Anscombe’s First Person


Elizabeth Anscombe’s notorious claim in The First Person, that “I” is not a referential term, has suffered an unfair history of discredit. Although, I will ultimately conclude that Anscombe’s position is untenable when argued to apply for all uses of “I”, to deny the irreferentiality of “I” in many common uses is equally wrong-minded. The assumption which undermines both Anscombe’s argument and criticisms thereof is that “I” must always be either referential or not. While this claim seems to be intuitively true, our clinging to the fixity of “I” is purely a result of a fear that to sacrifice the fixity of “I” would be to sacrifice the fixity of self. I will show in what follows that we latently accept an “I” that is, at times, referential, and, at others, not. In doing so, we shall salvage Anscombe’s argument in part.

First, we must dispel a possible misunderstanding of Anscombe’s argument: that its putative failure arises from an unclear notion of “I-reference.” Perhaps, objectors to Anscombe’s argument merely afford “I-reference” a greater latitude of meaning than Anscombe, and this is to account for the fact that what Anscombe finds to be true runs absolutely counter to the common-sense conceptions of the matters. However, Anscombe defines, quite clearly, what she has in mind by saying that “I” is irreferential.

Reference simply is the indicating of some object by some word or expression. The general logical formulation of “reference”-a position Anscombe is not satisfied with-is that for a word or expression to refer to something, it must be exchangeable salva veritate with another name for that thing, when that thing is the subject of some assertion. To this definition of “reference” Anscombe wishes to add that referring terms are in some way intentioned to their objects, although the reference may be incorrect. When she says that the “I” does not refer, she is denying that “I” is a true subject of a sentence in that it does not signal something in the world. There is nothing which “I” “gets hold of” and, furthermore, the “I” is not intended to get a hold of any such object.

Anscombe begins her argument against “I” as a referring term by asking to what “I” would refer were it to indeed refer. She eliminates any other possibilities other than the Cartesian Ego by illustrating with her “man in a tank” example that even if we were completely unaware of our bodies we could still use “I.” If we were blinded and anesthetized in a tepid tank, we would presumably be unaware of our bodies but we would still have “I” thoughts such as “I will never let myself get in this situation again.” Therefore, our bodies cannot be the referent of “I”. The only option left for which to the “I” to refer is some immaterial soul or mind or ego.

At this point, Anscombe recognizes her argument as something of reductio ad absurdum in that if “I” does refer, Descartes must have been right, a position she sees as impossible. If “I” always refers to one’s Cartesian Ego, there is required some identification of the referent across “I” thoughts. This consistency of self-identification is an improper one that requires the positing of some continuous “self” which only leads to confusion. Take the amnesiac. Is his “self” then changed by his loss of memory because it presents itself to him as a discontinuity? The amnesiac uses “I” fine even though his earlier identity is not accessible to him. Many such difficulties arise when we try to posit identification as a continuous self required by the Cartesian Ego. Such an idea leads to a dispute that is, “…self-perpetuating, endless, irresolvable…” (Anscombe 58).

Furthermore, Anscombe notes that “I” might even refer to more than one subject, were it to refer at all. There is certainly a possibility that “I” could have collective reference, a position which Anscombe reinforces by reference to religious life.

From these two objections with “I” as reference, Anscombe concludes that “I” is not a referring term. As she has just demonstrated, “I” does not always “get hold” of the right object because either, in her Cartesian Ego objection, the “self” is not properly an object, or, in her several referent objection, the “I” may refer to sometimes one thing, sometimes more. However, the “I” can also never get hold of the wrong object because the thought or use of “I” demands that there be something which speaks or uses “I”. Even though cases are imaginable where what one says of themselves with “I”, this only shows that one can be wrong about what they self-attribute and not about the (alleged) reference of I. Therefore, the only resolution is to accept that “I” does not get hold of any object whatsoever; “I” is not referential.

This conclusion, as often argued, is patently incorrect. Anscombe’s argument is largely compromised by her insistence that when someone makes an I-statement they must possess and assert a full sense of self. Although her tank argument is convincing, in that I-statements can refer to the speaker in a way that does not require bodily reference, it does not entail that some self must be posited and grasped by the speaker. Gareth Evans writes:

“…it does not follow that in order to elucidate the intention of satisfying ‘x refers to x’, we need a grasp of the self-conscious Idea-type that we have of ourselves. Indeed, it seems plausible that the explanatory direction goes the other way: the fully self-conscious use of ‘I’ can be partly explained, precisely, as a use in which the subject knowingly and intentionally refers to himself (satisfies lx (x refers to x))” (Evans 258-9).

While Anscombe’s argument seems convincing for statements such as “I am John Smith”, statement such as “I am six feet tall” do little to uphold Anscombe. Truthfully, though, even the former sentence may avoid any apprehending of the self. Imagine the actor who is playing John Smith or someone who has assumed the anonymous identity of “John Smith.” In such cases, “I am Jon Smith” may be as simple as “My name is John Smith” and does not require Anscombe’s grasping hold on the “self.” Clearly, Anscombe is incorrect in resolving that if ‘I’ refers it must refer to a Cartesian Ego, her reductio conclusion. [Could be more clearly worded. Something like “Clearly, Anscombe is wrong to conclude that if ‘I’ refers it must refer to a Cartesian Ego.” MH]

However, it is from this failure in Anscombe’s argument that many conclude the opposite of Anscombe, that ‘I’ is referential. To do so is to make the unfortunate mistake of throwing the baby out with the bath water and neglecting the value of Anscombe argument, namely her demonstration of how I-statements could be meaningful and understood without being referential. What is necessary to reconcile Anscombe with her objectors is to adopt a model of I-statements in which “I” may be referential in some cases and non-referential in others.

If the notion that “I” may be sometimes referential and sometimes not seems to run against linguistic sensibility, this is only because one has not separated words themselves from their uses. Strawson writes, “We are apt to fancy we are talking about sentences and expressions when we are talking about the uses of sentences and expressions…Meaning…is a function of the sentence or expression; mentioning and referring and truth and falsity, are functions of the use of the sentence or expression” (Strawson 7). With this model we will demonstrate that the problems we have with “I” result from conflating the word with its use. It should prove uncontroversial that a word (or in Strawson’s terminology, an expression) can be used to sometimes refers, sometimes not.

Consider, for reference, the word “it”. As Anscombe writes in a parenthetical, “…no one thinks that “it is raining” contains a referring expression, “it”…” (Anscombe 55). [Interestingly, Shakespeare did: “The rain it raineth every day”] Clearly, there is a use of “it” and, indeed, many uses of “it” in which the expression “it” does not refer. “It is cold today”, “It is five o’clock”, “It is ten miles to the store”. All such “impersonal expressions” as they are commonly called, feature an “it” that is not used to refer to anything. “It” is merely a grammatical construction to indicate a state of affairs in a certain way, much like “I” does in Anscombe’s proposed model of self-attribution.

However, in other cases, the far more common uses of “it”, the expression is certainly used referentially. “What color is the ball?” “It is blue.” In this exchange, “it” is used to clearly refer to the ball. We naturally countenance such a dual use of “it”, why, then, should “I” not enjoy a similar double use character?

Most of the cases presented when “I” is discussed feature some abstract existentiality such as “I am John Smith.” From these statements, all kinds of elaborate thought experiments are concocted in which the speaker is confused about his identity, unaware what his name is, joking about his name, etc. This led Anscombe to posit that were “I” to be referential it would have to indicate some “self” which is untenable when pushed to these extremes, these odd cases of language. This is an apt reaction and indeed these “John Smith” examples do seem to run counter to an “I” which refers to a Cartesian Ego. There seems to be no problem in denying that “I” is referential in these cases. Anscombe’s assertion that these sentences have no subject and are merely construction of self-attribution seems correct.

However, from this conclusion, we cannot infer that “I” must always be irreferential. The preceding cases were merely instances of “I” used without reference. Consider the sentence “I am six feet tall.” No “self” needs to be posited to assert the height of the subject. “I am six feet tall” seems to be no different from “The body of the speaker of this sentence is six feet tall” or (to use Anscombe A-users example) from “A is six feet tall”. In fact, were it possible for a body to speak this sentence without having a mind, we would not find it strange. If a machine reported “I am six feet tall,” we would likely happily countenance the usage. However, a machine saying “I am John” would be met with greater apprehension and would likely be translated in the listener’s receptivity into “My name is John” or “I have been called John.” Thus, Anscombe’s reductio falls apart.

This is simply because “I” is being used in a different way. We must not assume that referentiality is stored within a word. Rather it is a product of referential use. If we adopt such a model of “I” we will quickly see the problems with its grammar vanish. Like most problems that arise when we prod our language past the conventional usage which does not trouble us, the problem of “I” proves to be a specter born out of muddied conception of word and use.

Works Cited

Anscombe, G.E. M. “The First Person.” Mind and Language. Ed. Samuel Guttenplan. Oxford:

Clarion P, 1975.

Evans, Gareth. The Varieties of Reference. Ed. John McDowell. New York: Oxford UP, 1982.

Strawson, P. F. Logico-Linguistic Papers. Grand Rapids: Ashgate, Limited, 2004.

Erik Hinton (’10) is a Philosophy & Film Studies double major at University of Pittsburgh.



Cover art courtesy of winstoncamille.

Leave a Reply

Your email address will not be published. Required fields are marked *