SORITES, ISSN 1135-1349

Issue #02. July 1995. Pp. 57-75.

Textual Identity

Copyright © by SORITES and Jorge J. E. Gracia

Textual Identity

Jorge J. E. Gracia

There are several volumes in the University library with the title Don Quixote, whose author is identified as Cervantes. These volumes have important differences among them. They occupy different spatio-temporal locations, the typescript they use is different, the number of words per page they have varies, the paper on which the words are printed has different consistency, and so on. In spite of these differences, however, most of us regard these volumes as copies of the same text, and, indeed, most users of the library have no trouble in identifying them as such.Foot note 99 So we may ask: What is it that makes them the same. And more generally, we may pose the question: What makes texts the same. These questions raise the issue of what constitutes textual identity.

Apart from the importance and interest that the question of textual identity may have for other issues involved in textuality, it is itself puzzling because the basic approaches frequently adopted to answer it pose difficulties. Five of these views suggest themselves.Foot note 100

-- One identifies a text with the entities that are used to convey meaning considered separately from that or any other meaning, namely, the entites which constitute a text, or ECTs for short. In a text composed of marks made on a piece of paper, it is the conditions of identity of the marks, apart from the meaning they are supposed to convey, that are considered to be the conditons of the identity of the text. This view encounters difficulties, however. Were this view to be accepted, for example, we would lack a way of distinguishing texts from entities which are not texts. Moreover, this view would allow a text to have contradictory meanings, since the same entities can be used to convey such meanings.Foot note 101

-- The second view identifies a text with the meaning considered independently of the entities (ECTs) used to convey it. In this way, the conditions of identity of a text apply to the meaning only and not to the entities used to convey such meaning. It is not the conditions of identity of the marks made on the paper that determine the identity of a written text, for example, but rather the conditions wich determine the identity of what it signifies.Foot note 102 The difficulty with this position is that different texts, including texts in different languages that have the same meaning but different ECTs, would have to be considered the same text if this view were correct.

-- The third possible view holds that texts are to be identified with the entities that are used to convey meaning (ECTs) considered together with meaning, when meaning is taken in general and not identified with a particular meaning. In this sense, the conditions of identity include meaning but not any meaning in particular. The conditions of identity of a written text represented by marks on a paper include the conditions of identity of the marks plus a condition that the marks have meaning, but not that the marks have any meaning in particular.Foot note 103 The difficulty with this view is similar to the difficulty mentioned in connection with the first view presented, for in accordance with it the same text could have any meaning whatever, and that does not seem acceptable.

-- The fourth view identifies texts with certain acts.Foot note 104 This view is derivative of Austin's well-known conception of language in terms of speech acts.Foot note 105 A text, then, would be a series of acts someone performs. Since Austin distinguished among three different kinds of pertinent speech acts, the question arises as to which of these constitutes a text. The locutionary act, for Austin, is the act of uttering that takes place when someone says, for example, `Pick up the ball, please.' It is the act performed when one utters the sounds which constitute an oral text. (This could be applied as well to writing, of course.) The perlocutionary act is the act of getting whoever is asked to pick up the ball to do so. It is the act performed when the loctutionary act produces the desired effect. And the illocutionary act in this case is the act of asking someone to pick up the ball. It is the act performed when one says something, that is, when one performs a locutionary act. Within this framework one could identify the text `Pick up the ball, please' as a set of locutionary, perlocutionary, or illocutionary acts or as a set composed of all or some of these acts. In any case, the imporant point is that a text becomes a set of acts performed by a speaker or writer. One of the problems with this positon is that it leaves no place for meaning to play a role in textual identity. Moreover, it confuses the use (i.e., an act) of a text with the text, just as it confuses the act of uttering with the utterance. Yet it is not the act of uttering, but the utterance -- just as it is not the act of writing, but the writing, with which one communicates meaning. So the text cannot be the act of uttering or writing, even if one were to add to these perlocutionary and illocutionary acts. The text must be the utterance or the writing.Foot note 106

Finally, there is the view I shall defend, according to which the conditions of identity of a text include not only the conditions of identity of the entities (ECTs) used to convey its meaning, but also the conditions of identity of the particular meaning they are used to convey. But this position is not entirely without difficulty. In the first place, this view seems to preclude the possibility that a text may have different meanings depending on its context and how it is used. And, in the second place, it also seems to preclude the possibility that different audiences understand the same text differently.

In spite of the importance of the issues involved in and related to the identity of texts, and the considerable attention that texts are receiving in recent literature, the question of textual identity is seldom explicitly raised by philosophers.Foot note 107 Textual critics by contrast are much concerned with this issue. But their concern relates more to the question of the identity conditions of particular texts, rather than of texts in general.Foot note 108

Before I begin the discussion proper, I would like to make an important qualification. In this paper I intend to discuss the question of identity from an ontological rather than an epistemological perspective. My issue is with the identity of texts, not with the discernment of that identity. I am aware that the epistemic question of identity is as important as the ontological one and that some regard it as necessarily propaedeutic to the latter. Nevertheless, I shall omit consideration of it in the present context, leaving its determination for another time.

1. Sameness

The notion of `sameness' is one of the most versatile in our ordinary conceptual framework. We apply it to all sorts of things, such as colors, persons, times, spaces, relations, essences, experiences, events, concepts, and so on. We speak of persons and/or their lives as being the same or of the same type: we say that a daughter is the same as her father with respect to this or that characteristic; we refer to the use of the same concepts in discourse; we agree that sometimes we have the same experiences; and we talk about being in places at the same time, being essentially the same, and witnessing the same events. Indeed, an endless number of examples could be given to illustrate the usefulness and pervasiveness of this notion in ordinary discourse, but for our purposes the examples provided should be sufficient.

The notion of sameness has obvious relationships to the notion of similarity. Indeed, it is not unusual to find that authors use the terms `same' and `similar' interchangeably. Yet there are important distinctions between the two notions. Perhaps the key distinction is that similarity occurs always in the context of difference. That is, in order for two things to be similar, they must also be different, although the difference in question must refer to aspects other than those on which the similarity is based. Thus one may speak of two persons as being similar provided that they differ in some way. If they do not differ in any way, then they are regarded as the same. The conditions of similarity may be expressed in the following way:

X is similar to Y if and only if X and Y (1) have at least one feature F that is the same in both and (2) also have at least one feature F1 that is not the same in both.

Sameness, on the other hand, does not require, indeed it precludes, difference. That does not mean that two things could not be regarded as the same with respect to some feature or other and different with respect to something else. A daughter, for example, may be the same as her father with respect to stubborness while being different, as is obvious, with respect to gender. The point is, however, that in order for the daughter and father to be the same with respect to stubborness, their stubbornesses must not involve any difference whatsoever. If there were some difference, say that their stubbornesses were not exactly the same in every respect, one would speak instead of a `similarity of stubborness.' We might express this understanding of sameness of things and sameness of their features in the following two propositions:

X is the same as Y if and only if there is nothing that pertains to X that does not pertain to Y and vice versa.

X is the same as Y with respect to F if and only if there is nothing that pertains to F of X that does not pertain to F of Y, and vice versa.

Part of the reason for the frequent blurring of the distinction between sameness and similarity is that the term that is often used as the opposite of both is `difference,' even though there exists a term which is used more properly to express the opposite of similarity, namely, `dissimilarity.' Similar/different and same/different are generally regarded as pairs of opposites. This usage has not always been prevalent, however. In the Middle Ages, for example, there was a concerted effort to keep the notions of similarity and sameness separate, and this was supported by the use of different opposite terms for them. `Difference' (differentia was used, at least in technical philosophical discourse, as the opposite of `similarity' (similaritas), while `diversity' (diversitas was used as the opposite of `sameness' (identitas).

Apart from `similarity,' there are also other terms that are sometimes exchanged with `sameness' in both ordinary and philosophical discourse. Perhaps the most commonly used ones are `identity' and `continuity.' There is very little difference in ordinary discourse between the notions of identity and sameness. `Identity' is a learned term derived from the late Latin identitas (in turn a derivative of idem, which means «the same»), while `sameness' comes from an Old Norse common root. In technical discourse there can be differences in the usage of these terms, but since those are idiosyncratic to particular authors, they are irrelevant to our present purposes. With respect to continuity, things are otherwise, however. The notion of continuity carries the implication of noninterruption either spatially or temporally, while the notion of sameness, as we shall see, is much broader. Continuity turns out to be interchangeable with only one type of sameness.

Not all sameness about which we speak is of the same sort. There are at least three fundamental but distinct types of sameness, which I shall respectively call achronic, synchronic, and diachronic. Achronic sameness is sameness irrespective of time -- it may be understood as follows:

X is achronically the same as Y if and only if X is the same as Y.

By contrast, synchronic sameness and diachronic sameness have to do with time. The first may be taken thus:

X is synchronically the same as Y if and only if X is the same as Y at time t.

Diachronic sameness may be understood in the following way:

X is diachronically the same as Y if and only if X is the same as Y at times tn and tn+1.

The distinctions between achronic, synchronic, and diachronic sameness, then, have to do with time. In the first case, sameness has no reference to time at all; in the second case, sameness applies to a specified particular time; and in the third, it applies to two (or more) different times. These three sorts of sameness generate three different problems for those who wish to account for them. In the case of achronic sameness what is osught is to determine the necessary and sufficient conditions that make a thing to be the same irrespective of time. This is another way of asking for what makes a thing to be what it is and the answer involves identifying its necessary and sufficient conditions. For this reason, I like to call this issue the problem of identity. Because of the atemporal character of the inquiry involved in solving this problem, such an inquiry may concern atemporal entities, such as universals, in addition to temporal ones. Indeed, this sort of investigation can be applied to anything which may become the subject of philosophical discourse. We may ask about the necessary and sufficient conditions not only of an individual person, but also of universals, concepts, propositions, events, and so on.

The case of synchronic sameness is different from that of achronic sameness insofar as what is sought for in this case is an account of the necessary and sufficient conditions that make a thing to be the entity it is at a particular time. This difference is significant because it restricts the relevant types of entities to temporal ones. It would make no sense to ask for an account of the sameness of atemporal entities at a particular time. Thus, for example, questions concerning synchronic sameness could not apply to universals, mathematical entities, or even to God if God is conceived as being outside of time, as Augustine thought. Apart from this significant difference, achronic and synchronic sameness are similar because their analyses abstract from the passage of time; this abstraction is what distinguishes them both from diachronic sameness.

In diachronic sameness what is at stake is the determination of the necessary and sufficient conditions that make a thing to be the same at two (or more) different times. Indeed, it is usual for philosophers to speak of this as the problem of accounting for «identity through time» or that of «temporal continuity.Foot note 109 From this it should be clear that diachronic sameness may apply only to those things to which temporal passage applies. It would make no sense to talk about the diachronic sameness of instantaneous entities, that is, of entities that exist only at an instance of time, or of atemporal entities such as universals, mathematical entities, and God.

2. Conditions of the Achronic Sameness of Texts

In the previous section we have seen that there are three different kinds of sameness. For our present purposes, however, we will be concerned only with achronic sameness. The problem of achronic sameness has to do with the identification of the necessary and sufficient conditions that make entities the same apart from any consideration of time. The question involved in the case of texts is the identification of the necessary and sufficient conditions that make a text the text that it is. Note, however, that we are not dealing here with token texts; we are not concerned with the identity of the various copies of, say, the text of Cervantes's Don Quixote or the American Declaration of Independence. That is, we are not concerned with determining the conditions that make a copy of a text the individual copy it is. Rather, our concern is with whatever makes different individual copies be the same text. The question of the individual identity of texts is both interesting and important, but it is not the one that shall be discussed here.

In order to bring out the problem of the achronic sameness of texts more clearly, let us consider the following examples of texts:Foot note 110

1. 2 + 2 = 4

2. 2 + 2 = 4

3. Two and two make four.

4. Two plus two add up to four.

5. Dos y dos son cuatro.

6. Dos ma's dos son cuatro.


8. Four two and two make.

9. 3 + 3 = 6

Our ordinary intuitions would seem to dictate that we consider (1) and (2) as the same text, and likewise with (3) and (7). Indeed, when we speak about the text of Thomas Aquinas' Summa theologiae we make no distinction between the manuscript copies of it written in different medieval hands and those printed on a page. Nor is the color of the ink used relevant or of the paper or parchment, the size of the letters, or even whether those letters are all capital or not. We are also quite certain on the basis of our ordinary intuitions that (9) is not the same as any of the other members in the group, and the reason given would be most likely that it means something different than the others. Moreover, most people, I believe, would not regard (5) and (6) as the same texts as (1), (2), (3), (4), (7), (8) or (9). They would argue that (5) is a translation of (3) and (6) of (4) into a different language. The matter of whether (3) and (4) or (5) and (6) are different texts or not would probably elicit some disagreement. Some would argue that since they mean the same thing, are written in the same language and contain the same key words (`two' and `four' in the English text; `dos' and `cuatro' in the Spanish text) or functionally synonymous ones (`and' and `plus,' `y' and `ma's,' etc.) they are the same text. But others would argue that they cannot be regarded as the same even under those conditions, because they are composed of different signs even if those signs are synonymous. Besides, they might point out, there are different physical characteristics to contend with as well.

From all this it should be clear that the matter of the identity of texts is by no means easy to establish. A list of the necessary and sufficient conditions for the achronic sameness of texts does not seem to be readily available. The most likely candidates are the sameness of meaning, author, audience, context, arrangement of signs and signs themselves. Let us look at these possibilities in the order given.

A. Sameness of Meaning

The condition that appears at first as most obvious in connection with the sameness of texts is sameness of meaning. I take sameness of meaning to indicate at least that two texts have the same meaning if their truth conditions are the same. Thus, `Dos y dos son cuatro' and `Two and two make four' have the same meaning because the conditions under which one would be true are the same conditions under which the other would be true, and the conditions under which one would be false are also the same conditions under which the other would be false. It is, of course, common to find texts which have meanings to which truth value cannot apply, e.g., commands. And there may be other exceptions to this conception of meaning sameness. But the minimal understanding of meaning sameness provided here may serve to give a rough idea of what is involved. The presentation of a satisfactory view of meaning identity would require more space than I can provide for it in a discussion of textual identity, thus I cannot address the many problems it poses at this point.

One could easily see why it might be argued that sameness of meaning is both a necessary and sufficient condition of the achronic sameness of texts, that is, that texts that mean the same are the same text. Indeed, this condition allows us to distinguish (9) from (1)-(8) above, for its meaning is differet from that of the other texts on the list. This condition is particularly attractive, of course, to those who identify a text with its meaning.

There are various ways in which this view may be assailed, however. Some of these ways are ineffective, but there is at least one I consider effective against the view that sameness of meaning is both a necessary and sufficient condition of textual sameness.

Although sameness of meaning does appear to be a necessary condition of textual sameness in the sense that texts that do not mean the same coulE not possibly be the same texts, it would be difficult to argue that it is also a sufficient condition. That it is a necessary condition can be seen clearly in the reasons why (9), for example, is not the same text as any of the texts (1)-(8) it does not share the same meaning. On the other hand, to have the same meaning does not ensure textual sameness. If that were the case, texts (1)-(8) would all be the same text, something which very few would be prepared to accept, and something I certainly do not wish to defend. To do so would imply that two texts composed by different authors in different languages could be regarded as the same text provided their meanings were the same. It would also suggest that a painting and a written text would have to be regarded as the same text if their meanings were the same. But none of this seems acceptable. Therefore, we must conclude that sameness of meaning is a necessary condition of textual sameness under the specified strictures, but it is not a sufficient condition of it.Foot note 111

B. Sameness of Author

Apart from sameness of meaning, there are still other alternatives to account for textual sameness that may be explored, however. Some of these do not seem very promising. Take, for example, the author.Foot note 112 It is possible to argue that a text is the same if the author is the same, but that would not make much sense. Such a view would imply that all the texts an author writes are one and the same regardless of the differences that may exist among them. One could, of course, argue that by «the same text» in this case is meant «part of the same text.» And indeed this is a sense that is sometimes used in discourse. We sometimes speak of everything an author has produced as a single work. But, obviously, that is something different from saying that every text an author has produced is the same text in the sense that it is identical with every other text produced by him. Thus, sameness of author could not be a sufficient condition of sameness of text, and the main reason is that the condition of sameness of meaning would be missing. But what if that condition were added. Would sameness of meaning and sameness of author combined ensure textual identity?

Unfortunately this combination does not seem to do the trick either. For the same author may create two different texts that have the same meaning, say a poem and an essay.Foot note 113 Examples that illustrate this point abound in everyday experience, where we use different sentences to mean the same thing. Indeed, texts (1), (3), (4), (5), and (8) could have the same author and yet in spite of the sameness of meaning and author they would share could not be considered the same text.

So much then for the sufficiency of authors for textual identity. But what of necessity. Is sameness of author a necessary condition of the sameness of texts. Could there be two instances of the same text produced by two different authors. This is one of the puzzling questions that Borges explores in his «Pierre Menard, Author of the Quixote.» His answer there is negative, although Borges seldom gives an unambiguous answer. He assumes that the authors in question are separated by important temporal and cultural differences which alter the meaning of the text. So that, although the signs of which the texts are composed are the same, the meanings of those signs are different because of the cultural distance between them.Foot note 114

But what about contemporary authors. Indeed, what about authors who are alike inasmuch as that is possible, say identical twins raised in the same environment and so on. Couldn't we say that in that case the authors of the text are different but the text is the same. Indeed, sameness of texts does not seem to require sameness of author if by «sameness of author» is meant the same individual person.

In short, sameness of author is neither a necessary nor a sufficient condition of the sameness of texts. But when we say «sameness» of author, we are speaking of numerical sameness. That is, it is altogether possible for two similar, but numerically different authors to produce the same text. This is obvious from the case of twins. But does it make sense to say that persons who are not only numerically different, but also different in other respects, could produce two instances of the same text?

In one way it is obvious that this can happen, for two persons may differ in respects that would have no relevance to their composing a text. For example, they may differ in the fact that one of them has a tiny birth mark on his back and the other does not. But would it make sense to say that two persons could be authors of the same text even though they had substantial differences in outlook, education, and so on?

Logically speaking, I do not see how this question can be answered negatively. Indeed, when it comes to short, simple texts, it does not seem difficult to think of counterexamples. There is no reason why two authors could not have produced two instances of the text «Please, do not smoke» independently of each other. But it is difficult to accept the real possibility of this happening with long and complicated texts, such as the text of Cervantes's Don Quixote.

C. Sameness of Audience

Another not very promising factor that may be used to account for textual identity is the contemporaneous audience. What I mean by the contemporaneous audience is the audience the author intended to reach with the text during his or her own time.Foot note 115 The question is, can sameness of audience be a necessary and/or sufficient condition of the sameness of texts. As far as a sufficient condition is concerned, it is clear that it cannot, for the same audience can be the audience of different texts, that is, it can be meant to be the audience, by different or the same authors, of different texts.

On the other hand, when it comes to being a necessary condition, the situation is different. The difference comes about because the particular audience the author has in mind for a text influences in important ways what the author produces as a text. All texts are enthymematic; they contain lacunae that are meant to be filled by the audience. The meaning the author intends to convey through the text to an audience is incomplete unless what the audience is meant to supply is taken into account. It is not necessary for the audience to be the individual audience the author had in mind; the audience that is pertinent is the type of audience the author had in mind, just as it is not the individual author that is pertinent for the achronic sameness of texts but rather the type of author. It is the type of audience that can supply the needed elements for the text. This means that, although sameness of audience is not a sufficient condition of textual sameness, it is indeed a necessary condition in cases where the text contains lacunae to be filled by the audience.

Now, someone may wish to argue that if a particular type of audience is a necessary condition of the identity of a text, then the author (or type of author) should also be one. Indeed, it is the author who leaves out, intentionally or unintentionally, the parts of the text missing in the lacunae that the audience must fill. Moreover, the author's subjectivity has much to do with a text and its meaning. So how can the author be left out if the audience is thought to be necessary. Either both are left out or both are put in.

I do not want to argue that the author is not closely related to the text or not necessary for it. Indeed, the author is not only responsible for the selection and arrangement of signs that compose a text but also for the overall meaning.Foot note 116 Moreover, the author is also responsible for the lacunae that the audience has to fill, as already noted. Nonetheless, there is an important difference between author and audience, namely, that in composing a text the author, consciously or unconsciously, takes into account the audience and what it is meant to supply. Signs are selected, arrangements are made, and materials are included or excluded with the audience in mind. Thus the audience intended by the author is a necessary part of the puzzle that reveals the meaning of a text in this special sense.

D. Sameness of Context

What applies to the author and the audience also applies to context. Context is always important for the meaning of texts. For what appears to be the same text may have very different meanings depending on context and thus may turn out to be a different text. The threat, «Do not touch that or I will kill you» means quite different things when it is addressed by a mother to a child reaching for a fragile object than when it is said by a policeman to a burglar reaching for a gun. But sameness of context does not insure textual sameness. It is obvious from everyday experience that different texts can be or are uttered under the same (in all pertinent respects) conditions.

Moreover, one may want to argue that, unlike the case of the audience and the author, sameness of context is not even a necessary condition of textual sameness, for contexts may play no role in determining the meaning of some texts. Take (1) above. It would appear that the meaning of this text cannot be altered by surrounding circumstances provided, of course, that the signs of which it is composed and the arrangements in which they are organized have the determinate meaning we associate with them. The example that has been given, however, is an unusual one, for the texts we normally use in communication are not mathematical. Most frequently we communicate with texts that take for granted the context as a determinant of their meaning. A more sensible view, then, would be to argue that sameness of context is not pertinent for all texts and as such it is not a necessary condition of their sameness, but that it is certainly necessary in the case of texts where it is pertinent for the determination of their meaning. Note again that, as in the case of the author and the audience, the sameness of context that is pertinent is the sameness of type of context, not of individual context. In short, then, we may conclude that sameness of context is not a sufficient condition of textual identity, but that it is a necessary condition whenever the meaning of the text depends on it.

E. Sameness of Sign Arrangement

Another candidate for necessary and sufficient condition of the achronic sameness of texts is the arrangement of the signs that compose a text. But is sameness of arrangement a sufficient condition of textual sameness. Obviously not. Syntař is not enough to make two texts the same. This is quite clear from texts (1) and (9) above. The same syntactical structure may be common to different texts and, thus, cannot insure textual sameness. But is it a necessary condition. The question concerns, for example, texts (3) and (8) above. Can texts that follow a different arrangement be considered the same texts. If we are going to follow our ordinary intuition on this matter, I would answer negatively; in other words, arrangement must be the same. Yet the matter is not that simple.

In short texts like (1)-(9) above, it appears that any change in order results in a different text. In some cases the reason is evident: the meaning of the text is destroyed or modified. Consider:

1. 2 + 2 = 4

and let us exchange the `4' for the first `2' that appears in the text. The result is:

10. 4 + 2 = 2

whose truth value is different from that of (1). Now consider:

3. Two and two make four.

and let us scramble its words thus:

11. Make and two four two.

Obviously the result is jibberish. However, there are changes of order that do not change the meaning, even though we do seem to regard them as implying a change of text. Consider:

8. Four two and two make.

On the other hand, in very extensive texts, say Cervantes's Don Quixote, some changes of order would not generally be regarded as sufficient to warrant a change of identity. The reason for this seems to be that the relevance and importance of a change in order has to be seen in a total context. A change that results in a change of meaning clearly will have to be regarded as implying a change in the identity of the text at least in cases where the change of meaning is significant. If the change of meaning makes little difference for the overall meaning of a text, then the text can be regarded as the same. But if there is no change of meaning, changes can still be regarded as sufficient to change the identity of the text in cases where those changes alter the nature of the text in some sense. The change of (3) to (8) is a good example, for that change implies a change in the function of the text. Whereas (3) is primarily scientific, (8) appears to be literary. A change that implies a change of function, then, implies a change in identity. From this we can conclude that sameness of arrangement is a necessary condition of textual sameness except when the changes in question are such that neither the substantial meaning of the text nor its function are altered by it.

F. Sameness of Signs

Finally we come to signs, the components of texts. As before, the question we have to answer is whether sameness of signs is a necessary and/or sufficient condition of the sameness of texts. The first problem that we encounter with this question has to do with the meaning of `sameness of signs,' for it is by no means clear what conditions apply to it. Interestingly enough, the search for those conditions is surprisingly similar to the search for the conditions of the sameness of texts.

The main differences between signs and texts are, first, that texts are necessarily composed of signs, while signs are not necessarily composed of other signs, and, second, that the meaning of signs is relatively simple if compared with the meaning of texts. Thus, for example, `I' is a sign meaning «I» but is not a text, while `No smoking' is both a sign and a text. Signs composed of other signs do not differ essentially from texts except in terms of degree.

The fact that some signs may not be composed of signs does not mean that they are necessarily simple. Indeed, no sign is simple, strictly speaking. Even a dot has features and therefore presents some composition and complexity. That means that signs, like texts, involve an arrangement, as well as an author, an audience, and a meaning. The relative semantic simplicity of signs makes irrelevant the consideration of author or audience as conditions of sameness, however, for two different authors can very well use the same thing or type of thing as a sign of some meaning. The audience is generally irrelevant also because simplicity diminishes the role of the audience. On the other hand, context is very important. It is one thing for an officer next to a cannon to yell «Fire!» and another for a joker to yell «Fire!» in a crowded theater.

Having said that, let me give a few examples of signs to see if we can determine on what basis they may be regarded as the same.

a. bear (the noun)

b. bear (the noun)

c. BEAR (the noun)

d. bear (the verb)

e. oso

f. bare

g. rbea

On the basis of ordinary custom I would say that (a)-(c) are the same sign, (d), (e) and (f) are each different from each other and from (a)-(c), and (g) is not a sign at all, but simply a set of letters. (a) and (b) are different instances of the same sign and, therefore, are the same as far as the type to which they belong. (c) has the same meaning as (a) and (b) but has a different physical appearance, so we may ask whether it belongs to the same type as (a) and (b). (d) has the same appearance of (a) and (b) but has a different meaning and grammatical function. (e) is the Spanish translation of (a), (b) and (c). (f) is clearly a different sign altogether, even though when pronounced it sounds the same as (a)-(d). The difference in pronunciation is irrelevant, however, since sounds must be regarded as different signs from written signs although they may be used to convey the same meaning. The important thing is that (f) has the same components as (a) and (b) but the arrangement is different, and it has a different meaning. Finally, (g) again has the same components as (a)-(d) and (f) but it has a different arrangement which results in no meaning; that it has no meaning entails that it is not a sign at all.

From all this it would seem that the key factors to be considered in the sameness of signs are meaning, function, components, appearance, arrangement, and context. As in the case of texts, sameness of meaning seems to be a necessary condition of the sameness of signs, but not a sufficient condition. `Oso' and `bear' mean the same thing but are not the same sign. But `bear' (the noun) and `bear' (the verb) are exactly alike in all aspects but have different meanings, and therefore are not instances of the same sign.

Something similar could be said about function, for function is closely related to meaning. The meaning of `bear' (the animal) and of `bear' (the action) are different in part at least because `bear' functions as a noun in some situations and as a verb in others. However, it is also true that different signs with different meanings may have the same function, v.g., `bear' and `cat' in a sentence such as `The X is an animal,' where `X' is substituted by either one of them. Moreover, different signs with the same meaning can have different functions, as is clear in paraphrases and circumlocutions. For these reasons it would seem that sameness of function does not insure sign identity and thus that sameness of function is not a sufficient condition of it. On the other hand, it would appear that the identity of signs requires identity of function.

The case with appearance is likewise not simple. Indeed, `bear' and `bare' have the same appearance in sound, and `bear' (the noun) and `bear' (the verb) have the same visual appearance, and yet are not the same signs. So appearance cannot be a sufficient condition of the sameness of signs. But is it a necessary condition. Not in all cases, since `BEAR' and `bear' are the same sign and yet look different. This indicates that it is only some aspects of the appearance of a sign that are relevant for the sign, and these are those aspects that have been determined by the author and/or are generally accepted to be so in a particular context. Thus color, arrangement, design, size, and so on, are all features that can become necessary conditions of the sameness of signs, but they are not sufficient conditions, for sameness of meaning seems also to be necessary.

We may say, then, that the necessary and sufficient conditions of the achronic sameness of signs are three: 1) sameness of meaning, 2) sameness of function, and 3) sameness of features identified by the author and/or accepted as such in a particular context as relevant for meaning. Note that context should not be underestimated. Indeed, the difference between `bear' (the verb) and `bear' (the noun) depends on context. The two are different because the first is part of sentences such as «To bear such a burden is a virtue,» and the second is part of sentences such as «The bear liked the honey it found in the jar.»

Before I leave the discussion of signs I should make explicit a rather serious implication of the view presented here. The requirement that signs have the same meaning in order to be achronically the same implies that words which have different meanings are not the same signs. This seems counterintuitive, for we frequently regard a sign as the same even if it is used to mean different things. Take, for example, the word `father.' In a sentence such as «Philip was the father of Alexander,» the word is used to indicate biological paternity, but in sentences such as «Thales is the father of philosophy» it is used to mean that Thales was «the first» philosopher. The only answer I have to this problem is that, in order to preserve this intuition, we would have to give up too much. For giving up the requirement of sameness of meaning in the case of signs and also, as a consequence in the case of texts, creates too many problems, making it very difficult to account for sameness.

Having identified what `sameness of signs' means, we can return to the question that prompted the discussion of signs in the first place: Whether sameness of signs is a necessary and/or a sufficient condition of sameness of texts. And the answer is that it could not be a sufficient condition for the reasons already stated in connection with (3) and (11) above. Nonetheless, it would seem that sameness of signs can be a necessary condition of sameness of texts, since a difference of signs may affect both meaning and appearance. Consider the following two sentences:

12. He was a respectable man.

13. He was a dignified man.

Clearly these two sentences, although having the same structure, and so on, mean different things if the terms of which they are composed are being used in the ordinary sense. Thus, they constitute not one but two texts. On the other hand, what do we make of the following?

14. The Angelic Doctor wrote the Summa theologiae.

15. Thomas Aquinas wrote the Summa theologiae.

And of the following:

16. He made a contribution to the fund.

17. He made a donation to the fund.

In (14) and (15) we have two sentences which are exactly the same except for the fact that (14) uses an honorific title to refer to Thomas Aquinas and (15) uses his proper name. In (16) and (17) we have a similar case except that here the difference concerns the use of two different but synonymous signs, `contribution' and `donation.' Is (15) the same text as (14) and (17) the same as (16). I believe most of us would want to answer negatively because the texts are not composed of the same signs, even if those signs have the same meaning. And, indeed, some scholars would find it objectionable if someone were systematically to exchange all instances of `Thomas Aquinas' in their writings by `the Angelic Doctor.' They might object that, although the referent of the term is the same, they chose `Thomas Aquinas' and not `the Angelic Doctor' because they wanted to look at Thomas as a philosopher rather than as a doctor of the Roman Church. Whether this makes sense or not is debatable. However, for our purposes what is important is that authors would object to the exchange of expressions.Foot note 117 Similarly, authors would find objectionable the substitution of `contribution' for `donation,' even if they could not think of the reason they had in the first place for why they did not use the former term rather than the latter. Once the question is posed, however, they might say that they object to the exchange because of the differing features of the words, and so on. For example, in a poem, the sound difference between the two words may be important for the intended rhyme.

Still we might want to argue that the texts mean the same thing and thus there is no reason why (14) could not be regarded as the same as (15) and (16) as the same as (17). And indeed, as already noted above, they would be the same if texts were the same as their meanings. But if, as I have argued elsewhere, texts are not their meanings, but groups of signs selected, arranged and intended by authors to convey specific meanings to an audience in a certain context, then (14) and (15) cannot be the same, nor can (16) and (17).Foot note 118 The reason is that they are composed of different signs. That (14) and (15) or (16) and (17) may actually turn out to do the same job does not change the fact that they are different texts, just as some of the signs of which they are composed are different signs having the same meaning. Of course, the sameness of signs in turn depends on what the author regards as semantically significant, or is so in a particular context, as already noted earlier.

3. Conclusion

In conclusion, then, we have examined various conditions that appeared to be good candidates of the achronic sameness of texts, but we found that none of them taken by itself constitutes a sufficient condition of textual identity. Moreover, we found that the author, the audience and the context were related to the identity of texts only insofar as they affected meaning. Sameness of meaning incorporates, then, these conditions when they are pertinent and thus there is no need to list them as conditions separate from the sameness of meaning. That is not the case with the sameness of arrangement and sameness of sign composition. For we found that texts with the same meaning but composed of different signs, or of the same signs arranged differently, cannot be considered the same text. Thus arrangement and sign composition, although not sufficient conditions of textual sameness, may become independently necessary conditions of it. The reason is quite simple. Texts are mixed entities. They are artifacts with meaning. As such, their conditions of sameness must include conditions of artifactual sameness (arrangement and composition) and of meaning. It is all these conditions put together that constitute the necessary and sufficient conditions of their achronic sameness. We may formulate them thus:

A text X is achronically the same as a text Y if and only if (1) X has the same meaning as Y, (2) X and Y have the same syntactical arrangement (with the proviso noted in E), and (3) X and Y are composed of the same type signs.

The question that I set out to explore in this paper had to do with what makes texts the same, that is, with textual identity. For its answer, three types of sameness were distinguished: achronic, synchronic and diachronic. The latter two involve time and so are more restrictive; thus I concentrated on achronic sameness. After examining various possible views we reached the conclusion that there are three conditions which, taken together, constitute the necessary and sufficient conditions of the achronic sameness of texts and hence explain their identity: sameness of meaning, of syntactical arrangement and of type-sign composition. Going back to the example used at the beginning of the paper to introduce the problem of identity we can now understand how different copies of the text of Cervantes's Don Quixote are the same text, for they have the same meaning and they are composed of the same type signs arranged in the same way. Thus, in spite of the many differences that characterize them, they are still to be regarded as copies of the same text.

Jorge J. E. Gracia

State University of New York at Buffalo

Fac. of Social Sciences

607 Baldy Hall

Buffalo, New York 14260-1010