,” those utterances are word-for-word duplicates, but whichever is later is not likely to be an iteration of the other. In earlier versions of this essay, we took the issue to be whether a duplicate was “the same text” or not. But there are too many acceptable notions of “text” for this question to be of interest. The issue Derrida was concerned with was the possibility that iterations would be instances of what the speaker said, even though differing in force and in the referents of some expressions.

29. Classical scholarship since the Renaissance has constructed “family trees” of errors that are copied and give rise to further copies with further errors.

30. How exactly this “internal pointing” takes place is a topic in psychology. See Zenon Pylyshyn, “The role of location indexes in spatial perception: A sketch of the FINST spatial-index model,” Cognition 32 (1989): 65–97. Pylyshyn’s research program is focused on perceptual “pointing” but suggests generalization.

31. Utterances are sound-sequences, and that aspect may be copied. For decades, until my wife straightened me out, I thought Creedence Clearwater Revival had a famous song including the lyrics, “There’s a bathroom on the right.” My version was a very poor copy of the song as a sound sequence, but not a copy of the meaningful utterance at all. Likewise, Walt Kelley, Deck us all with Boston Charlie (New York: Simon and Schuster, 1963).

32. Copies may or may not be intentional. Tolhurst and Wheeler give examples such as accidental carbon copies. Tolhurst and Wheeler were concerned with “textual identity.” “Same text,” though, can mean different things. In a Google search, you type in a text and get inscriptions which are the same text, but duplicates, not iterations, of what you typed. We meant “iteration” by “same text.”

33. At some level of the event, robot emails and phones calls have an intender as source.

34. Steven Knapp and Walter Benn Michaels, “Against Theory,” Critical Inquiry 8, no. 4 (1982): 723–42.

35. “If A, then probably B” has the same feature: that one step is pretty reliable, but the generalization, which holds with “equals,” can lead from truth to falsehood.

36. In ancient times, there was a party game “telephone” where children sat in a circle, one of them concocted a sentence, and whispered it to their neighbor. The utterance that came around to the originator was very different from the original.

37. Such syntactic ambiguities, sentences with a given surface form but distinct underlying forms, have been part of philosophical semantics at least since Bertrand Russell’s “On Denoting,” Mind 14 (1905): 479–493. Russell’s example, “The present king of France is not bald,” is false if it attributes not being bald to the present king of France, but true if read as “It is not the case that there is a present king of France.”

38. We can distinguish “readings” by positing covert parentheses in semantic form. So “pretty little girls’ school” can be “pretty (little (girls’(school),” a pretty little school for girls; “(pretty (little girls) school),” a pretty school for little girls; or “(pretty(little(girls))) school,” a school for pretty little girls.

39. Sincere thanks to Joshua Kates, whose observation (by email) showed that differences in indexical referents do not always yield iterations. Indexicals are not special, but just different from demonstratives. My resulting theory has, thanks to his remark, come to be that all differences of a re-occurrence of an uttering or inscribing can yield an uttering that is not an iteration or a copy.

40. See Tyler Burge, “Reference and Proper Names,” Journal of Philosophy 70, no. 14 (1973). We discuss demonstratives below. That names in natural language are not unique labels for individuals is quite obvious.

41. “Copy” and “imitation” are essential to acquisition of a language. In Davidson’s “The Second Person,” a child who calls a chair “chair” by imitating what the adult has said in a chair environment is producing a copy of the adult’s utterance. A child is an innately-gifted syntax guesser, and imitator, as innumerable studies have shown, with a surprising small corpus of utterings to copy. A child is a competent language-user by age four. See Noam Chomsky, Syntactic Structures (The Hague: Mouton & Co., 1957); and Noam Chomsky, Aspects of the Theory of Syntax (Cambridge, MA: MIT Press, 1965).

42. Some indexicals are impure. An “impure” indexical is one whose reference depends partly on the speaker and partly on demonstrative components. Plurals such as “we” can refer to innumerable pluralities including the speaker. “We always send thank-you notes when we receive gifts” assigns an extension to “we.” A difference with the accusative “us” is discussed below.

43. See John Perry, “The Problem of the Essential Indexical,” Nous 13, no. 1 (1979): 13–21.

44. Saul Kripke, Naming and Necessity (Cambridge, MA: Harvard University Press, 1980). Wheeler was present at Kripke’s lectures at the Woodrow Wilson center in 1970. The lectures were famous before publication and a transcript had been published before 1980.

45. Kripke’s examples show that no analysis of the reference of a speaker’s use of a name in terms of the speaker having a description are plausible. Someone whose only opinion about Einstein is that he invented the atomic bomb can refer to Albert Einstein. He is not attributing that invention to the great musicologist Alfred Einstein. Naming depends on a chain of connection to a referent, not on having a true description.

46. Another example: Fred first heard of Thomas Browne when I told him I sold a four-volume set of Browne’s works. That Thomas Browne, somewhat younger than the Thomas Browne of Pseudodoxia Epidemica, wrote light, humorous pieces. So, Fred has heard of Thomas Browne. When Celeste then told Fred that Thomas Browne wrote Religio Medici, Fred acquired a false belief about the Thomas Browne I had referred to. Fred’s “Thomas Browne wrote Religio Medici” is false. It is a duplicate, but not a copy, of Celeste’s true utterance. It differs too much to be an iteration of Celeste’s remark.

47. With imperfect copies (as all iterations are), transitivity breaks down. At the end of a chain of iterations of iterations, an expression is not an iteration of the original. There is no non-arbitrary line between “W is an iteration of A” and its negation. This illustrates the vagueness of the “is a copy” relation, a vagueness shared by all relations outside of mathematics and perhaps some other hard sciences.

48. See Samuel Wheeler, “Reference and Vagueness,” Synthese 30 (1975): 367–79. The topic of vague predicates has since become a vast literature in philosophical logic.

49. Translation from The New English Bible (Oxford: Oxford University Press, 1972).

50. See James Kugel, The God of Old (New York: The Free Press, 2003); and James Kugel, How to Read the Bible (New York: The Free Press, 2007); and many other of his works for a detailed account of this transformation and of numerous similar ones. Kugel describes the changed understanding of this inscription. Roughly, many documents, including Genesis 1–3, came to be regarded as divinely inspired. This change in perception took place over several hundred years after Genesis 3. Scribes and redactors combined sacred documents into what became Torah by the time of Ezra.

51. The same is true in Luke and Acts. In Acts, Peter goes to the temple after the resurrection every day.

52. Among many others, before and after. See Josephus, The Jewish War (New York: Dorset Press, 1986); and Acts 5:34–40, where Gamaliel recounts other Messianic cults.

53. We are here simplifying a complicated history.

54. Curses applied to whole lineages have famous precedents. Noah curses Ham and all Ham’s descendants for an obscure sin in Genesis 9:22–26.

55. Here are some Tanakh examples of ambiguous reference of a demonstrative. In Exodus 4:24, God tries to kill Moses. Zipporah circumcises their son and “touched him with it,” thus saving Moses’s life. Whether she touched Moses or God with the foreskin is not clear in the narrative. The god who tries to kill Moses but is foiled by a woman’s knowledge of magic cannot be the god of Genesis 1. The original speaker or inscriber meant one or the other. This inscription that has been copied and remains part of Torah despite its theological issues. The remarkable Exodus 3:14 “ehyeh asher ehyeh” is translated as “I am, that is who I am” with a note that “or ‘I will be what I will be.’” What does it mean in my truth-definition into my idiolect of English? Hebrew has (of course) a non-Indo-European tense system. The passage strictly would have to be translated into English as a disjunction, which cannot be what the original inscriber meant.