Inconsistency, paradox and linguistic content. Did Bhartrhari offer a solution for truth-paradoxes?
The present paper contests an assertion made by some scholars to the effect that a specific textual passage in a treatise dedicated to linguistic philosophy provides a way to the solution of truth-paradoxes with the objective of shedding light on their nature. In this connection a particular chapter possesses relevance which is traditionally named Sambandha-Samuddesa (henceforward 'SS') and which belongs to a treatise with the title Vakyapadiya or Trikandi dealing with topics of grammar, linguistics and philosophy of language. It is attributed to a writer, called Bhartrhari, who supposedly flourished in the 5th or 6th century A.D. (1). According to the position advocated here, the renowned sequence of words sarvam mithya bravimi occurring in verse 25 of the Sambandha-Samuddesa was not meant to raise the issue of a truth- or liar-paradox but belongs to a different argumentative context where the internal consistency of proclaiming a linguistic tenet is at stake which contends that meaningful linguistic items never express something about any way in which they are related to that which they mean or express. However, an investigation pertaining to certain objective qualities of statements leading to paradoxes or antinomies of truth is indispensable for vindicating the result that, even if the proposed interpretation should not be adopted, the text cannot offer any solution of truth-paradoxes, notwithstanding the fact that an investigation of the Sambandha-Samuddesa is apt to yield some relevant clarifications. It is possible to demonstrate that statements pertaining to truth possess themselves truth-conditions which are more complex than corresponding linguistic formulations and that they exhibit features of context-sensitivity which are not caused by the occurrence of indexical constituents. If one assigns to ascriptions of truth or untruth a content which mirrors their truth-conditions and supposes that this type of content is the primary bearer of a property of absolute truth it follows that items ascribing in a direct or indirect manner untruth to themselves are definitely assessable as untrue or false because their truth-relevant content embodies a contradiction. Accepting the proposed suggestions permits to preserve certain pre-theoretical intuitions and implies that at least truth-paradoxes do not furnish a reliable philosophical basis for varieties of 'paraconsistent' logic, and specifically for the position of 'dialetheism'. On the other hand the implicit contention that treatments of truth-paradoxes ought to address issues of content supports the dismissal of the contention that the Sambandha-Samuddesa offers us a way to their solution.
This study deals partly with a topic of philosophical history and partly with matters of philosophical analysis. It contains both 'exegetical' and 'analytical' ingredients. The initial objective lies in showing the groundlessness of claims to the effect that one of the most prominent representatives of the philosophy of language in the Indian tradition has indicated a fruitful way of solving some varieties of truth-paradox. The rejection of such contentions is regarded as warranted even in consideration of the fact that the wording of pertinent textual passages allows for diverging interpretations in accordance with linguistic and philological criteria. On the other hand precisely the assumption that the relevant textual source deals with a problem of making consistent statements about meaning or content and with an issue that clearly differs from the predicament of truth-paradoxes permits to identify thoughts which are apt to render assistance for a better understanding of the nature of truth-paradoxes. It will be undertaken to demonstrate that a textual exegesis which yields that result is feasible and justified despite divergences from traditional interpretations.
As suggested by its name (Sambandha-Samuddesa), the chapter which constitutes the object of exegetical investigation treats relation as its topic, in particular relation between linguistic items and some or other type of content ascribable to them. Allegations to the effect that the author of the text has not merely dealt with a truth- or liar-paradox but even indicated a way for its solution are based on a subsection of this chapter and can be found in fairly recent publications. In Houben (1995a, 382) it is even claimed that Bhartrhari "provides us with the key to an interesting and elegant solution" to the family of paradoxes to which the Liar Paradox belongs. It might be noted in passing that the employment of the definite article in 'the key' intimates that there is no other key to an 'interesting and elegant solution' of some group of paradoxes than the one which this philosopher offers us.
Although the entire SS encompasses 88 metrical units it is only the initial segment up to verse 29 which matters for the present investigation. The exegetical part of the present study pursues the goal of supporting the presumption that in the SS Bhartrhari did not even intend to deal with any variety of truth- or liar-paradox and that he was concerned with a different theoretical issue. In contrast, the second part pertains to philosophical or analytical matters inasmuch as it explores some consequences which can be reasonably derived from the existence of truth-paradoxes. Despite the marked difference of their character the two sections are significantly related: It has been suggested above that the investigations of the latter part are apt to strengthen the conclusion that the SS did not present a basis for solving any variety of truth-paradox and intimate that the same verdict should be appropriate even if one would prefer different alternatives regarding details of interpretation with respect to the relevant textual segment. This outcome is based on the supposition that issues possessing central relevance for assessments of truth-paradoxes have not been considered in the investigated text. Hence it can be merely conceded that the SS indirectly contributes to some clarification as far as truth-paradoxes are concerned.
Although topics which are explored here were dealt with in a previous publication of mine (henceforward 'AL') (2), the present article is not a mere mirror or summary of the earlier study. On the one hand, it differs by presenting a more concise delineation of the interpretation of the relevant textual source involving some modifications in matters of detail and on the other hand it furnishes additional explications regarding the alleged consequences of truth-paradoxes, partly against the backdrop of a recent publication in the field of theoretical philosophy.
Before addressing the two major topics mentioned above, it appears apposite to avert a precipitous equation of the terms 'liar-paradox' and 'truth-paradox'. We may designate by the expression 'liar-paradox' any variety of paradox which results from assumptions in which the properties of lying or being a liar are attributed to persons or the quality of being a lie to linguistic units or acts of employing them. In contrast, the term 'truth-paradox' ought relate to paradoxes resulting from the supposition that linguistic items or their content exhibit properties related to truth, specifically, apart from the characteristic of being true or expressing a truth, to the qualities of being false, of expressing a falsehood, of lacking the property of being true or of failing to express a(ny) truth. The expression 'paradox' can refer to any sentence or proposition which seems prima facie impossible to consider as true whenever it can be derived from assumptions which prima facie appear undeniable by employing principles of derivation whose validity seems indubitable. In the present context the word 'antinomy' should denote varieties of paradoxes which are logically inconsistent. By extension it might be permitted to call 'paradox' also any argument implying a paradox or to denote by this word sets of assumptions and principles of derivation which entail a paradox in the narrower sense.
There is no need to add further annotations to the elucidation of the expression 'paradox' because the underlying idea is on the one hand sufficiently clear for the current purposes and on the other hand not novel. In view of the fact that the precise content of the notions of a liar, of a lie and related ones is not immediately obvious it is conceivable that different persons (are inclined to) cherish deviant views as far as the question of the extension of the concept of 'liar-paradox' is concerned. Nevertheless, the following points should be incontestable:
First, ascriptions of lying or being a liar encompass several varieties relating to the extent of ascription. One can assign to a person the attribute of lying either with respect to some particular (assertive) utterance or with respect to a plurality of utterances and in the extreme case to the totality of his utterances. Classifying somebody as a liar might mean that the concerned subject is always or practically always lying; it can, however, also mean that somebody is unusually often lying or that he is disposed to lie under certain circumstances. Evidently if somebody should make the statement
(1) I am a liar.
intending to communicate that he is always disposed to uttering a lie if this entails significant advantages for himself, we could credit that person with a capacity of achieving insights about himself and--even supposed the subject of utterance is not a politician--would consider his statement as true rather than as paradoxical.
Second, given that not only a relation of entailment between being untrue and being a lie as well as uttering an untruth and uttering a lie, but also a reverse relation between uttering a lie and uttering an untruth fails to hold good, it must be regarded as probable that the extensions of the terms 'liar-paradox' and 'truth-paradox', as explicated above, significantly differ. A denial of the reverse entailment can be motivated both by common intuitions and by reference to the tradition of Western philosophy. Linguistic intuitions suggest that the concept of a lie is intimately connected with intentions of linguistic subjects and their beliefs. Specifically, it is not absurd to classify as a lie a statement with a true content, given that the subject making that statement was convinced of its untruth. This conception appears to underlie the statement to be found in footnote 8 of G. Frege's article 'Uber Sinn und Bedeutung' according to which we attribute to persons the quality of lying precisely if they assert as true some content and are at the same time convinced of its falseness (3). There is no need to explore statements concerning the idea of lying in the tradition of philosophy in the present context. For in a recent publication--viz. W. Kunne, Epimenides und andere Lugner (henceforward 'EL')--it has been demonstrated that several authors of the Western tradition did not regard the actual truth-value of an asserted content, but the opinion of the asserting subject concerning its truth or untruth as decisive for the question whether or not an utterance constitutes a lie. Since on the other hand, EL provides relevant considerations supporting the tenet that the actual untruth of an asserted content is not a necessary condition of lying, it is not needed here to present a more detailed account of the reasons for rejecting the assumption that being a lie entails objective untruth (4). Even if that position were not accepted it would not be justifiable to equate the concepts of 'liar-paradox' and 'truth-paradox' from the outset. After all, the contention that not all truth-paradoxes are equally liar paradoxes in the sense explicated above can be hardly denied. If, on the other hand, one accepts the proposition that actual truth of an asserted content is compatible with classifying something as a lie, then it is not immediately obvious that specimens classifiable as liar paradox really exist. Particularly with respect to an utterance of sentences like:
(2) Everything which I ever utter (i.e. have uttered, am uttering now and will utter in the future) is a lie.
it is difficult to discern a paradoxical consequence. Supposed that the subject uttering the sentence has in fact always lied before and will always lie later and that his present utterance does equally not reflect what he actually believes at the time of his utterance, it might follow that the uttering subject possesses mistaken beliefs about his previous biography or his future career. Such a situation might be improbable, but is it impossible? If it is, however, possible, why should the consequence be paradoxical? As it seems, the situation dramatically changes if the predicate 'is a lie' in (2) were replaced by 'is an untruth' or 'fails to express anything true' or something similar. It will emerge later that those questions as well as the relation between the notions 'liar-paradox' and 'truth-paradox' are relevant for an assessment of the section of the SS which is considered below.
The focus will lie on the question whether the SS provides a solution for truth-paradoxes because this issue is most relevant.
Notwithstanding our presumption that the pertinent section of the SS constitutes a coherent passage dedicated to a particular topic there is no need to analyse in this connection all verses which the segment contains. Only those remarks will be scrutinized which are of potential relevance for the determination of the central underlying issue and the assessment of the question whether the passage addresses a truth-paradox and indicates a way to its solution. A survey and account of the section in its entirety has been provided in AL, and an extensive study can be found in Houben (1995). The segment which is of pivotal importance in the present context consists of the verses 1 and 3-4 as well as the section of the stanzas 19-29. These read as follows:
1 By words that are uttered a cognition of the one who uses [them], an external object and the own form [of the words] are comprehended. Their relation is fixed.
3 'This is a signifier/an expression of this [and this] is its [correlate] [which is] to be signified/expressed', thus the connection between word and [its] object is comprehended by the sixth case ending (= the genitive suffix); hence also a state of things is pointed out.
4 There is no term which is a signifier/ expression of the relation in accordance with its own characteristic; due to complete dependence its form is not indicated.
19 Yet neither the utterer nor the hearer approaches through words the [relation of] attainment [between words and their objects] called 'inherence', which goes beyond the characteristics that are to be expressed.
20 If that which is to be signified/expressed by '[it is] not to be signified/expressed' would be ascertained as [something which] is to be signified/expressed, then it would become something which is to be signified/expressed.
21 If on the other hand that is not ascertained in this manner as something which is not to be signified/expressed, [then] the very situation of it which is to be signified/expressed is not ascertained.
22 It is definitely not so that by those words this situation is negated even regarding that about which it is said that it is not to be signified/expressed in this, in another and in every manner.
23 For regarding an object which has the nature of a doubt and is subordinate [with respect to its object] no other doubt is operative in so far as it does not lose its own nature.
24 And when a cognition of ascertainment is ascertained as an ascertainment then the cognition does not persist in its own character.
25 [With the words:] 'Everything [which I speak] I speak falsely' this sentence is not intended to be meant. For if it is said falsely the intended object is not attained.
26 And what is operative in the form of a signifier/a [means of] expression is not something to be signified/expressed. That by which something else is communicated is not there (in the same context) to be communicated.
27 Just as [with the words:] 'An assertion is not proving' this very assertion is not designated, in the same manner also no characteristic of it is ascertained [there].
28 Because there is no further operation of an operation therefore one should not resort to a contradiction or an infinite regress in all cases.
29 In the same manner as the beginningless fitness of the sense-faculties with respect to their objects the relation of words with their objects is a beginningless fitness. (5)
We presume that the central concern of the whole section is embodied in the formulations 'Their relation is fixed' (tesam sambandhah samavasthitah) of the initial verse and 'the relation of words with their objects is a beginningless fitness' (andair arthair sabdanam sambandho yogyata). To be sure, the import of the two remarks is far from obvious. Nevertheless, it is (in consideration of various details) not unreasonable to distil from them the tenet that the relation between words--and meaningful linguistic expressions in general--on the one hand and whatever they mean on the other hand is established in advance before their employment in individual utterances. In the light of the initial verse one can suppose that the author of the text considered in the first place linguistic expressions occurring in oral utterances. Hence it is not implausible to conjecture that this stanza should refer to items which in modern terminology are called 'linguistic tokens'--or to entities akin to linguistic tokens (6)--and that the remark represents the claim that from the oral production of (meaningful) linguistic tokens one can recognize three types of items, viz. 1. some cognitive state of the producer of the utterance (jnanam prayoktur). 2. some external object and 3. something which can be regarded as the 'own form' of linguistic tokens.
Since the original correlate of the expression 'object' is artha-, and artha- amalgamates the imports of 'object' and 'meaning', one is entitled to suppose that the expression rendered by 'external object' should refer to items which can be characterized as something which linguistic expressions mean, if they are meaningful. To be sure, this suggests an outlook on meaning which could appear outdated from the modern perspective, namely the opinion that expressions possess meaning in virtue of standing for particular objects (7). There is no need to address the question of whether the writer of the text in fact accepted that view. For it will turn out that the validity of the central claim(s) of this section is not affected by this problem. Since the precise determination of the import of 'own form' is rather irrelevant for the present topic, it should suffice to point out that identifications either with sound-patterns or with linguistic types deserve to be taken into consideration. Central importance is to be accredited to the claim embodied by the remark that their relation is 'well-established'. Although this statement involves various problems of interpretation it is legitimate to disregard certain intricacies of this issue here. Various alternative interpretations differing in details possess the common implication that the relation between meaningful linguistic expressions and what they mean is 'steady' or 'remaining fixed' (samavasthita-). The acceptance of this tenet is corroborated by the remark of the last verse where it is said that the (beginningless) relation between words and (their) meaning-objects is a beginningless fitness or suitability like the beginningless fitness between sense-faculties with respect to their objects (8). The central section of this segment, in particular the segment of the verses 8-18 (which is not investigated in the present article), indicates that Bhartrhari was concerned with the question of how it is possible that the relation between uttered words and that which they mean is not arbitrary, that there is a restriction to the effect that if some uttered linguistic token instantiates a particular type its communicative value is, to some extent at least, determined by the fact that it instantiates this and no other type. Seen in this light one can surmise that in the pertinent section of the SS the view is advocated that the observable restrictions between individual linguistic tokens and their import is due to the fact that their general form or type is endowed with a specific expressive capacity which regulates the connection between (uttered) linguistic tokens and their meaning-contents, if they are meaningful. As that regulative capacity does not originate simultaneously with the production of individual tokens it is beginningless at least in the sense that it has not a beginning which coincides with the beginning of the existence of the produced item. By supposing that a vindication of this thesis is the overriding aim of the whole segment one is in a position to interpret the entire passage of the SS up to verse 29 as embodying a coherent train of thought.
This assumption could, however, evoke the objection that it hypothesizes a fairly trifling tenet as a major object of concern. Against this
it can be retorted first that if a proposition appears commonplace from the perspective of a modern interpreter it does not need to constitute a commonplace from the perspective of the author of the text and his contemporaries. In fact the pertinent proposition is by no means trifling against the background of the milieu in which the text had been written. It is, for example, in disaccord with views advocated in Buddhist schools, in particular with the tenet that everything which exists possesses merely momentary existence. Presumably this view had been already advocated before the time the SS was written, even if particular ways of proving this theorem were developed later. Second, it need not be admitted that establishing a thesis to the effect that meanings of linguistic tokens are regulated by semantic properties of linguistic types which they exemplify constitutes the ultimate goal of the undertaking. Schools existed which maintained that words possess everlasting existence insinuating an eternal relationship between words and their meaning-objects. Thus Bhartrhari, even if he had not intended to advocate this particular claim in the pertinent section of the SS, might have aspired to pave a way for the acceptance of those or similar stronger dogmas by indicating the existence of affinities between them and propositions which ought be acknowledged on the basis of common sense views. Thus the occurrence of the word 'beginningless' (anadi-) permitting various explications in verse 29 is possibly no coincidence. Anyhow, the fact that the tenet formulated in the last verse of the segment is combined with non-trivial consequences holds true irrespective of whether one hypothesizes a weaker or stronger reading.
For someone who advocates the proposition that meanings of uttered expressions are (fully or partly) determined by semantic properties of linguistic items which they instantiate (such as corresponding linguistic types) it is natural to dismiss the idea that the pertinent determination should be accomplished by definitions or stipulations pertaining to their meanings. But this implication creates a problem: It appears undeniable that meanings of expressions can be explained, and in principle even stipulated. To be sure, such meaning-explanations are observed to be accomplished by employing expressions which differ from those whose meaning is explained or stipulated. But if it is in principle possible to explain meanings, why must one discard the hypothesis that linguistic tokens could themselves explain their own meanings? If, on the other hand, one acknowledges that self-explanations might occur at least in some cases, what could legitimate a rejection of the assumption that this happens in all cases, and that accordingly the postulation of expressive capacities of items instantiated by linguistic tokens is unwarranted. In this context one ought not invoke the fact that in cases of ordinary meaning explanations the explaining items are themselves meaningful. For the crucial question is how in general expressions including those used for purposes of explanation can be meaningful, so that by declaring that the explaining expressions possess meaning on account of semantic qualities of items they instantiate one is entangled in an argumentative circle.
We presume that the writer of the SS addresses precisely this problem in the verses 3 and 4. In the first of those stanzas it is acknowledged that one can indicate relations between expressions and their meaning-objects by certain linguistic forms and thereby communicate that such and such expressions possess such and such meaning. In contrast, the following verse modifies, or rather specifies, the preceding statement by asserting that linguistic expressions cannot perform with respect to themselves the same explanatory function which other expressions could perform. It needs to be emphasized that the formulation of the relevant verse is extremely equivocal and that in the tradition of both Indian and Western exegesis altogether different interpretations have been advocated. Hence the above proposed account does not presuppose that deviant interpretations are impossible on linguistic and philological grounds but merely that the recommended alternative matches with the linguistic details of the pertinent textual segment at least to the same degree as envisaged alternatives.
As the verse presents a justification for the thesis in the form of the expression atyantaparatantratvad, which has been rendered above by 'due to complete dependence'--and was translated as 'because it is extremely dependent' in Houben (1995) and similarly elsewhere--one can control the admissibility of any interpretation of the thesis by the criterion of its conformity with the explicitly communicated justification. As the formulation of the reason is equally equivocal one cannot definitely settle its intended import by purely linguistic criteria. One can, however, demand that any admissible interpretation must permit the construction of an intelligible argument for a thesis which is in accordance with the linguistic facts. I allege that the following account satisfies this postulate.
A binary relation can be regarded as completely or 'extremely' dependent in the sense that its identity does not only depend on the items which are related by it to something or the other, but also on the items to which objects are related by the relation. In other words, if it holds good that 'a' but not 'c' is related by '[R.sub.1]' to 'b', whereas both 'a' and 'c' are related by '[R.sub.2]' to 'b', then '[R.sub.1]' and '[R.sub.2]' are different relations. If, on the other hand, some '[R.sub.3]' holds good between 'a' and 'b', but not between 'a' and 'd', whereas '[R.sub.4]' holds true between 'a' and 'b' as well as between 'a' and 'd' then '[R.sub.3]' and '[R.sub.4]' differ. We might in a slightly technical diction say that both the domain and the range matter for the identity of a relation. Analogous statements could be made concerning relations between three or more items, which apparently do not lie in the focus of the SS and its writer. Now, if we suppose that a certain expression 'e' were related by some 'meaning-relation' '[R.sub.m]' to some 'meaning-object' 'o', assuming that 'o' exhausts all the meanings of 'e' and add to this the supposition that 'e' were also related by '[R.sup.*]' to 'o' and '[R.sub.m]', then the assumption that '[R.sub.m]' and '[R.sup.*]' are the same forces us to reject the original assumption that 'o' exhausts all the meanings of 'e'. To avoid this result we would need to assume that '[R.sub.m]' and '[R.sup.*]' are different relations. But the supposition that 'o' as well as '[R.sub.m]' are 'meaning-objects' of 'e' would compel us to acknowledge that there is a meaning-relation '[R.sup.*.sub.m]' connecting 'e' both with 'o' and '[R.sub.m]', and this relation ('[R.sup.*.sub.m]') needs to differ from '[R.sub.m]', if the original assumption that 'o' exhausts the meaning-objects of 'e' is maintained. Thus the idea that linguistic expressions are related to meaning-objects due to the fact that the relations between them and the meaning-objects are also meaning-objects for the very same expressions engenders an infinite regress. Due to the circumstance that in the Indian tradition it was commonly acknowledged --whether rightly or not is a different matter-- that if some supposition entails an infinite regress, technically called anavastha, then the supposition cannot be true, we can safely assume that the writer of the SS would be disposed to reject the above envisaged situation on this account. That Bhartrhari should be in fact disinclined to accept the idea that every expression possessing a meaning-object possesses as its meaning-object any meaning-relation between itself and any of its meaning-objects is vindicated by the circumstance that it militates against the view that linguistic expressions possess only a finite number of meanings. Thus a reluctance to accept the envisaged hypothesis does not presuppose a rejection or ignorance of linguistic ambiguity.
The crucial point can be detached from its presentation in terms of meaning-objects and meaning-relations. It appears that Bhartrhari does not distinguish between different categories of linguistic expressions, such as proper names, general terms, predicates and sentences. To be sure, the description in terms of relations between linguistic expressions and objects fits best with a semantic account of proper names. But there can be hardly any doubt that the theorem formulated in verse 4 of the SS was not meant to relate to any specific category of linguistic expressions. Given that the point can be applied to all categories of linguistic expressions, it can be illustrated with respect to the linguistic category which matters most in the present context, namely (declarative) sentences. This can be achieved without employing the expressions 'object' and 'relation' at all. The relevant claim is embodied in the following formulation:
(T) For any sentence s, if s means/expresses that p, then s does not mean/express that s means/ expresses that p.
On the assumption that only sentences are at stake which are not ambiguous the theorem is extremely plausible. For the contents of 's means that p' and 's means that s means that p' appear to differ. The justification of this intuition is vindicated by the circumstance that it is conceivable that the truth-values of the contents expressed by such pairs of sentences, hypothesizing an identical reference of the subject terms, differ. The same holds good if 'express' is substituted for 'mean'. Hence the assumption that any sentence expresses some (propositional) content or the other and simultaneously expresses (the fact) that it expresses that content entails the consequence that it expresses (at least) two different contents at the same time and hence cannot be free of ambiguity. If we connect this result with the previously considered phenomenon of meaning explanations, we recognize that the hypothesis of sentences' explaining their own meaning, just by expressing what they mean or should mean--something which other sentences can do--, necessitates the acknowledgment of ambiguity for such sentences. Against the background sketched before, this is significant because it follows that the supposition that all meaningful linguistic units are self-explanatory in the described manner militates against the familiar assumption that there are or at least can be non-ambiguous sentences in a language. If, on the other hand, one drops the presupposition of lack of ambiguity, a variant theorem results, which can be formulated as follows:
([T.sup.*]) For every sentence s, if s possesses any content as its meaning at all, then there is at least one content c, such that s possesses c as its content but does not express that it possesses c as its content.
The acknowledgment of ([T.sup.*]) is mandatory on condition that sentences do not possess an infinite amount of contents. If this theorem is retranslated to the diction of 'meaning-objects' and 'meaning-relations' it means that there is at least one meaning-relation connecting linguistic expressions and what they mean which is not a meaning-object of the expressions concerned.
According to a received reading of verse 4 of the SS the stanza embodies a claim to the effect that no expression designates the relation between word and meaning (in general) or that the particular relation called 'inherence' (samavaya) cannot be designated by any expression (9). In contrast (T) embodies a more modest contention according to which words cannot designate their own meaning-relation, which enjoys a higher degree of prima facie plausibility than the alternatives.
([T.sup.*]), however, provokes the problem whether the hypothesis of its propagation can be reconciled with the wording of the text because it confines itself to a mere existential statement, whereas the formulation of the verse does not insinuate that the exclusion should pertain to some member of possibly several meaning-relations associated with a meaningful unit. One might presume that the expression which has been rendered above by 'in accordance with its own characteristic' and which could be equally understood in the sense of 'in accordance with its own quality', svadharmena, should indicate that expressions cannot specify the character of their meaning-relation, in the generic sense, in its entirety; that means that they can provide at best a partial characterization, which would bring the reading in closer accordance with ([T.sup.*]). In spite of this, it is recommendable to abstain from this conjecture not only because it is doubtful, but also because it is dispensable in the present context of discussion.
It might be reasonable to assume that, even if Bhartrhari's argument does not strictly support the stronger conclusion, the writer of the text intended to advocate the view that no expression designates any of its own specific meaning-relations and that this holds true particularly for sentences and the meanings they express. It appears hardly possible to construe a sentence which in one reading expresses some content and in a second reading expresses some comment about this content, in particular a comment to the effect that the concerned content is expressed by the pertinent sentence.
It is true that the formulation does not explicitly indicate that not meaning-relations in general, but own meaning-relations of expressions are at stake, except on the hypothesis that the component 'own' (sva) in svadharmena is co-referential with the preceding expression abhidhana- ('term', 'expression') and not with the subsequent word sambandha- ('relation'). Since it is impossible to decide the question on the basis of purely linguistic considerations, only additional criteria could provide support for one or the other alternative. In the preceding section the interpretation according to which own semantic properties of meaningful linguistic expressions are at stake was presented against the background of maximal theoretical plausibility, leaving undecided whether the vital specification has been explicitly indicated. Therefore the circumstance is important that the section constituted by the verses 19-29 provides supplementary corroboration for the envisaged reading.
Given that the noun qualified by the expressions 'called 'inherence'' (samavayakhya-) and 'which goes beyond the characteristics to be expressed' (vacyadharmativartin-) in verse 19 refers to the meaning-relation which was in the focus of the previous deliberation, part of the sentence could be understood as representing a remark which comes close to a truism. This would be the case if the sequence of words rendered above by 'yet neither the utterer nor the hearer approaches through words the [relation] of attainment [between words and their objects]' (praptim tu ... prayoktapratipatta va na sabdair anugacchati) should convey the thought that generally neither a speaker nor an interpreter envisage meaning-relations as something which ought be communicated or conveyed. If a sentence, such as:
(3) Managua is the capital of Nicaragua.
is uttered, hardly anybody would be disposed to consider as a content which is or should be communicated the fact that (3) means or expresses that Managua is the capital of Nicaragua instead of or in addition to the fact that Managua is the capital of Nicaragua. However, the statement represented by the pertinent concatenation of words would evoke the appearance of a truism to a lesser degree, if it were understood as entailing as a matter of necessity that the semantic fact concerning the meaning of (3) cannot be something which (3) means--either exclusively or in addition to something else--and that the same holds good mutatis mutandis for any other meaningful linguistic item. Remarkably the verse has been rendered in the translation of Houben 1995 as However, neither the speaker nor the hearer can approach with words the relation (prapti) called samavaya (inherence), which is beyond the property of things that can be signified.
Presumably the addition of the modal element represented by 'can' is fully justified, and the subsequent context supports this reading. Against the background of the preceding considerations the addition of the modal component should however not yet suffice for making the relevant theorem fully explicit. For if the preceding analysis is mainly correct, the writer of the text did not intend to disclaim that facts concerning semantic relations can be formulated in language at all but merely that linguistic units cannot state regarding themselves that they possess the semantic relations which they in fact possess.
We may, however, concede that Bhartrhari was willing to advocate a proposition which is a slight generalization of (T) above and which could be formulated as follows:
([T.sub.G]) For any sentence s, if s expresses that p, then s does not equally express any comment concerning itself (viz. s) to the effect that for some [PHI]: s [PHI] that p.
The intended import of ([T.sub.G]) can be illustrated with respect to any token of (3) that, given that (a token of) (3) means or expresses that Managua is the capital of Nicaragua it does not express in addition that (3) is used to assert that Managua is the capital of Nicaragua even if it should happen that (3) is in fact employed to make this assertion (10). Anyhow, even against the background of the special variant of ([T.sub.G]) which results by replacing 'for some [PHI]: s [PHI] that p' with 's expresses that p' the point of the qualification 'which goes beyond the characteristics that are to be expressed' (or: 'which is beyond the property of things that can be signified') can be made intelligible: Even in the case of some finite chain of linguistic units, such that each member explains the meaning of some other member there is at least some content of some unit concerning which it is not expressed by any member that the pertinent unit expresses that content. Although this outcome is most obvious under the hypothesis that the elements constituting the chain are free from ambiguity it can be ascertained that the consequence holds also good if that supposition is dropped. As the qualification embodied by 'called 'inherence' (samavaya)' relates to matters of the terminology of the Indian scholastic tradition, it can be disregarded in the present context of discussion.
Although it must be assumed that the writer of the SS intended to advocate claims relating to all types of meaningful linguistic units, it suffices to restrict the scope of consideration to the special case of sentences for presenting an intelligible account of the remainder of the relevant section of the SS.
Difficulties emerge if one alters (TG) into:
([T.sub.C]) For any sentence s, if s possesses c as its content, then s does not equally express any comment concerning itself (viz. s) to the effect that for some [PHI]: s [PHI] c.
Here the term 'content' functions as a convenient variant for the previously employed expression 'meaning-content', which was used as an equivalent for the original word artha-. The idea of ([T.sub.C]) can be clarified by the following re-formulations:
For all sentences s: if c is the content of s, then s does not express that s [PHI] c, for any syntactically acceptable replacement of '[PHI]'
For all sentences s: if c is the content of s, then s does not express any comment concerning itself and that content to the effect that it (=s) is related in such and such way to this content (=c).
A simpler and more convenient way of putting the point would be:
For all sentences s: s does not express anything about the way it is related to its own content.
In view of examples like (3) no reason exists to call into doubt the validity of the tenet embodied by ([T.sub.C]). The crux is, however, that ([T.sub.C]), like the previously formulated propositions ([T.sub.G]), (T) and ([T.sup.*]), represents a general claim about all sentences and that it supposedly expresses something about linguistic units which holds true of them with necessity. But such a contention appears untenable. This is vindicated by examples like:
(4) ThisT sentence (the sentence which I am uttering just now) does not express anything about the way it is related to its own content.
Here the symbol '[up arrow]' is a device signalling token-reflexivity so that 'this [up arrow]' has an effect comparable to the expression occurring in brackets (11).
But does not (4) by expressing that it does not express anything about the (meaning-)relation between itself and its content, express something about this relation, specifically about the way in which the sentence is related to its own content, and thereby invalidate ([T.sub.C])? (4) is self-refuting on account of the generality of the quantification concerning the relation between expressions and their content(s)--embodied in the expressions 'any' or 'anything'. On the other hand, ([T.sub.C]) is endangered by inconsistency due to the generality of the quantification concerning sentences--or linguistic items in general. This generates the following dilemma:
(1) If the quantification is absolutely unrestricted it must concern the formulation representing ([T.sub.C]) itself. Accordingly a proposition ought hold true which is expressed by replacing '([T.sub.C])' for 's' and which amounts to:
([T.sub.C]') ([T.sub.C]) does not express anything about the way it is related to its own content.
But is ([T.sub.C]') not refuted by the content of ([T.sub.C])? If one supposes that the quantification concerning linguistic items is not restricted, so that ([T.sub.C]) can only be true if ([T.sub.C]') holds true, it communicates something about the way in which it is related to its content, namely the fact that the way in which it is related to its own content is not expressed by it. It might be objected that this circumstance must not refute ([T.sub.C]) because the relevant sense of 'communicate' is different from the import of the first occurrence of 'express' in ([T.sub.C]). But this would merely underscore the fact that a difficulty exists because only an explication of the pertinent sense of 'express' or of the nature of the pertinent meaning-relation could offer a way out of the impasse into which the pertinent assumption regarding the character of the quantification leads.
(2) If the quantification is taken to be restricted in a way which implies that the formulations representing the relevant theorems are exempt from its domain then they cannot communicate that which they should communicate. For it has been assumed that the point concerns linguistic items in general and that the statement of verse (19) ought convey something which holds necessarily true of all meaningful expressions and of all declarative sentences in particular.
We suppose that exactly this dilemma is addressed in the verses (20) and (21) of the SS. It is commonly acknowledged in the tradition of exegesis that those stanzas represent an objection. Against the background of the preceding exposition one can specify the character of the objection as one which is specifically directed against the contention expressed in the immediately preceding verse. Whereas verse 20 corresponds to the first member of the dilemma, the subsequent verse addresses the second alternative and points out that the underlying hypothesis is incompatible with the actual intentions of the statement of verse (19) by saying that the very situation or state-of-affairs which is to be expressed, i.e., which ought be expressed, would not be ascertained, i.e., would not be identified as the intended import (vivaksitasya yavastha saiva nadhyavasiyate).
The remarks of the subsequent verses harmonize with the suggested analysis. One can plausibly interpret verse 22 as embodying the beginning of the reply against the criticism pronounced in SS 20-21. Verse 22 does not offer any counterargument, but merely affirms that the words employed in verse (19) do not counteract the state of not being an ingredient of the expressed content as far as the meaning relation between the words and their actually expressed content is concerned. That means that it is not the case that by uttering the sequence of words occurring in verse (19) --or in similar statements to the same effect--a situation is created in which that which ought remain unexpressed according to the pronounced tenet is in fact expressed. To be sure, sentences which linguistically express that it holds true for all linguistic items that the fact that they express their content is not an ingredient of their content, possess an expressed content--namely precisely the proposition that it holds true for all linguistic items that the fact that they express their content is not an ingredient of their content. But the author of the text intends to assure us that this does not necessitate the consequence that even the fact that they express that propositional content is part of their content (12). There is no need to conceal the circumstance that the formulation of verse 23 leaves room for alternative interpretations and that certain details cannot be easily settled, given that the remaining uncertainties do not affect the main results of the ongoing analysis (13).
It is pretty obvious that none of the theorems considered before could be maintained if an iteration-principle were valid, which reads:
If x [PHI] that [PHI] then x [PHI] that x [PHI] that p
If such a principle is applied to sentences and their expressed content, a consequence results which is represented by the formulation:
(For every sentence s:) If s expresses that p then s expresses that it (=s) expresses that p.
In this light the remark of verse 23 can be understood as a means of clarifying that the central tenet of the writer of the SS is not endangered by such a principle. The fact that whenever somebody doubts that something is the case it does not follow that he equally doubts that he possesses this doubt, destroys the basis for refuting the author's contention in this manner.
On the other hand, the assumption that iteration occurs is not inconsistent. If somebody asserts that such and such is the case he may (presumably at a later time) also assert that he asserts or has asserted that such and such is the case. Verse 24 accounts for this possibility and indirectly indicates why this possibility does not disprove the thesis. For the remark that a cognition does not persist in its own character (svadharme navatisthate) if an ascertainment is ascertained as an ascertainment can be taken as a manifestation of the insight that in any pair of ascertainment described by sentences of the form
x ascertains that p
x ascertains that x ascertains that p
the character of the concerned acts of ascertainment necessarily differs because they differ regarding their content--given that the content of (replacements of) 'p' is the same in both cases. Transferring this situation to pairs of sentences of the form:
s expresses that p
s expresses that s expresses that p
one can ascertain that the supposition that 's' refers to the same item in both cases necessitates (under the same presuppositions regarding replacements of 'p') the consequence that it expresses (at least) two different contents. Accordingly an embedding of the latter phrase, viz. 's expresses that s expresses that p', in 's expresses that' would entail that the item possesses an additional third content, and so on. The point can even be generalized, envisaging a replacement of the verb 'expresses' by a variable, i.e., considering pairs of the form:
s expresses that p
s expresses that s [PHI] that p
as well as the following pair where '[PHI]' is replaced by the negated correlate of 'express', viz.
s expresses that p
s expresses that s does not express that p.
This indicates the crucial reason for denying that linguistic expressions expressing some content also express that they express or do not express it. Every comment about an expression's expressive capacities generates some content which differs from the one which is expressed in the first place. Thus the hypothesis that for all contents which an expression possesses also the fact of expressing or not expressing that content is expressed by the pertinent linguistic unit entails the existence of an infinite amount of expressed contents as a consequence. Hence the hypothesis deserves to be rejected if one does not want to accept such an infinite series.
To be sure, a solution of the difficulty raised in the objection is not provided thereby. But if we follow a generally shared assumption of the exegetical tradition the proponent's reply does not end at this point. Now the attention must be directed to the question whether the above considered theorems
([T.sub.G]) For any sentence s, if s expresses that p, then s does not equally express any comment concerning itself (viz. s) to the effect that for some [PHI]: s [PHI] that p.
([T.sub.C]) For any sentence s, if s possesses c as its content, then s does not equally express any comment concerning itself (viz. s) to the effect that for some [PHI]: s [PHI] c.
amount to the same. It can be ascertained that the problem which sentences like (4) present to (TC) on the basis of a reading saying that no sentence can express anything about a manner in which the sentence is related to its own content does not affect ([T.sub.G]). For given that (4) expresses that it (i.e., (4) itself) does not express anything about the way it is related to its own content it can be plausibly maintained that (4) does not express in addition that it expresses that it does not express anything about the way it is related to its own content, and the same verdict appears appropriate regarding any proposition expressed by a formulation in which the word 'expresses' in the preceding phrase were replaced by some other corresponding verb or verbal phrase. There is, after all not the slightest reason to attribute to (4) any ambiguity of this sort.
For a proper understanding of the situation it is needed to draw a distinction between two types of referring to a content which could be called 'content-representing reference' or short 'representing reference' and 'non-representing reference' respectively. The first variety is exemplified by using expressions of the form 'that p'--or more technically 'the proposition that p'. Whenever 'p' is replaced by meaningful sentences the resulting expression reveals what the constituent sentence means. Evidently, if one attributes to the item purportedly referred to by an expression of the form '(the proposition) that p' some property, such as the property of being asserted by such and such person or the property of being expressed by such and such sentence, by using some expression of the form 'F that p'--'It is asserted by Descartes that there is a God', '(3) expresses that Managua is the capital of Nicaragua' etc.--the resulting sentences are bound to possess a meaning and possess a content which differ from the meaning and the content of the embedded sentences. It seems hardly deniable that the author of the SS recognized this fact and emphasized the point in the verses 23-24, even if he referred to it in terms of mental acts and their content-objects and not in terms of linguistic items and their meaning-objects. The circumstance that in view of pairs of 'p' and 'It is true that p' it is not certain that (substitutes of) 'p' and corresponding expressions of the form 'F that p' exhibit always a difference of content does not present a relevant problem in the present connection. For first, the content-identity between 'p' and 'It is true that p' has been advocated by a number of philosophers, but was also denied by others. Second, and this is more important, the writer of the SS could legitimately ignore such cases because, as long as the generality of the difference holds good in the realm of cases in which relational properties are attributed to contents, his central concern is not imperilled.
The second variety of reference is exemplified by using expressions of the form 'the content of s', 'that which s expresses' and others. It is distinguished by the circumstance that those expressions do not by themselves reveal the identity of a content or proposition that is referred to.
A crucial feature of devices of the second type is that they can be used for referring to precisely that content--and mutatis mutandis to any other quality--which a linguistic item possesses by which some property is attributed to it. The most straightforward, but not the only, tool for accomplishing such an effect provide functional expressions of the form 'the F of the sentence which I am uttering just now', where the constituent 'the sentence which I am uttering just now' could alternatively relate to the token which is being produced or to its corresponding linguistic type. With this help it is possible to construe various 'self-defeating' utterances, such as:
(5) Nothing is asserted by me concerning the sentence which I am uttering just now.
(6) The sentence number (6) in the present article expresses absolutely nothing with respect to its content.
(7) Absolutely nothing is ever stated about the relation between this [up arrow] sentence and its content.
etc. The reason of the ensuing inconsistencies is akin to an inconsistency that can be observed as regards linguistic units which do not exhibit any self-reference, as for example:
(8) Absolutely nothing has ever been said about Managua.
If the author of the SS had uttered or written any item like (5), (6) or (7) he could be definitely accused of making an inconsistent assertion in view of the fact that he acknowledges the existence of linguistic items, their contents or meaning-objects and relations between them. But nothing of the sort of those sentences is formulated in the preceding verses.
Regarding the critical statement of verse 19 it can be easily ascertained that its formulation does not contain any presenting reference to its own content. It could be argued, nevertheless, that it involves a 'vague reference' to its own meaning relation due to the intended generality of the statement, which involves a universal quantification concerning such relations. But given that the claim envisaged by SS 19 implies merely that no linguistic item expresses anything about its own content by including a presenting reference to that content, the objection turns out to be completely irrelevant as long as it can be maintained that a universal quantification does not entail a presenting reference to the items which constitute the quantifier's domain.
Precisely this point is illustrated by the following famous stanza, which has been regarded by the tradition of exegesis as addressing the 'liar-paradox'--and which is usually translated as if it should present a truth-paradox. The segment sarvam mithya bravimi is rendered by various authors as expressing a speaker's statement to the effect that everything which he says is false (14). In fact, however, the formulation linguistically permits a reading according to which a speaker asserts about himself that he never speaks truthfully, which must not entail that everything which he says is untrue. It might be argued that the interpretation adopted by those authors ought be accepted because it is linguistically permissible. But this is not a valid contention. If one encounters a formulation which linguistically allows for different readings it is not legitimate to adopt one of the alternatives without argument and disregard the others. The occurrence of an equivocal formulation could, but need not derive from the fact that an author was careless regarding his wording or that he failed to obtain full clarity about a pertinent matter. It might rather indicate that a remark should possess a point which is not affected by any particular choice among linguistically admissible alternatives. The following account not only complies with the demand, but would be plausible even if it were supposed that an interpretation in the sense of 'Everything which I say is false' represents the only acceptable reading of the phrase sarvam mithya bravimi (15).
If by virtue of its meaning a universal quantification over contents of linguistic items entailed a representing reference to all members of the domain--if it were e.g. identical to a conjunction of sentences, each of which refers in this way to some member of the domain and such that no member is not referred to by some sentence--the consequence would be unavoidable that it involves a representing reference to its own content. But sentences, such as:
(9) Everything which I say is untrue.
(10) Everything which I say is regarded by me as untrue.
vindicate that this cannot be the correct account of universal quantifications (16). The reason is that if the meaning of universal quantifications concerning contents of linguistic objects were tantamount to conjunctions implying a representing reference to the content of the sentence expressing the quantification itself, then statements made by employing the above quoted sentences were bound to be self-defeating. But evidently this is not the case. Both (9) and (10) can be employed in a manner which does not communicate that the content of (9) is untrue or that (10) has been uttered with an insincere intention. The remark occurring in the second half of verse 25 to the effect that if the expression sarvam mithya bravimi or the statement made by it would be declared as untrue or untruthful then the intended object or the intended purpose would not be understood or attained (prakranto 'rtho na gamyate) (17), indicates that it is not bound to be self-contradictory since the sentence can be employed in a manner which does not frustrate its intended import or purpose. Possibly verse 25 provides a pertinent example for the dangerousness of interpreting a textual passage in the light of a cultural tradition foreign to it due to the fact that one and the same phenomenon can illustrate different points. Given that the author of the text intended to demonstrate a fact about universal quantification and related means of expressing generality the issue of threat of paradox is irrelevant.
It might be conceded that by exempting sentences like (9) or (10) and their content from the range of the quantification one is able to safeguard that inconsistencies do not originate. But it would be faulty to conclude from this that Bhartrhari recommended quantifier restriction as a means of avoiding inconsistency or even paradox. After all, SS 25 does not provide any indication of this. It could seem that the subsequent stanza 26 communicates this idea by the remark that something which is operative in the form of a means of expression (vacaka-) is not to be expressed (vacya-) and that an item by which something else is communicated is not to be communicated there. However, a closer analysis reveals that this conclusion is unwarranted. First, it is uncertain that this remark should express any kind of recommendation or indicate a possible way of solving a problem. Against the present background it can be understood as a mere statement of a fact vindicated by the preceding observation, namely the fact that as far as universal quantifications and general statements are concerned they do not contain a reference to themselves or their content as an essential ingredient. Second, even if one accepted the supposition that the thought which the writer intended to impart by SS 26 involves some non-explicit modal ingredient of necessity, the formulation does not dictate any particular verdict concerning the scope of the modal operator. To be sure, one could interpret the import of the remark as equivalent to:
It is necessarily the case that anything which is operative as a means of expression is not something to be expressed and it is necessarily the case that anything by which something else is communicated is not to be communicated in the same context.
However, the following alternative explanation which reads:
Something which is operative as a means of expression is not necessarily something to be expressed and something by which something else is communicated is not necessarily to be communicated in the same context.
Is not less natural from a linguistic point of view. Thus the point could be that whenever a linguistic item is used as a means of expressing something it is not eo ipso something about which something is said (18) and that some content by which something else, e.g. a fact about a person or some other object, is made known to a hearer is not eo ipso something about which something should be made known (19). Third, the point of the occurrence of the expression 'there' (laird) can be seen as an indication to the effect that an entity which ought not be referred to by a sentence containing a universal quantifier or in a statement of a general nature might nevertheless be an object of reference, possibly even of representing reference, in other contexts, viz. other expressions communicating certain facts about the former expressions or their contents.
It ought be fairly obvious, however, that the remarks of the verses 22-26, even if they are taken together, do not provide a definite solution of the problem presented in the immediately preceding two verses 20-21. After all, by showing that making a general statement without restrictions in the domain of some particular category does not need to entail an inconsistency it is not shown that the particular claim pronounced in verse 19 does not imply an inconsistency. Moreover, unlike in the example of verse 25 no linguistic element with the meaning of 'all' or 'every' appears in SS 19. Therefore it is appropriate to make some further comment on the matter. One can suppose that the subsequent stanza satisfies the need of additional clarification: If one assertively utters the sentence
An assertion is (by itself) not probative.
one does not refer to the assertion which is being made in particular and attributes a property to it (20). The situation is alike in the case of the assertion embodied in verse 19. It represents a general statement without an explicit universal quantifier, using instead a noun-phrase corresponding to an indefinite noun-phrase in English and some other Western languages. It does not refer to itself in particular and just as in the example--where it is not explicitly said that the assertion that an assertion is not probative is not probative--does not specify its nature in a representing reference. However, by admitting that nothing true is asserted regarding some item in particular one is not committed to deny that the quality which is attributed to some type of objects in general is true of all objects of the pertinent type (21). Therefore it can be maintained with respect to the formulation occurring in verse 19 that, like other sentences, it does not say of itself that it has the content which it actually possesses. Thus the objection voiced in the verses 20 and 21 must be regarded as based on a misconception about the real character of the theorem presented in verse 19, and the threat of inconsistency is removed.
Verse 28 furnishes a generalization of the crucial point: An act of linguistic expression does not (automatically) involve an additional act of commenting on that act or on an expressed content, in particular a comment to the effect that such and such content is expressed by such and such an item. Hence a regress of expression or, with respect to assertions like the one which is made in verse 19, an internal inconsistency can never (sarvaira na) arise. Given that not only the formulation of the theorem of SS 19 entails self-defeating consequences but that its content is also true, the conclusion can be drawn which is pronounced in stanza 29.
According to the expounded analysis the aim of the entire section beginning with verse 20 and ending with verse 29 lies in safeguarding the consistency of the tenet proclaimed in SS 19, foreshadowed in verse 4 of the same chapter, and of its formulation. On this condition there is no compelling reason to suppose that the writer of the text had any intention to deal with some sort of liar- or truth-paradox or to even offer a solution to the problem. As a matter of fact, the textual source does not even present an exposition of the nature of some paradox of truth or lying. The remark that a frustration of communicative intentions can occur is so vague and unspecific that the reproach would be appropriate that the writer of the text treated an important problem in a cursory and superficial manner if it was his intention to deal with it at all. The basis of this criticism would however collapse if the assumption represented by the preceding conditional clause were rejected. The account presented here suggests a dismissal of that supposition (22).
It might be tempting to object that the preceding exposition refers to distinctions and notions which are not traceable in the investigated source itself. This demurral is unwarranted. It might be in fact true that, for example, the concept of presenting reference lay beyond the conceptual resources of the author who wrote the text (23). But even if this were the case it can be plausibly assumed that the writer of the work possessed a capacity of theoretical intuitions which was strong enough to sharply discern the nature of the subject matter which lay in the focus of his concern.
Even if the thesis is true that the concern of the writer in the textual passage containing an example evoking associations with liar- or truth-paradoxes lies in some different issue it does not follow that the pertinent section of the SS cannot contain ingredients which either throw some light on the nature of truth-paradoxes or indicate a promising way for some solution of the problem.
It will be attempted to demonstrate below that at most the first member of the disjunction, viz. that certain ingredients throw light on the nature of certain paradoxes of truth, is tenable.
It emerged that a particular sort of inconsistency can arise due to the circumstance that some item is characterized as exhibiting a certain property which it cannot exhibit precisely because it is characterized in that way. The occurrence of this phenomenon is not confined to cases in which contents of linguistic items are objects of attribution. It is observable in connection with different sorts of linguistic objects, e.g., with linguistic tokens as exemplified by:
(11) The sentence which I am uttering now is a grammatically well-formed sentence of Classical Tibetan.
The same content which is expressed in (11) can be expressed in a way which does not lead to any inconsistency just by employing Classical Tibetan for its formulation. In the realm of attributions of properties to content inconsistencies of this sort can only arise if non-presenting forms of reference to a content are employed because otherwise the referring term cannot refer to the content of the unit which is used for expressing the pertinent attribution. Hence it can be surmised that paradoxes of truth do not arise as long as only representing forms of reference to contents are employed. On the other hand it is certainly not true that employment of non-presenting references to a sentence's own content is a necessary condition for generating paradoxes because such inconsistencies can be also created by employing expressions which do not contain any reference to themselves at all, for example by referring to a different linguistic item which in its turn refers to the original item. Nevertheless, the pertinent inconsistencies arise only if the reference to the content of an item is accomplished in an indirect way by implying a description which does itself not reveal what the content exactly is. In a similar manner certain antinomies of set theory arise not by specifying the quality of a set directly by enumeration of its members but indirectly by specifying qualities of the members which it contains (24). No need exists to pursue the issue further in the present context because only the following fact matters here: Even if one grants, maybe only for the sake of argument, that non-presenting reference to content is a characteristic or even indispensable feature of items that engender paradoxes of truth it would be at best a gross exaggeration to contend that the investigated passage of the SS offers us 'an interesting and elegant solution to paradoxes of truth' or to 'the family of paradoxes to which the Liar Paradox belongs'.
To be sure, in view of the remark pronounced in verse 26 of the SS, one could surmise that a possible solution is insinuated to the effect that statements pertaining to linguistic items or features connected with them must always exempt the statement itself, the items used for formulating the statement and various other entities intrinsically connected with the formulation, in particular its expressed content or relations between the formulation and its content(s). This would specifically entail a general prohibition of linguistic self-reference and a demand of restricting (explicit or implicit) general quantifiers. As a matter of fact, this suggestion possesses little value.
As far as the crucial example of making an assertion by uttering the words sarvam mithya bravimi is concerned, a restriction of the universal quantifier is pretty useless as a means of 'solving' a problem of paradox. For the problem arises because the universal quantification can be employed without being restricted. Hence the suggestion that the same expression can be equally employed in some other way does not offer any help. If a linguistic item possesses at least one reading on the basis of which unbelievable consequences follow from plausible premises and plausible rules of derivation then the existence of additional readings is irrelevant. Only if it could be regarded as absolutely impossible that a pertinent item possesses the crucial reading the problem could be dismissed as insignificant. However, as long as exhibition of the problematic reading is a possibility the problem must be taken seriously. For the mere supposition that something meaningful can be expressed in language which apparently can neither exhibit truth nor lack of truth is disconcerting.
Moreover, a general prohibition of linguistic self-reference is hardly commendable. Why should one disclaim that truths are expressed by sentences, such as:
(12) The sentence which I am uttering just now is a grammatically well-formed sentence of English.
(13) Everything which was, which is or which will be ever asserted by me is asserted after the origination of life on earth.
Supposed that the author of the SS spoke during his lifetime only Sanskrit or some variety of Prakrit, would he not assert a truth if he uttered in Sanskrit a sentence possessing the same content as the English sentence:
(14) Everything which is asserting by me in my lifetime is asserted by using either Sanskrit or some variety of Prakrit.
Examples of indirect self-reference support this conclusion. Let us suppose that A utters the sentence:
(15) The next sentence which will be uttered in this room will be uttered by B and expresses a truth.
Now, B in fact utters immediately after this the sentence:
(16) The sentence which has been uttered immediately before in this room has been uttered by A and expresses a truth.
Why should we deny that both (15) and (16) are used for the expression of truths in the supposed situation? Why should we yield to the command that self-reference must be avoided at all costs? If verse 26 of the SS did not express a demand but a statement of fact to the effect that something which is expressing is never at the same time an object of comment then it could be only regarded as a bold contention which is presented without any argument. It had emerged, however, that the thesis advocated in verse 19 does not commit its proponent to prohibit linguistic self-reference and self-reference with respect to expressed contents in general. Its truth only implies that making comments about a a content expressed by a linguistic item by employing a particular variety of self-reference is impossible. Moreover, a closer analysis reveals that verse 26 does not even provide a compelling indication to the effect that the writer of the text believed that linguistic self-reference is impossible.
For vindicating the contention that the investigated segment of the SS is far from offering a solution to paradoxes of truth it is apposite to consider certain consequences which can be reasonably derived from their existence. The first consequence to be suggested is that the notions of meaning and content deserve to be kept apart. Accordingly with respect to sentences one ought avoid a general equation of their meaning with the content expressed by them. This thesis is pivotal but fairly unpretentious. The demand to distinguish between the notions of thought content and sentence-meaning is by no means a novelty. For example in Dummet (1981) the relevance of a distinction between sense (as conceived by Frege) and linguistic meaning or conventional significance is emphasized at various places (25). But the tenet that truth or lack of truth are never attributable to conventional linguistic meanings of declarative sentences, except perhaps in a secondary transferred sense, is not a triviality. It has been contended that, as long as only sentences involving no significant demonstrative or indexical features are considered, one may equate linguistic meaning and sense or thought content as something which is true or untrue absolutely, not relatively to a person or time etc. 26. In contrast we suggest that the difference between conventional meaning and thought content is pertinent even in the domain of declarative sentences which do not exhibit deictic constituents, that generally declarative sentences possess a content which categorically always and materially sometimes differs from their linguistic meaning. In the realm of declarative sentences without indexical elements this content can be calculated according to the formula:
p and this, namely that p, is ... the case ...
Here 'p' is replaceable by an expression exhibiting the linguistic meaning of a pertinent (declarative) sentence and the gaps indicated by '...' have to be replaced in accordance of whether or not the sentence implies a self-comment relating to truth as well as the nature of the truth-comment which it involves, if it involves this. In cases in which a sentence expresses or implies lack of truth of its content 'is ... the case ...' ought be replaced by 'is not the case'. If it expresses or implies with respect to itself that that which it expresses is true if and only if Managua is the capital of Costa Rica, then 'is ... the case ...' has to be replaced by 'is the case if and only if Managua is the capital of Costa Rica', and so on. Whenever a sentence neither expresses nor entails any self-comment its content is represented by
p and this, namely that p, is the case.
Accordingly it is permissible to replace the gaps by zero, and it ought be replaced in this way if a sentence does not imply any comment concerning the question of its own truth. In those cases the thought content which is evaluable of being categorically true or not is logically equivalent to the content represented by the conventional meaning of the sentence--or its meanings, if the sentence is linguistically ambiguous. This suggestion is compatible with the assumption that the linguistic meaning is never assessable as being true or untrue, at least not in a manner which does not rely on transference or metaphor. On that condition a considerable number of declarative and not ambiguous sentences are related to different contents which differ with respect to type but coincide materially. Among those however only one is (genuinely) assessable as a bearer of truth or lack of truth.
Without considering truth-comments in general it might be more difficult to attribute plausibility to the outlined account. Therefore it is appropriate to look at truth-comments which do not imply any sort of self-reference. Here the term 'truth-comment' should designate any item which --directly or indirectly--attributes exhibition of truth or lack of truth to some item, which by its nature permits attributions of this sort. Thus in the context of the following pairs of sentences:
(17) Managua is the capital of Costa Rica.
(18) This is not true.
(19) That which (17) expresses is not true.
Both (18) and (19) can be classified as embodying truth-comments in the relevant sense. In the present content it suffices to rely on this intuitive explication.
Granted that truth-comments can exhibit the properties of being true or untrue--or, if the linguistic vehicles are considered, of expressing something true or failing to do so--it can be easily assessed that their literal import only partly determines conditions of their truth. As far as (18) is concerned this kind of incompleteness is evident due to the occurrence of an indexical element. Even if 'this' is taken not to refer to a linguistic unit, in particular a sentence, but to the content of a sentence (type or token), (18) is assessable under the aspect of its truth or lack of truth only if a particular content is specified. Given that (18) occurs in the context of (17) the most natural explication could be represented by the formulation:
This, namely that Managua is the capital of Costa Rica, is not true.
On the other hand, in the case of (19) the literal content permits itself the identification of a necessary requirement of truth, namely that the content which the item designated by '(17)' expresses is not a truth. That assessment can be safely made even without taking into consideration what (17) actually expresses (27). However, a definite verdict about questions of truth cannot be made on this basis alone. It is additionally required to recognize what the pertinent item in fact expresses. In the case under consideration it is the fact that (17) expresses that Managua is the capital of Costa Rica. Accordingly a fully explicit specification of the condition of the truth of (19) would be obtained by a conjunction, where the second member coincides with the explication given for (18), namely:
That which (17) expresses is not true and this, namely that Managua is the capital of Costa Rica, is not true.
--The cumbersome character of the formulation is immaterial as long as it represents the requirements in a correct manner, and it could be, if desired, replaced by stylistically more attractive wordings 28.
The thought itself that necessary requirements for truth are ingredients of sentence-content is not unnatural. Intuitively it would not imply a substantial change of content if, say
(20) In Siberia larches are common.
were replaced by:
(21) In Siberia trees which are larches are common.
The main difference between the pair of (20)(21) and the pair of (19) and its explication is that the latter one contains an ingredient which cannot be read off from its literal import. However, this merely highlights the affinity between truth-comments and sentences containing indexical elements even regarding specimens of truth-comment which do not explicitly contain indexicals. As long as truth-comments neither contain deictic elements nor imply self-reference the specification of conditions of truth need not exceed their conventional linguistic meaning because in such cases linguistic meaning reveals a content which holds true exactly if the requirements imposed on the contents of the items or items which are (direct) objects of comment are satisfied (29). It is, however, important to recognize that this circumstance does not compellingly entail that in those cases linguistic meaning is the content which exhibits the specified property. Specimens involving self-reference insinuate that linguistic contents do not always possess the feature that expressing a correct ascription guarantees truth.
Let us suppose that '(S1)' and '(S2)' are designations of the sentences
That which (S2) expresses is not true.
That which (S1) expresses is true.
respectively. None of the considered sentences contains an indexical expression or a designation of itself as a component. By analogy with the previously considered example it is to be ascertained as a truth-requirement for (S2):
That which (S1) expresses is true and this (what (S1) expresses), namely that that which (S2) expresses is not true, is true.
This can be--granted that 'it is true that p' and 'p' are equivalent--replaced by:
That which (S1) expresses is true and that which (S2) expresses is not true.
By iteration one obtains:
That which (S1) expresses is true and this (what (S2) expresses), namely that that which (S1) expresses is true, is not true.
which can be taken as equivalent to
That which (S1) expresses is true and it is not the case that that which (S1) expresses is true.
and this amounts to:
That which (S1) expresses is true and that which (S1) expresses is not true.
Since the formulation reveals a patent contradiction, (S2) must be credited with a contradictory import and accordingly as untrue or false if it is supposed that the above presented explication represents the content which matters for truth-assessment. This would imply that the content of (S2) needs to be regarded as untrue no matter whether or not (S1) expresses a truth. With respect to the envisaged situation one might, however, surmise that (S2) in fact equally fails to express a truth.
The presented example is biased, however, because it conceals the fact that reluctance of ascribing expression of truth to an item could also be based on the consideration that it lacks any content which is capable of being true or that the entity to which it is linked by the relation of expressing lacks the capability of being true in principle. Hence by replacing the original formulation of (S1) by
(S1)* (S2) does not express a truth.
--and accordingly replacing '(S1)' by '(S1)*' in (S2)--room is provided for the possibility of assessing the unit to which the comment of (S2) pertains as expressing a truth without accrediting the same quality to (S2)30. The exposition can serve as a model for evaluating examples of direct truth denial, as illustrated by:
(S0) That which (S0) expresses is not true. as well as situations depicted by:
(S[1.sub.M]) That which (S[2.sub.M]) expresses is true if and only if Managua is the capital of Costa Rica.
(S[2.sub.M]) That which (S[1.sub.M]) expresses is true. and various others.
A consequence of the portrayed view is that linguistically synonymous tokens containing no indexical ingredients can differ with respect to truth-value of content. If '(S3)' were the designation of a linguistic token which reads:
That which (S1) expresses is a truth.
no inconsistency results, if '(S1)' in (S3) referred to (S1) in the envisaged context--or to a token which reads like (S1), where '(S2)' refers to a token which reads like (S2)--provided that the tokens designated by '(S2)' and '(S3)' differ. Against the background outlined above this circumstance ought be attributed to the fact that (S2) and (S3) differ regarding the content relevant for attribution or denial of truth because the latter item does not imply a truth-comment on itself. In so far as linguistic synonymy harmonizes with difference of truth-value the pair of (S2) and (S3) resemble the pair consisting of (11) and its possible equivalent in Classical Tibetan, although there is no reason to explain the deviance by any difference of content in the latter case. The difference between (S2) and (S3) could be characterized by saying that (S3), if it expresses a truth, represents an external comment in contrast to (S2) which is itself a factor for generating an inconsistency (31). This difference must be induced by factors lying beyond linguistic meaning.
Consideration of some consequences serves for clarifying the nature of the outlined suggestions:
1) Asserted content is not the same as that which someone intends to assert.
In EL pp. 55-56 the following sequence of derivational steps is presented:
1 (1) K[[for all]x (Kx [right arrow] Fx)] Assumption 2 (2) W[[for all]x (Kx [right arrow] Fx)] Assumption 2 (3) [for all]x (Kx [right arrow] Fx) From (line) 2, by W-elimination 2 (4) K[[for all]x (Kx [right arrow] Fx)] From 3, by [right arrow] F[[for all]x [for all] (Kx [right arrow] Fx)] -elimination 1,2 (5) F[[for all]x (Kx [right arrow] Fx)] From 2 and 4, by Modus Ponens 1 (6) W[[for all]x (Kx [right arrow] Fx)] From 1 and 5 by [right arrow] F[[for all]x [right arrow] (Kx [right arrow] Fx)] Introduction 1 (7) W[[for all]x (Kx [right arrow] Fx)] From 6 by [right arrow] [logical not] substitution F/ W[[for all]x (Kx [right arrow] [logical not] W Fx)] 1 (8) [logical not] W[[for all]x From 7 by (Kx [right arrow] Fx)] Consequentia Mirabilis 9 (9) W[[for all]x (Kx [right arrow] Fx)] Assumption v F[[for all]x (Kx [right arrow] Fx)] 1,9 (10) F[[for all]x (Kx [right arrow] Fx)] From 8 and 9, by Disjunctive Syllogism 9 (11) K[[for all]x (Kx [right arrow] Fx)] From 1 and 10, by [right arrow] F[[for all]x [right arrow] (Kx [right arrow] Fx)] Introduction
The numbers on the utmost left, refer to the assumptions on which the corresponding line depends. The intended reading of 'W' corresponds to 'it is true (that)', of 'F' to 'it is false (that)'. This suffices for understanding that the step in line 3 refers to the theorem that 'It is true that p' entails 'p', the step of line 7 to the tenet that 'It is false that p' entails 'It is not true that p' and the step of line 8 to the rule that if from an assumption a contradiction is derivable, then the negation of the pertinent assumption can be deduced. The remaining steps rely on familiar rules of derivation. The envisaged reading of the symbol 'K' is 'It is asserted by a Cretan (that)' (32). It is, however, easy to see that the question of the validity of the derivation does not depend at all on this supposition. If the conclusion of line 8 validly follows from the assumption of line 1, and if the conclusion of line 11 validly follows from the assumption of line 9, 'K' could be also interpreted in any other way, which does not imply a violation of syntactic and semantic rules, for example as 'Kurt asserts (that)', 'Kurt denies (that)', 'Kurt believes (that)' etc. Since, presupposing the intended interpretation of 'K', the expression occurring in the final line (11) says that if a(ny) Cretan asserts that everything which a(ny) Cretan asserts is false then it is false that everything which a(ny) Cretan asserts is false, a disconcerting fact can be ascertained: From the circumstance that somebody, e.g., the Cretan Epimenides has asserted something, a different historical fact appears to follow, namely that something else which is not false has been asserted by a Cretan. After all, in view of line 8 it cannot be supposed that the pertinent statement of the Cretan is true. To be sure, nothing of the sort of an antinomy results in the present case. But, as pointed out in EL (p. 57), the consequence can be regarded as paradoxical if one considers the assumption of line 1 as involving an assertion of a possibility: It militates against common standards of credibility to assume that the situation that a Cretan asserts that everything which a Cretan asserts is false can only occur if there is something else asserted by a Cretan which is not false. Mutatis mutandis it should be absolutely impossible that somebody asserts something like:
Everything which is asserted by me today at 12.30 p.m. is untrue
at the specified time supposing that nothing else is asserted by the same person at the same time.
However, in view of the preceding exposition it is legitimate to ask: Is it really sure that the situation that a Cretan asserts that everything asserted by a Cretan is false (or untrue)--without asserting anything more--represents a possibility? This is far from incontestable as long as one supposes that asserting something requires the existence of a content which is (by its nature) capable of being true. If the above suggested view were accepted a person who asserts with respect to himself that everything (ever) asserted by him is untrue can in fact assert a content which is capable of being true, but this content is not directly manifested by the linguistic formulation--e.g., 'Everything ever asserted by me is untrue'--but would be represented by a more complex structure, such as 'Everything ever asserted by me is untrue and this, namely that everything ever asserted by me is untrue, is untrue'. This content, which embodies a contradiction, must be assessed as untrue or false. Thus the above described paradox does not arise as before because the central assumption as to what is possible to assert has been dropped. Thereby a motivation is removed to question any of the logical rules employed in the portrayed derivation or to assail the assumed connections between 'It is true that p' and 'p' or 'it is false that' and 'it is untrue that'.
A possibility which can be hardly denied is that somebody by using a formulation can express that everything which he himself ever expresses (or which any sentences used by him express) is untrue. In this situation, as Bhartihari assures us, we have to presume that the employed formulation does not express that that which the actual content of his formulation expresses is untrue, that, in other words, the formulation expresses that it is untrue that everything which he (the speaker) himself ever expresses (or which any sentences used by him express) is untrue. But precisely for this reason it ought be supposed that the vague notion of expressing something does not determine a relation that holds good between meaningful linguistic items and a content which exhibits the quality of absolute (not relative) truth. For precisely this proposition must constitute a necessary requirement of truth if the quantifier is not restricted, and a complete specification of truth-requirements demands a presenting reference to the sentence's own content. The diagnosis is corroborated by the circumstance that common linguistic usage and intuitions license the statement that two or more tokens of
It has not rained today
uttered on different days, and possibly in distant places, express the same (33).
The crucial issue is whether plausibility can be attributed to the thesis that nobody can assert of himself not to assert any truth. Certain reasons apt to induce impressions of implausibility are dubitable. It is surely incontestable that somebody can utter chains of words which mean that the person who is making the utterance never asserts a truth, or something similar, and that such word-concatenations can be uttered without attributing to them any deviant import. But this would refute the tenet under consideration only if it could be taken for granted that attributable meaning is a primary bearer of (absolute) truth. As this equation lacks plausibility at least in the domain of linguistic units containing indexicals it should require a specific argument to establish that the situation is different in the pertinent case.
It would be also an invalid objection to allege that by an assertive utterance of, e.g.,
(22) All sentences uttered by me do not express anything which is true.
it is, among other things, asserted that all sentences uttered by the speaker do not express any truth. This contention involves a conflation between being a part or ingredient of an asserted content and being asserted among other things. If somebody assertively utters the sentence
(23) Managua is the capital of Nicaragua and that which is expressed in the preceding phrase is not a truth.
the speaker has not asserted by (23) the true proposition that Managua is the capital of Nicaragua because this is definitely not the content of the entire sentence. To be sure, according to a common notion it would be permissible to say, for example with respect to an assertive utterance of
(24) Managua is the capital of Nicaragua and Stockholm is the capital of Austria.
that by asserting (24) both a truth and an untruth have been asserted. But obviously this relies on a different concept of assertion than the one which denies (24) the property of expressing a truth on account of the fact that not both members of the conjunction are true. Against the background of the 'liberal' concept of asserting nothing paradoxical lies in the supposition that one and the same item can express or be used to state both a truth and an untruth.--The possibility and occurrence of this concept should hardly provide a compelling motivation for adopting 'paraconsistent' varieties of logic.
Nevertheless, the following objection deserves serious consideration: Is a doctrine realistic which implies that somebody who makes a general statement about his own assertions attributes a property to the assertion he is making? The envisaged account insinuates in particular that an assumption to the effect that some individual K asserts that everything asserted by him (K) is untrue implies that he attributes untruth to the actual content of the item by which the assertion is made. This impression is evoked by the suggestion that disclaimer of truth with respect to the linguistically expressed content is an ingredient of that which is asserted. In this connection it is of utmost importance to discern an equivocation in the notion of assertion. Turning the attention again to the example (19), by hypothesis assertively uttered in the context of (17), it emerges that common intuitions license divergent verdicts about the identity of that which is asserted by someone who assertively utters (19). One alternative lies in supposing that the proposition is asserted that that which (17) expresses is not true. According to an other alternative one could say that apart from this also the proposition that Managua is not the capital of Costa Rica is an ingredient of that which is asserted by assertively uttering (19) in the context of (17). Which of the options one inclines to prefer depends on additional circumstances. Presumably one would be inclined to accept the second alternative if one supposes that the utterer of (19) knows the meaning and content of the item to which reference is made in the utterance, whereas one would rather reject this idea if one supposes that the utterer does not possess such knowledge. However, given that the referring term '(17)' in (19) refers to a sentence expressing that Managua is the capital of Costa Rica, it is guaranteed that a truth can be asserted by (19) only if (17) fails to express a truth and Managua is not the capital of Costa Rica. Under this aspect the truth-comment of (19) behaves like a truth-comment involving an indexical ingredient, such as (18).
But the concept of asserted content which is pertinent does not reflect what persons intend or have in mind but the requirements for an assertion's being categorically true.
This concept is clearly also implicit in the above outlined derivation of EL, where it is presented as something that exhibits the properties of being true or untrue (or false). Requirements of truth for (19) encompass not merely that the content of (17), whatever it may be, is not a truth but also that it is not true that that which the commented item in fact expresses is a truth. Accordingly, whenever somebody makes an assertion to the effect that everything asserted by himself is untrue (false), then, if the range of the universal quantifier is not restricted, truth requires on the one hand that it is the case that everything asserted by the subject is untrue as well as that the content that everything asserted by the subject is untrue is untrue. Hence it requires a fortiori that it is not true that everything asserted by the subject is untrue. The circumstance, however, that something which is inconsistent entails an unwarranted proposition pertaining to history should not be surprising. It is not astounding either that an assertion by a sentence with the same linguistic meaning for making an 'external' comment need not be beset by inconsistency. Here no denial of the content which the item linguistically expresses is involved. This very fact prompts a difference with respect to requirements of truth together with a divergence regarding a type of content which does not coincide with that which a sentence conventionally expresses (34). In this case the pertinent requirements of truth are in principle satisfiable. Thus the circumstance that the disconcerting consequence regarding historical facts is not derivable any more need not surprise us.
2) Content is determined apart from conventional meaning by contextual factors different from communicative intentions.
The thesis is not meant to stipulate something concerning content in general. It should merely imply that some relevant type of content exists most intimately connected with conditions of truth for which the tenet holds good in the realm of truth-comments. This could be extended to declarative sentences in general. For the pertinent type of content the circumstance whether a linguistic unit does or does not imply a comment on its linguistically expressed import is a determining factor apart from its conventional meaning and its, possibly disambiguated, linguistically conveyed import. Against the background of the fact that determination of content relevant for truth or lack of truth by circumstances of context is widely acknowledged, the tenet represents only a moderate revision of those views by implying that even sentences (35) containing no pronouns or other indexical expressions can exhibit similar features (36). The modification appears necessary because truth-comments exhibit a kind of context-sensitivity which is not restricted to expressions containing indexical ingredients. The thesis does not preclude an acknowledgment of other types of content which are not subject to the same kind of dependence on contextual factors. However, in view of phenomena like conversational implicatures the idea that non-equivocal linguistic expressions can exhibit several varieties of content, even different varieties which are evaluable under the aspect of truth or lack of truth, is not unfamiliar.
3) The dichotomy of truth and lack of truth can be regarded as exhaustive.
The considered account permits to ascribe to items implying a denial of the truth of their own content the property of failing to express a truth and to their truth-relevant content the quality of being untrue. The view that all linguistic items, including units which are not declarative sentences, either express a (at least one) truth or fail to do so and that no third possibility exists, conforms to pre-theoretical intuitions.
4) The validity of two intuitive principles concerning saying a truth or expressing a truth is preserved by means of explication.
In EL, p. 40 two conditional principles are presented and assessed as 'intuitively very plausible' ('intuitiv sehr einleuchtend'). They read as follows:
[P1] if (x says that p and x thereby says something which is true), then p.
[P2] if (x says that p and p), then x says something which is true (37).
A similar pair of principles, referring to sentences instead of persons, is introduced in AL, p. 253. It says:
(TE1) [For any linguistic unit s:] (if s expresses that p and s thereby expresses a truth), then p.
(TE2) [For any linguistic unit s:] (if s expresses that p and p), then s expresses thereby a truth (38).
The crux is that with respect to the above quoted examples (S1) and (S2) a contradiction can be derived by taking as assumptions that they express that which they linguistically and intuitively must express relying on the rule of and-introduction and the principles (TE1) and (TE2). By virtue of the rule of reductio ad absurdum it follows that at least one of the two items cannot express, what it evidently does express. This means that either (S2) does not express that that which (S1) expresses is true or (S1) does not express that that which (S2) expresses is not true or both. A similar result is derivable for sentences which attribute untruth to themselves, or more exactly to their own linguistically expressed content, in a direct manner (39). This makes mandatory to assume that at least one of the principles (TE1) or (TE2) must be given up if the rules of and-introduction and reductio ad absurdum should not be questioned. Hence the solution has been advocated that the content which corresponds to the relation of (linguistically) expressing can not comply with both (TE1) and (TE2) and that (TE2) is particularly vulnerable. On the other hand, however, it can be shown that (TE1) and (TE2) can be considered as valid with respect to a different type of content which corresponds to the concept of content delineated above.
Due to the analogy between [P1] and (TE1) and [P2] and (TE2) the same predicament affects mutatis mutandis the pair of [P1] and [P2]. Accordingly the most pertinent question with respect to those principles is, how the word 'say' has to be understood to preserve the validity of both [P1] and [P2]. For on the basis of every reading of saying they can definitely not be valid. However, the conception of a content exhibiting a more complex structure in accordance with the previous exposition permits to specify a concept of saying which complies with [P1] as well as with [P2]. Surely, the resulting idea of saying something does not correspond to that of what a speaker has in mind when saying something. The correlating concept of content represents rather an intuitive notion of what is said by virtue of objective contextual circumstances. But the occurrence of linguistic items possessing indexical components calls for a concept of this kind anyway (40).
In this manner the intuition that the principle [P1] and [P2] possess validity can be accounted for. It is equally possible to explain why the need of elaborating more specific concepts of saying and content manifests itself in view of rather unfamiliar examples involving (direct or indirect) self-reference. The result is a replacement of the vague notion of saying something by a plurality of more definite concepts. The same holds true regarding the theorems (TE1) and (TE2) and their role for elaborating more definite concepts of expressing (41).
The considerations of the preceding chapter strengthen the thesis that the previously investigated passage of the SS does not offer any solution for paradoxes of truth or lying. This is relevant because from a purely linguistic and grammatical perspective the wording of the textual source permits often divergent interpretations. It is accordingly impossible to definitely prove the correctness of the presented analysis against the background of linguistic and other philological criteria alone. It can be at most asserted that it differs from previous interpretations by connecting the investigated text with an intelligible theoretical point and permits to understand the section of SS 1-29 as representing a highly coherent train of thoughts.
Even admitting the possibility of entirely different interpretations it can be safely maintained that the text does not address issues which one can reasonably expect in the context of discussions of paradoxes of truth or--if there are such paradoxes--paradoxes of lying. It might be illegitimate to demand that a treatment of paradoxes of truth connects its topic with that of other paradoxes, such as antinomies of set theory. It cannot even be taken for granted that any solution of truth-paradoxes provides an easy solution for other kinds of paradox. But the postulate that any hint for a solution of truth-paradoxes must, at least implicitly, account for cases of mediated self-reference, as exemplified by (S1) and (S2) appears fully appropriate. In view of the preceding considerations one is entitled to say that the topic of the SS contains a point of contact with an issue that matters for the assessment of truth-paradoxes. For the text brings into play the relation between linguistic items and what they mean, something which could be regarded as their meaning or semantic content. However, at least as far as the investigated passage is concerned, not the slightest indication can be found to the effect that a distinction between different relations between linguistic items and semantic objects might be called for. It might be also noted that the text does not bring the idea of context-sensitivity into play and that it nowhere alludes to possible affinities between truth-comments and sentences containing deictic ingredients.
If the author of the SS had the intention to deal with the topic of truth paradox, then it would have been appropriate for him to pay attention to the question of the nature of content related to linguistic expressions, particularly declarative sentences. Inasmuch as truth-paradoxes do not only raise questions of the nature of truth but indirectly also questions about the nature of the items which do and can exhibit the quality of being true they call for some clarification about the issue whether something that can be regarded as a content of declarative sentences is a potential bearer of a truth-property. As far as the concept of absolute truth is concerned one ought not only investigate matters from the perspective of whether or what sort of truth a given type of content can exhibit, but also the other way around under the aspect of the question of how content has to be conceived if it were something that can be absolutely true. The latter perspective was determinative in the exposition of the preceding chapter with the result that content must exhibit a more complex structure than anything which is expressed by virtue of linguistic meaning if it should be credited with a quality of absolute truth without inconsistency. To be sure, it is theoretically possible that one reaches the conclusion that the resulting conception of content is too implausible or useless and that a concept of absolute truth entailing it deserves to be rejected. But even in this case investigating the issue under the considered perspective would not lose its relevance because it clarifies the grounds for a dismissal. For the present no compelling reason for discarding absolute truth has been detected.
Aklujkar, A. and Potter, K.H. (1990). Trikandi or Vakyapadiya, with Vrtti on Books 1 and 2 (Summary and analysis of the Trikandi), in: Coward, H.G. and Kunjunni Raja, K. (ed.) Encyclopedia of Indian Philosophies. The Philosophy of the Grammarians. Princeton: Princeton University Press.
AL See Oetke, C. (2012)
Coward, H.G. and Kunjunni Raja, K. (1990) Encyclopedia of Indian Philosophies. The Philosophy of the Grammarians. Princeton: Princeton University Press.
Dummett, M.--(1981). The Interpretation of Frege's Philosophy. Cambridge, Massachusetts: Harvard University Press.
EL See Kunne W. (2013)
Frege, G. (1892). Uber Sinn und Bedeutung, Zeitschrift fur Philosophie und philosophische Kritik, vol C, 25-50.
--(1969). Nachgelassene Schriften. (ed. by Gabriel, G. et al) Hamburg.
Houben, J.E.M. (1995). The Sambandha-samuddesa (Chapter on Relation) and Bhartrhari's Philosophy of Language. Gonda Indological Studies, volume II) Groningen: Egbert Forsten.
--(1995a). Bhartrhari's Solution to the Liar and Some Other Paradoxes, Journal of Indian Philosophy 23, 381-401.
Iyer, K.A.S--(1971). The Vakyapadiya of Bhartrhari. Chapter III, pt.i. (English Translation). Poona: Deccan College.
Kunne, W. (2013). Epimenides und andere Lugner. Frankfurt am Main: Klostermann Rote Reihe. (= EL)
Oetke C. (2012). Ascription of Linguistic Properties and Varieties of Content. Two Studies on Problems of Self-Reference. (Acta Universitatis Stockholmiensis, Stockholm Oriental Studies. 20) Stockholm (= AL)
Rau. W. (2002). Bhartrharis Vakyapadiya. Versuch einer vollstandigen deutschen Ubersetzung nach der kritischen Edition der Mula-Karikas. (Akademie der Wissenschaften und der Literatur. Abhandlungen der Geistes- und sozialwissenschaftlichen Klasse. Einzelveroffentlichung, Nr. 8. Stuttgart: Franz Steiner Verlag.
Claus Oetke (firstname.lastname@example.org).
Professor emeritus of the University of Stockholm.
Orotina 20901. Costa Rica.
Received: Monday, june 30, 2014.
Approved: Thursday, july 24, 2014.
(1.) In the section on Bhartrhari in Coward/Kunjunni Raja (1990, 121) a dating between A.D. 450 and 510 is proposed.
(2.) Ascription of Linguistic Properties and Varieties of Content. Two Studies on Problems of Self-Reference. Stockholm 2012.
(3.) Frege (1892, 37)
In "A log, dass er den B gesehen habe" bedeutet der Nebensatz einen Gedanken, von dem erstens gesagt wird, dass A ihn als wahr behauptete, und zweitens, dass A von seiner Falschheit uberzeugt war.
(4.) See in particular Kunne, W. (2013, 24-33).
(5.) In the original those verses read as follows:
1 jnanam prayoktur bahyo 'rthah svarupam ca pratiyate /
sabdair uccaritais tesam sambandhah samavasthitah //
3 asyayam vacako vacya iti sasthya pratiyate /
yogafr sabdarthayos tattvam apy ato vyapadisyate //
nabhidhanam svadharmena sambandhasyasti vacakam /
19 praptim tu samavayakhyam vacyadharmativartinim /
prayokta pratipatta va na sabdair anugacchati //
20 avacyam iti yad vacyam tad avacyataya yada /
vacyam ity avasiyeta vacyam eva tada bhavet //
21 athapy avacyam ity evam na tad vacyam pratiyate /
vivaksitasya yavastha saiva nadhyavasiyate //
22 tathanyatha sarvatha ca yasyavacyatvam ucyate /
tatrapi naiva savastha taifa sabdaifa pratisidhyate //
23 na hi samsayarupe 'rthe sesatvena vyavasthite /
avyudase svarupasya samsayo 'nyafa pravartate //
24 yada ca nirnayajnane nirnayatvena nirnayafa /
prakramyate tada jnanam svadharme navatisthate //
25 sarvam mithya bravimiti naitad vakyam vivaksyate /
tasya mithyabhidhane hi prakranto 'rtho na gamyate //
26 na ca vacakarupena pravrttasyasti vacyata / pratipadyam na tat tatra yenanyat pratipadyate //
27 asadhika pratijneti neyam evabhidhiyate / yatha tathasya dharmo 'pi naiva kascit pratiyate //
28 vyaparasyaparo yasman na vyaparo 'sti kas cana /
virodham anavastham va tasmat sarvatra nasrayet //
29 indriyanam svavisayesv anadir yogyata yatha /
anadir arthair sabdanam sambandho yogyata tatha //
Since it can be safely assumed that the textual variants possess no bearing on the issues which are pertinent in the present context, they are not mentioned here.
(6.) The reason why the identification with linguistic tokens is not completely certain lies in the circumstance that the possibility cannot be ruled out that the author intended to refer by the term 'word' not to linguistic but to physical entities, such as sounds or sound-patterns, or did not attach importance to that difference in the present context. Anyhow, it appears that this question has no relevance for the issue which is at stake in the current investigation.
(7.) The attribute 'external' can, but must not imply that the objects which expressions mean belong to the 'outer world', that they are physical or non-mental objects. It might be employed to communicate that the meaning-objects are external relative to the linguistic expressions possessing a meaning. Such a statement could be motivated by the phenomenon of linguistic synonymy, by the consideration that different expressions can share one and the same meaning and are to this extent detachable from individual linguistic items. It is however far from certain that the author of the text had this fact actually in his mind.--One could also connect the remark with the idea that different linguistic tokens of a common linguistic type possess identical meanings.
(8.) This refers to the view that the sense-faculty of vision perceives visual qualities, the faculty of hearing acoustic qualities etc.
(9.) It is hard to see how those tenets can be protected against the menace of being self-defeating.
(10.) It is important to keep in mind that '[PHI]' is replaceable by complex syntagmas representing binary relations, in particular by 'x does not express that x expresses that y'.
(11.) The point could also be illustrated by a formulation like:
If c is the content of this [up arrow] sentence (the sentence which I am uttering just now) and R a relation between this [up arrow] sentence and c, then it (i.e. this [up arrow] sentence) does not express any comment about R.
(12.) Presumably this is the import of the expression tatrapi ('even with respect to that') in the verse.
(13.) This holds good in particular for the expression 'in this, in another and in every manner' (tathanyatha sarvatha ca). Notwithstanding the fact that it is not syntactically compelling to connect this constituent with the expression which means 'the quality of being something which is not to be expressed' (avacyatvam), there is on the other hand no reason to question this analysis. However, even if this is granted, the exact import is difficult to settle with certainty. Possibly the writer desired to convey that the fact concerning the relation between words and content is not part of any variety of content, alluding to a distinction between literal content and other contents which could be imparted, for example by virtue of conversational implicatures. This is not necessarily anachronistic because the notion of metaphorical or 'transferred' import, embodied in the technical term upacara, was familiar in the Indian tradition. Hence the first half of the verse could entail a clarification to the effect that meaning-relations are not ingredients of any kind of content, neither a literal nor some non-literal content-variety. It appears, however, that in the present connection it is not required to give a definite verdict about this issue.
(14.) Examples are Houben (1995), Aklujkar in Coward and Kunjunni Raja (1990, 158), Iyer (1971, 90) and (presumably) Rau (2002, 179).
(15.) In AL the demand to keep the different interpretations of the phrase clearly apart was also based on a rejection of the view that the notions of 'liar-paradox' and 'truth-paradox' are equivalent --although it was presumed that in a number of contexts the difference can be ignored.
(16.) This remark should surely not preclude that the equation between universal quantifications and conjunctions of (atomic) sentences is inappropriate also for other reasons.
(17.) The vagueness of the phrase prakranto 'rtho na gamyate might not be accidental if the verse ought communicate a point which can be acknowledged on the hypothesis of various explications of the formulation.
(18.) In this context an equivocation inherent in the expression vacya-, which might signify either something which is or should be expressed by a meaningful item or something about which something is or should be said, could attain relevance.
(19.) There is no need to make known anything about the content of an expression to a hearer because, if he is a competent speaker of a language, he can identify the pertinent content without any explanatory comment. It should go without saying that the hypothesis concerning the relative scopes of the negations and modal operators do not necessitate the specific elucidation envisaged above.
(20.) In this way one can account for the particle eva in neyam evabhidhiyate.
(21.) Thus it can be in fact true of the assertion to the effect that an assertion is not probative that it is not probative.
(22.) It is not improbable that Bhartrhari attributed importance to his theorem of inexpressibility not only because of the consequence pronounced in verse 29 but also because he considered it to be in opposition to a doctrine adopted in other schools according to which everything which exists is a meaning-object of linguistic items, in technical terms a padartha. In fact, if the theorem is true it follows under the perspective of an ontology acknowledging meaning-relations as objects that at any time there is at least one object which is not also a meaning-object. Thus the domain of what exists always exceeds the realm of objects which are denoted. This thought is insinuated in the segment of the verses 8-11 (which has not been considered above).--One can formulate the consequence of the pertinent theorem of inexpressibility also in terms of facts: To be sure the theorem does not entail that there are facts which can never be expressed, but it implies the weaker proposition that at any time there is at least one semantic fact which is not expressed (at that time).
(23.) It could be supposed that a grasp of this idea is indicated by the expression svadharmena occurring in verse 4. If the element sva- ('own') should be co-referential with the following noun (sambandha-) it might convey the thought that a relation between words and their meaning-objects cannot be specified according to its own quality in the sense that the specific nature of all the items connected by it is made manifest. It must not be overlooked in this context that 'poetic' ingredients, such as intended equivocations of formulations are not uncommon in the Indian philosophical literature.
(24.) One can certainly say:
M is the set which contains the number 'seven, the capital of Nicaragua and M as its members.
But given that the identity of sets depends on the identity of its elements it is questionable that by saying this one has succeeded in specifying a particular set. In the considered case any attempt of specifying the identity of all the members of the characterized set leads to an infinite regress. However, even supposing that an endeavour of specifying a particular set is bound to be unsuccessful here one needs to acknowledge that inescapable failure of an attempt is not tantamount to inconsistency.
(25.) See Dummet (1981, 86ff, in particular 128, or 487).
(26.) Cf. Dummet (1981, 487).
(27.) This might be regarded as a counterpart of the partial specification of conditions of truth regarding sentences with indexical elements when it is ascertained for example that
Today is a Sunday
expresses a truth if the day of an utterance of the sentence (whenever it may be uttered) is a Sunday
(28.) The appearance of difference with respect to (18) and (19) and similar examples is further diminished by taking into consideration that thought-contents or propositions are (and presumably have to be) identified as contents of linguistic items.
(29.) The addition of 'or items' is appropriate because truth-comments can simultaneously express verdicts about several objects, e.g. 'That which S expresses is true and that which S' expresses is not true (and that which ...)'. Because of essential similarities it is not necessary to account separately for cases involving longer chains of truth-comments where truth-comments pertain to other truth-comments.
(30.) It appears that the same verdict could be given if (S2) were replaced by:
(S2)* (S1)* expresses a truth.
(31.) If two entities express something in different ways, e.g. express that (S1) expresses a truth as an internal and as an external comment, it is consistent to suppose that by expressing this fact they establish relations with respect to different contents. It could be tempting to regard those latter relations as relations of expressing. In this case, however, one ought be aware of the resulting equivocation: There is nothing paradoxical in presuming that even one and the same item might be related to a truth in one sense of the word 'express' and to something which is or cannot be true, in a different sense.
(32.) We have put 'that' in brackets because for a plausible interpretation of the formulae it is either required to regard 'that' as inbuilt in the symbols 'K', 'W' or 'F' or to assume that the expressions following those symbols, such as '[[for all]x (Kx [right arrow] Fx)]' in line 1, incorporate a corresponding element. --The accompanying elucidations in EL suggest the second alternative.
(33.) From the circumstance that something is not true which due to its own nature cannot be true no conclusions regarding historical facts are derivable. --Particularly no legitimate deduction can be made from 'it is not true that all ...' to '(it is true that) there is something which is not ..'
(34.) Surely, very few items will without contradiction and correctly attribute to expressions synonymous with them failure of expressing a truth due to their contradictory import.
(35.) Here the expression 'sentence' is employed for the sake of convenience and should not rule out that expressions like 'Yes', 'No', 'Right' etc. can perform the same functions as sentences expressing agreement or disagreement with respect to truth.
(36.) On the other hand it implies the admission that relevant contextual factors determining the content of a linguistic unit can lie in features exhibited by the pertinent unit itself.
(37.) In the original German version they read:
[P1] wenn (x sagt, dass p, und x sagt damit etwas Wahres), dann p.
[P2] wenn (x sagt, dass p, und p), dann sagt x etwas Wahres.
(38.) In the original version the formulations are:
(TE1) [For any linguistic unit s:](x expresses that p & E*s) [right arrow] p
(TE2) [For any linguistic unit s:](s expresses that p & p) [right arrow] E*s
As an intended reading of the symbol 'E*' had been envisaged 'expresses something which is true' or 'expresses a truth'. It was presupposed that sentences free from linguistic ambiguities are under consideration. The purpose of the addition of 'thereby' is to eliminate this restriction.--The occurrence of 'thereby' in (TE2) is in fact dispensable.
(39.) The derivations are presented in AL pp. 246ff.
(40.) [P1] and [P2] as well as (TE1) and (TE2) are only plausible if 'p' can be replaced by formulations which do not exactly reflect the conventional linguistic meaning of original expressions if they contain indexical elements. This means that the relevant concepts of saying and expressing cannot be identified with relations between linguistic units and their conventional standing meaning.
(41.) In the history of Western philosophy, B. Bolzano has advocated a view which exhibits some affinities to the account presented here, but is not identical with it. Cf. EL, p. 88.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||articulo en ingles|
|Publication:||Revista de Filosofia de la Universidad de Costa Rica|
|Date:||Sep 1, 2013|
|Previous Article:||Acerca de asincronias: naturaleza, sociedad y cultura. Ensayos sobre el tiempo: (editor: George Ivan Garcia Quesada).|
|Next Article:||Oriente vs. Occidente. Un acercamiento al pensamiento filosofico oriental.|