On the Methodological Restriction of the Principle of Characterization

The subject of this article is the Principle of Characterization—the most controversial principle of Alexius Meinong’s Theory of Objects. The aim of this text is twofold. First of all, to show that Bertrand Russell’s well-known objection to the Principle of Characterization can be reformulated against contemporary unrestricted interpretations of it. Second, to propose an alternative formulation of this principle. This refers to the methodology of metaphysics and is based on the distinction between pre-theoretical and theoretical languages. The proposed formulation fits well with every type of contemporary interpretation of the Theory of Objects and helps to overcome Russell’s criticism.


Introductions
In 1904, Alexius Meinong presented the Theory of Objects (Gegenstandstheorie) (hereafter "TO"), whose thesis is that "there are objects of which it is true that there are no such objects" (Meinong 1904(Meinong /1960. 1 During the last century, Meinong's theory had a tough time. For long years it was commonly considered to be a textbook example of a false theory. This was mostly due to the influential critique by Bertrand Russell, who argued that the TO contradicts common sense, empirical knowledge, and the laws of (classical) logic. Consequently, some prominent philosophers labeled the metaphysical consequences of this theory "a slum" (Quine 1948) or "a jungle" (Kneal 1949), and the theory was finally pronounced dead and buried, with no chance for resurrection (Ryle 1973, p. 255).
After decades of bad press, some philosophers tried to rehabilitate Meinong's views and place it on the map of contemporary philosophy (Findlay 1963;Chisholm 1973a, b;Routley and Routley 1973). Along with increasing philosophical interest in the problems of intentionality, ontology of fiction, philosophy of modality, and interpretation of quantifiers, this eventually lead to the revival of the notion of nonexistent objects. This resulted in a variety of theories that are commonly labeled "Meinongian." Although each contemporary Meinongian theory is supposed to be immune to the criticism leveled against the original Theory of Objects, at least one charge seems to still be a threat for those who believe that not every object exists. This is the problem of the so-called Principle of Characterization (or the Characterization Postulate) (hereafter "CP"), i.e. the principle which states that an object characterized by certain properties has those properties. Its importance lies in the fact that this postulate is one of the basic assumptions of the TO, but, as Russell argued, it simultaneously leads to a paradox.
The aim of this paper is to argue for a reformulation of the CP that should help in avoiding this problem and, at the same time, could be equally well applied to every kind of Meinongianism. I should stress that I aim neither to deliver any new interpretation of the TO nor to propose an alternative analysis of non-existent objects. Rather, I wish to point out an important common feature of the paradoxes that arise in Meinongian theories. Since similar problems should have similar solutions, I will argue for one of them.
In order to expound the thesis of this paper, I begin with a section that focuses on some of the methodological aspects of metaphysics (Sect. 2). This will provide a basis for justifying my proposal for reformulating the CP. The second section is a brief reminder of the basic assumptions of the CP and Russell's argument against it (Sect. 3). Following that, I show how advocates of Meinongianism tried to refute Russell's critique through various reformulations of the CP; in particular, I focus on explicating two (unrestricted) versions of this postulate (Sect. 4). 2 Analysis will show that they are committed to paradoxical objects. A diagnosis of the problems following from the unrestricted CP will justify a new formulation of the restricted CP (Sect. 5). In the last section (Sect. 6), I will consider potential charges against the proposed view.

Pre-theoretical and Theoretical Languages
The Theory of Objects (as well as contemporary Meinongian theories) was created to explain an interesting and puzzling phenomenon-that of referring to objects that most would say do not exist. As Meinong stressed, we can (and in fact often do) think and talk not only about what does exist but also about what does not exist. This finds its manifestation in the fact that we take sentences such as (1) "Sherlock Holmes is a detective" to be true and (2) "Sherlock Holmes is a butcher" to be false. In other words, in our everyday language we often make non-vacuously true (or false) statements about objects that do not exist. Naturally, this raises many questions and serves as a basis for metaphysical disputes about the structure of reality and the nature of existence. In order to make those discussions as precise as they can be, philosophers propose detailed frameworks intended to explain the puzzling phenomenon. Consequently, some of them postulate a distinction between various types of predicates, others draw a distinction between two types of predication, and others still postulate possible and impossible worlds.
Thanks to this, we can explain why we consider some claims about non-existent objects to be non-vacuously true or false. This is achieved by virtue of paraphrasing everyday sentences into sentences formulated in the languages of these theories. Despite their common aim, these are very different frameworks with very different paraphrasing. In that sense, explanations of the truth of the sentence (1) "Sherlock Holmes is a detective," will differ according to the chosen approach. Below are a few examples that one might find: (1a) "Being a detective is a characterizing property of Sherlock Holmes." (1b) "Being a detective is one of the properties that Sherlock Holmes has internally." (1c) "There is a possible world in which Sherlock Holmes does exist, and in that world he is a detective." Although explanations (1a-1c) provide very different interpretations of (1), what they all have in common is the assumption that an analysis of (1) might give rise to understandable puzzles, especially by virtue of accepting the claim that Holmes does not exist. This is mainly due to the fact that (1) is formulated in everyday, pretheoretical language, which is a breeding ground for vagueness, misunderstandings, and equivocations. The proposals of various philosophers aim to avoid this by paraphrasing (1) in clearer, more precise theoretical language, i.e. the language of metaphysical theories. 3 This practice justifies a distinction between two types of language: (i) pre-theoretical (or folk, everyday) language (PTL), in which we formulate claims that are usually true or false regardless of basic metaphysical facts about reality; and (ii) theoretical language (TL), which provides an interpretation of the former language and which is also a description of the general metaphysical facts about reality. The relation between (1) and its paraphrases are examples of the relation between these two types of languages. 4 The reason why we assert that the truth-values of expressions of PTL are independent of those of TL is shown by the example of (1) and (1a-1c) as well. The first, (1), might be taken to be true regardless of whether we do or do not believe in the truth of (1a), (1b), or (1c). That is so because the truth-value of (1) is in some respects independent of what the metaphysical structure of reality is. It focuses on the plot of a particular fiction. Sentence (1a) on the other hand, is not so obviously true. It makes non-vacuous claims about the structure of reality, namely that there are characterizing properties. Analogously (1b) and (1c) make non-vacuous claims by providing an alternative explanation of (1) and an alternative characterization of metaphysical reality. 5 By describing a detailed picture of metaphysical reality, philosophers are able to offer a solution to (or at least a clarification of) puzzles formulated in everyday language. As has been mentioned, this is achieved by providing a framework (i.e. theoretical language) that often contains technical terms such as "worlds," "tropes," "non-existent objects," etc. Thus, TL plays a double role. On the one hand it clarifies and summarizes expressions of PTL, on the other it provides a picture of the deep metaphysical structure of reality. Some of these frameworks postulate two types of properties, abstract objects, non-existent objects, possible worlds, impossible worlds, etc. Others are philosophically more modest and hold that there are only individuals postulated by natural science and sets of such individuals. Regardless of differences between particular TLs, the aim of doing metaphysics remains the same.
A common aim and alternative ways of achieving it stirs rivalry between various metaphysical accounts. This usually leads to the temptation of comparing alterative frameworks and considering their theoretical virtues. Beside broadly understood simplicity, explanatory power, intuitiveness, coherence with data and many other criteria, one of the key features of any theory is its consistency. And lack of consistency is what the TO has been famously charged with. This is supposedly due to accepting the Principle of Characterization. 4 The reductive analysis of modalities in terms of modal realism ("it is possible that p iff there is a world w, and at w, p") seems to be a good example of such relations, where the left side of the biconditionals contain expressions of PTL, and the left those of TL (Lewis 1973, p. 84). 5 Although I focus on theories that in one sense or another postulate non-existent objects, it is obvious that there are also frameworks that explain the truth of (1) without committing to non-existent objects, see Kripke (1973/2013), van Inwagen (1977/2001, Thomasson (1999). Moreover, there are approaches according to which the proper analysis of (1) reveals that it is actually a false sentence (Russell 1905a). In this sense, one can speak in pre-theoretical (folk) language about what does not exist, while one's theoretical language will rule out the reality of these entities. For a comprehensive analysis of approaches to the problem of fictional objects, see Sainsbury (2010).

The Principle of Characterization
The Principle of Characterization is intended to do justice to the fact that we are not limited in terms of what can be an object of our intentional acts. As Meinong claims, every characterization corresponds to an object that satisfies this characterization. In other words, for every description or set of predicates, there is an object that satisfies that description or those very predicates. Accordingly, the set {"is a mountain," "is golden," …} corresponds with a golden mountain, and that of {"is round," "is square," …} with a round square. The most important difference between these two objects is that the first is merely possible, while the second is impossible. There could be a golden mountain, but there could not be a round square.
This principle has been famously criticized by Bertrand Russell (1905a, b). If it were the case that for every characterization there is an object that satisfies that very characterization, then, Russell argued, there should be an object that corresponds to the sets {"is golden," "is a mountain," "exists," …} or {"is round," "is square," "exists," …}. The consequence of this is that there should be, respectively, an object that is an existing golden mountain and an object that is an existing round square. Obviously neither of these exists, which makes the CP implausible. Since the Principle of Characterization is one of the fundamental assumptions of the Theory of Objects, the theory itself should therefore be taken to be false.
In virtue of the above, philosophers who find Meinong's theory to be an interesting and well-reasoned view are faced with the problem of coming up with a formulation of the CP that gets around Russell's charge. The vast majority of them claimed that this can be achieved by placing some restrictions on the CP. These restrictions came in two kinds. The first is based on the distinction between two types of predicate (Parsons 1980;Routley 1980;Jacquette 1996); the second is based on the distinction between two types of predication (Castañeda 1974;Rapaport 1978;Zalta 1983;Paśniczek 1999). Both originate in the works of Meinong's pupil, Ernst Mally (1904, 1912. Even though these restrictions helped to express the Theory of Objects in a more precise manner, their success in arguing against Russell's charge is sometimes put into question. Accordingly, some claim that the introduced distinctions are either unclear and introduced ad hoc (Griffin 2009) or lead to Clark's paradox, which is a variant of the "barber paradox" (Clark 1978). 6

Item Theory
Due to the controversies concerning versions of the restricted principle of characterization, it might be tempting to lean towards an unrestricted version of the CP. A proposal in favor of this can be found in the late Richard Routley's work (published under the name Richard Sylvan) Re-Exploring Item Theory: Object-Theory Liberalized, Pluralized and Simplified but Comprehensivized (Sylvan 1995). Sylvan argues that the theoretical potential of the TO has not been fully realized. This is supposedly due to the fact that so many have tried to restrict the CP, resulting in Meinong's theory reaching a dead end. 7 In order to avoid this, and to exploit the richness of "Meinong's jungle," one ought to lean towards an unrestricted version of the CP.
It should be noted that while Meinong's theory is labeled "Theory of Objects," Routley calls his view "Item Theory." Because of linguistic connotations, the term "object" is considered to be too narrow for Routley's theory and therefore he recommends switching to "item" instead. The relation between them is such that "all things and objects are items; but the converse may not hold. Mental items, which are (de facto) private and subjective, may not be objects but they are certainly items" (Sylvan 1995, p. 59). In light of this difference, the CP becomes as liberalized as it possibly could be: "Every well-formed subject signifies an item" (Sylvan 1995, p. 52). 8 Accepting this form of the CP, one is bound to admit that there is an object that corresponds to the descriptions supposed to be counterexamples for the TO, i.e. the existing round square and the existing golden mountain. Sylvan admits this, and by referring to an urban metaphor drew a distinction between "an old town" and "suburbia." While the old town is a place where the restricted CP applies and where "typical" Meinongian objects may be found, this is a very narrow area. Far from the old town one may find suburbia-various areas that are populated by unordinary items. Their being unordinary lies in the fact that they are "characterised or axiomatized through different postulates, appropriate to the suburb" (Sylvan 1995, p. 68).
The urban metaphor, along with examples of suburban regions given by Sylvan, suggests that these are areas where what is necessarily false in the old town becomes true in suburbia. Among the objects that populate suburbia one may find the existent round square and the existent golden mountain. This does not mean, however, that any of those objects actually exists. They exist merely in a suburban area that is far from the old-town zone, and as such they have no effect on what one may find in the old town (Sylvan 1995, pp. 78-79).
Although the liberalized CP allows the original Russell's critique to be addressed, one might query whether an analogous problem does not arise in Sylvan's framework. The potential problem is due to the fact that if every well-formed subject signifies an item, one might ask what corresponds to a description such as "an object that is not an item" (or "an item that is an element of no zone"). This characterization is well-formed, which suggests that it corresponds to an item. However, just as the characterization "existing round square" corresponds to an item that is an existing round square, this description should correspond to an object that is not an item.
However, if it satisfies this description, then it is not an item. If it is not an item, then it is not an object either and does not satisfy any characterization.
Because of its paradoxical nature, it is safe to assume that this "object" does not belong to the old town. However, it is not so clear that it belongs to suburbia either. That is because if the aforementioned characterization is satisfied, then this object is not an item. In that sense, there is no room for it in the suburban area any more so than there is in the old town. Considering a similar question, Sylvan claimed: Here is a new subject, Bugboo. Suppose Bugboo is not an item. Then Bugboo is an item, albeit an entirely nondescript one. For whatever is not an item is thereby an item […] Everything, whether existent or not, possible or not, objective or not, even absurd or not, is an item. (Sylvan 1995, pp. 55-6) What seems to be problematic with the above is that on the one hand it is claimed that Bugboo is not an item, and on the other that Bugboo is an entirely nondescript item. Assuming that "is not an item" is a predicate, there is a description of Bugboo after all ("something that is not an item and is called 'Bugboo'"). Secondly, if something that is supposed not to be an item is an item after all, then one may question whether any other item has the property that we have ascribed to it using the CP. Accordingly, a characterization like "something that is not a golden ring" could be satisfied by an object that is a golden ring. This calls into question both the plausibility of the CP and the consistency of the notion of "item" in general.
It is not easy to address these concerns based on Sylvan's work. This is mainly due to the fact that the presented liberalized Characterization Postulate appears to be merely the outline of a new approach to the problem of non-existent objects. However, what Sylvan sketched out finds itself developed in Graham Priest's view (2005, 2016. 9 Priest's interpretation of the CP is presented in terms of worlds semantics, as a result of which some label it "modal Meinongianism" (Berto 2011(Berto , 2013a. In order to distinguish between Priest's and Sylvan's views, I will go along with this change in terminology. 10

Modal Meinongianism
The aim of modal Meinongianism is to provide an interpretation of the CP that would be, on the one hand, unrestricted and, on the other, immune to Russell's argument. According to this version of Meinongianism, every description corresponds to an object that satisfies the description. However, thusly described objects do not have to possess ascribed properties in the actual world. For example, Russell's characterization "is an existing golden mountain" is satisfied by an existing golden mountain. Since "to exist" in this framework means "to be spatiotemporal," and since there is no such object in the actual world, something has these properties in a possible though unactualized world (Priest 2005, pp. 59-60).
Besides merely possible objects, the Meinongian universe contains objects that are impossible, such as the aforementioned existent round square. But this is not a problem for modal Meinongianism either, because along with possible worlds it also postulates a plenitude of impossible worlds, i.e. worlds where what is impossible (necessarily false) in the actual world, is true in those worlds. Under this assumption, the description "something that is an existing round square" is satisfied by an existent round square. Due to it being an impossible object, it is only to be found in impossible worlds.
Although objects such as the round square and the golden mountain do exist in unactualized worlds, they also belong to the actual world. However, they possess different properties in it; namely, neither of them exist in the actual world. Also, they have properties such as being an object of Meinong's thought, existing in an unactualized possible (or impossible) world, etc. This is because the domain of objects is constant for every world (Priest 2005, pp. 12-4). What differs between worlds is what properties a given object has in them. In this sense the domain of the world of Sherlock Holmes' stories is the same as the domain of the actual and any other world. 11 However, while in our world Sherlock Holmes is a non-existent object (i.e. does not have properties that entail being actually spatiotemporal), he does exist in some possible and impossible worlds. As has been mentioned, the non-existence of Holmes does not preclude him possessing some properties in the actual world as well. Examples of these are being a fictional character, being described by A.C. Doyle, and many others that do not imply being a spatiotemporal object.
It appears that if one has no qualms about accepting commitments to impossible worlds (and there are many arguments in favor of their theoretical virtues (Berto and Jago 2018)), then modal Meinongianism might be considered an attractive theory. This is mainly due to the fact that it avoids Russell's original criticism and preserves the main theoretical advantages of Meinong's theory. However, the unrestricted CP may lead to controversies that are very similar to those arising from the Theory of Objects and Item Theory.
What is puzzling about this modal approach is that one can reformulate Russell's original charge into a question about an object that satisfies descriptions such as "something that is an actually existing round square" or "something that is not an object." 12 This-an advocate of modal Meinongianism claims-does not have to be a problem at all. The unrestricted CP states that every characterization is satisfied in a given world. The world that is under consideration, however, does not have to be the actual world. In this sense, just as the fact that the description "something that is an actually existing golden mountain" is satisfied in a given world does not imply that it is actually true that the golden mountain exists, the fact that in some world the description "something that is not an object" is satisfied, does not imply that actually something is not an object. While the case of the first characterization results 11 While modal Meinongianism might be reformulated also in variable domain semantics (Priest 2016, p. 263), this is considered to be a superfluous change of the original view (Priest 2016, p. 13). 12 Further examples include "something that is outside the domain of objects," "something that is an element of no world" or "something that is golden and is not an object". in extended analyses in the literature, the problem of the second appears to find less interest. 13

Something that is not an Object
The details of Graham Priest's (2016) analysis of the characterization (Ch1) "is not an object" merits closer inspection. His treatment of Ch1 starts with the claim that being an object is a property, and "x is an object" (Ox) means the same as "x is something." Since, in terms of modal Meinongianism, "being something" means to be quantified over by neutral (i.e., existentially unloaded) particular quantifier ( ); "x is something/x is an object" is understood as yy = x. This quantifier ranges over everything, hence the property O is a universal property, and it is actually necessarily true that everything is an object. If that is so, then it is impossible for something to be not an object (¬O). This-according to modal Meinongianism-entails that at an impossible world w, it is true that something is not an object, x¬Ox. Hence, ¬Oo is true at w and Ch1 is satisfied after all. This, however, does not entail that ¬Oo is true at the actual world (Priest 2016, p. 245).
It should be noted that Graham Priest leaves the question of whether Oo is true at w open. Assuming the above, however, it seems that it should not be an open question. If Ch1 is meant to be satisfied at w by o, it is true at w that something (i.e., o) is ¬O, hence x¬Ox is true at w. Since "to be something" means the same as "to be an object," Oo is true at w. If this were not the case, one would not be able to quantify over o, and x¬Ox would not be true at w either. Consequently, if Ch1 is satisfied at w by o, both Oo and ¬Oo are true at w. This naturally makes o have contradictory properties, but o-as Priest argued-has those not in the actual world, but in an impossible world w.
What is crucial for the above analysis is how "¬Oo" should be understood. It seems that there are two options: (i) "o has the property of being a non-object" and (ii) "o does not have the property of being an object." 14 I am going to argue that regardless of the chosen interpretation, it is not the case that o satisfies Ch1.
Let us assume that (i) is the proper interpretation of ¬Oo. If this were the case, then it would be true at w that o has the property of being a non-object. This, however, does not allow for believing that o satisfies Ch1. After all, to satisfy Ch1 means to lack the property of being an object, and this is very different than having the property of being a non-object. Consider an analogy between Ch1 and (Ch2) "is not round." Ch2 is satisfied by r, such that r is not round. In other words, r does not belong to the set of round objects. This, however, does not entail that r is a member of the complement set, i.e., the set of non-round objects. After all, some objects are incomplete, and they are neither round (R) nor non-round (¬R). Consequently, in the case of Ch2, the question of whether "r is non-round" ([¬R]r) is true is beside the 14 For now, I will use the prefix "non-A" as indicating a negative property that is complementary to A. point. What is not beside the point, however, is that a world where r satisfies Ch2 is a world where r is not R, i.e., where Rr fails to hold.
Likewise, Ch1 does not ascribe the property of being a non-object, but rather indicates that something lacks the property of being an object. Hence, if something satisfies Ch1 at w, it should be something that does not have the property of being an object. Since Oo is true at w, it is not true that o lacks the property of being an object. Because of this-assuming that (i) is the proper understanding of ¬Oo-o does not satisfy Ch1, but rather (Ch1*) "is a non-object." 15 The above leads to the assumption that "¬Oo" stands for "o does not have the property of being an object." What seems to be problematic with this is that, along with the definition of being an object and the assumption that the domain of objects is constant, if Oo∧¬Oo were true at w, it would be true at every other world. That includes the actual world. The reason for this is that, if (ii) is the proper understanding of ¬Oo, "o is an object" is false at w. Since the truth of "x is an object" is understood as yy = x, one is justified in interpreting the falseness of "o is an object" as ¬ yy = o. This results in the claim that o is not an element of w, i.e. that o is not something that can be quantified over. Assuming that the domain of w is the same as the domain of any other world, o is not a member of any other world. 16 Since o does not belong to any world, and since satisfying a characterization means satisfying it in a given world, o satisfies no characterization. This includes Ch1. Moreover, if ¬ yy = o is true at w, Oo is not true at w. After all, if o does not belong to the domain of objects, one cannot quantify over o, hence yy = o (i.e., Oo) is not true at w.
In virtue of the above, if Oo∧¬Oo were true at w, the domain of objects of w would be inconsistent. This means that the domain would both contain and not contain o, i.e., both yy = o and ¬ yy = o would be true at w. Since the domain is constant, if Oo∧¬Oo were true at any w, it would be true at every world. That includes the actual world. Hence, either Ch1 remains unsatisfied and the unrestricted CP is false, or Ch1 is satisfied by o and each and every world (including the actual world) is such that it both contains and does not contain o.
This shows that the unrestricted version of the CP results in a consequence that seems to be no less problematic than the consequences of Meinong's original theory as pointed out by Bertrand Russell. However, it appears that there is something peculiar about the above counterexamples against the unrestricted CP of Item Theory or modal Meinongianism. In the following section I will try to reveal this 15 What it precisely means to be a non-object is no trivial problem, but I will not attempt to solve this here. 16 It should be stressed that this is different than saying that at a given world something does not belong to the domain of objects that are P. While the domain of objects that are P (e.g., objects that are presidents) varies from world to world, the domain of objects remains the same. Even though Donald Trump is a president at the actual world, this does not entail that he is a president in every other world. After all there is a possible world where he is a detective and an impossible world where he is a prime number. Contrary to this, the domain of objects is constant and "is an object"-as opposed to any other predicate-has the same extension across worlds. Hence, if something does not belong to the domain of objects in one world, it does not belong to the domain of any other world. peculiarity and suggest why Russell-like arguments against the CP might be considered misleading.

Methodological Restriction
Considering the remarks of the first paragraph, the role of Meinongian theories is to provide a philosophical explanation of common-sense intuitions according to which we can refer to non-existent objects. For modal Meinongianism, the TL is the language of philosophical theory, containing theoretical terms such as "actuality," "possible world," "impossible world," "open worlds," etc., together with axioms of the theory. The PTL is, of course, our folk language, in which we talk about nonexistent objects such as Sherlock Holmes, Batman, or round squares. If one accepts this distinction between our loose everyday language and the precise language of philosophical theory, then one can address the Russell-like argument against modal Meinongianism by pointing out that the criticism conflates the terms of PTL ("is round," "is square," "is golden," "exists," etc.) with the categories of TL ("is actual," "is an object," etc.). Hence, by including TL's terms in a given characterization one changes the subject of modal Meinongianism. That is, while originally the subject of this theory was the deep metaphysical structure that is revealed by the paraphrasing of PTL into TL, here the subject is the TL itself.
The reason why this change of subject is erroneous is that it leads to an unjustified expectation with respect to the theory-namely, an expectation of expressing the falseness of the TL in the very TL of the theory. This is illustrated through the example of objects that would satisfy descriptions such as "something that is an actually existing round square," "something that is not an object," "something that is an object that does not belong to the domain of objects," etc. In order to satisfy the above expectation, one would have to postulate objects that contradict the very basic assumption of modal Meinongianism, i.e. that everything is an object and that impossible objects do not exist in the actual world.
This unjustified expectation appears similar to the one that we raised in the case of Sylvan's Item Theory, i.e. an expectation concerning an object that would satisfy the characterization-"an object that is not an item." Since, according to Item Theory, everything is an item, no object fails to be an item. Again, the expectation of indicating such an object would require contradicting the basic assumptions of the TL. In this sense, the alleged counterexamples for the CP of modal Meinongianism or Item Theory should not be considered a real threat within those frameworks. That is because those counterexamples are based either on a mix of PTL and TL or on the expectation of expressing the falseness of a given TL in terms of that TL.
What might this mean for the debate on Meinongianism in general? It appears that Russell-like arguments against the CP are based upon the inclusion of terms of the TL in characterizations, such as "existent," "non-existent," "is an object," "is a set," "is incomplete," "is impossible," "encodes," "is a characterizing predicate," and so on. Depending on the given view, these terms belong to the theoretical language and not to the everyday language whose semantics we wish to explain. Thus, just as the description "something that is not an object" from modal Meinongianism's point of view might be considered improper because it contains the terms of theoretical language, so the set {"is round," "is square," "exists," …} might be considered from the TO's point of view to not be a well-formed characterization set.
On the basis of the considerations above, one can try to formulate a version of the principle of characterization that fits every type of Meinongianism. This is a kind of restricted principle, and its restriction is expressed in terms of the distinction between PTL and TL as outlined above.
Every characterization, which is expressed using the terms of pre-theoretical language, corresponds to an object that satisfies this characterization.
I believe that the above is a formulation of the CP that (1) highlights the role of metaphysical theories, which is to explain the use of our folk language and describe the metaphysical structure of reality, (2) is immune to variations of Russell's (and Clark's) charges against the CP, and (3) is less controversially restricted than the options presented by Parsons,Routley,or Jacquette. 17 That being said, three points should be stressed. Even though this formulation of CP might be applied to every Meinongian approach, it obviously does not make them all on par when it comes to their theoretical virtues. Secondly, the aforementioned formulation seeks to justify an exclusion of what was considered to be a counterexample against a given Meinongian approach. As such it does not exclude objects that have contradictory properties, such as a table that is both round and nonround. Objects of this kind, however, are not problematic for advocates of Meinongianism, and they find their place in each and every Meinongian account, including the original Theory of Objects. This changes in cases of characterizations such as "is both existent and non-existent," "is both possible and impossible," "is complete and incomplete," etc. The change, however, is grounded in the fact that these descriptions contain terms of TL and not in the fact that they ascribe contradictory properties to an object.
Finally, it should be noted that there is an important difference between a genuine characterization (e.g., "is both A and non-A") and what merely looks like one (e.g., "is A and it is not A"). While the first ascribes two contradictory properties to something, the characterization itself is consistent-it ascribes A and it ascribes non-A. The second one, however, is inconsistent and-as long as one does not assume that "is not A" entails "is non-A"-it simultaneously ascribes the property of being A to an object and denies that this object possesses this very same property. This exposes the second as an invalid characertization and should, in turn, not be taken to be counterexamples against the CP. After all, the characterization "is A" is satisfied by an object that is A. While this very object (if it is impossible) may also be non-A, it cannot fail to be A. In other words, if a satisfies "is A" it is not true that a is not A. 18

Possible Objections
In this section I am going to consider some of possible charges against my proposal. These are focused either on the proposed formulation of the CP or on the plausibility of the distinction between PTL and TL.

There is no Clear Distinction Between PTL and TL
The presented proposal is based on a distinction between PTL and TL. One could argue, however, that considering some of the aforementioned examples to be expressions of theoretical language and not of folk language contradicts apparent evidence. After all, it is quite common to use words such as "world," "property," "existence," "possible," "actual," "object," etc., without saying anything about the metaphysical structure of reality, e.g. "The world of European fauna is more complex than the world of Arctic fauna," "Copper has the property of conducting electricity," "After years of hard work, the Eiffel Tower came into existence," "Given that their star player is out injured, I think it is impossible for them to win tonight," "The actual stores offer local convenience, while the virtual stores offer variety for shoppers," "Animals are living beings and not objects." Consequently, the proposed restriction of CP may be considered to be based on an implausible assumption about the distinction between two types of languages.
Although it is hard to argue with the observation that some words belong to both PTL and TL, one should notice that the meanings ascribed to these terms are different than the meanings ascribed to similar words in theoretical languages. 19 This is mainly due to the imperfections of pre-theoretical language, which ought to be removed by clarifications made in the theoretical language. In this sense, there is no contradiction in using the term "property" in its folk meaning and rejecting the existence of properties when it comes to theoretical language. Just as there is no contradiction in being an animal rights activist and considering them objects (i.e. elements of the domain of quantification). Hence it is possible that from a syntactic point of view some words are elements of both PTL and TL. Nonetheless, the meanings of these words are different, which allows them to be considered part of TL and not PTL.
As I claimed, the difference between the terms of PTL and a given TL is that the latter ones are usually more precise. Thus, while we allow for the ambiguity of words such as "existence," "property," "object," etc. in PTL, preferably TL should avoid this and provide refined and univocal meanings for key-terms. Naturally, sometimes we want to be precise not only in doing philosophy but also in everyday situations. Thus, there are situations involving day-to-day communication where we use a given term while ascribing a more refined, theoretical meaning to it (especially if one of the speakers is a philosopher). After all, we sometimes use "impossible" to mean "inconsistent" and not merely "unbelievable" or "unlikely." The fact that we can mix terms and meanings isn't a reason for putting the distinction between TL and PTL into question, but rather provides a simple example of cases where we use precisely characterized terms in everyday communication. It is fair to say, however, that such situations are a negligible part of everyday conversations.

The Mere Difference Between Meaning Ascribed to a Term in Common and Metaphysical Discourse is Enough to Undermine the Value of Metaphysics
The discrepancies between plain pre-theoretical English and the theoretical English of philosophers may be taken as grounds for an argument against the plausibility of such philosophical theories. After all, this might be interpreted as a result of the work of "the philosophers who advance […] various anti-commonsensical doctrines [and who are] driven by an obsessive fascination with highly rarefied general principles and philosophical riddles, which invariably override any pre-philosophical intuitions of common sense" (Hirsch 2008: 371).
It should be noted that metaphysics (and philosophy in general) is not the only branch of inquiry where these kinds of discrepancies between common and refined meanings words take place. The botanical use of "fruit" differs from the use of this term in everyday communication. Consequently, what botanists consider to be fruit, many competent users of plain English consider to be vegetables (e.g., tomatoes or bean pods). This difference, however, is not taken to be a charge against the plausibility of the botanic use; rather, it shows that the common use of the term "fruit" is simply imprecise. It is safe to assume that something similar happens in cases of "exists" or "object," which refer to much more abstract concepts than the concept of fruit.
Moreover, it is not only the case that the meaning ascribed to some terms is different in PTL and TL; it is also the case that different theoretical languages give different characterizations of key notions. For example, "existence" has a different extension in Quine's (1948), Lewis' (1986), and Priest's (2005) frameworks. 20 This, however, does not have to entail that these philosophers merely talk past each other. The debate between them might be a substantial one and grounded in the fact that they argue in favor of different theoretical clarification of everyday "existence" (van Inwagen 2008). 21 This ultimately results in differences with respect to what elements of the domain of existent objects are. 22 20 Similarly, the "possible world" of concretism and abstractionism are characterized in very different ways. 21 Just as the fact that there are different theoretical characteristics of negation does not have to put into question the substantiality of the debate between intuitionistic and classical logicians (Priest 2006, ch.4.2.). 22 Another example of this phenomenon is the debate between Russell and Meinong. While both of them agreed that every name refers to an object, they ascribed very different theoretical characteristics to the notions of "name" and "object" (Chrudzimski 2007). Consequently, Russell concluded that some expressions, which we consider names, are merely hidden descriptions. Meinong, on the other hand, believed that some objects do not exist.
Considering various approaches to this question, it is sometimes argued that one's theoretical notion of "existence" is more intuitive for it fits better the common use of the term than alternative proposals do. This can be (and often is) used as an argument in favor of one's approach compared to others (Hirsch 2008). It should be noticed, however, that since ordinary language has to cope with metaphors, vagueness, speech acts, context dependence, and so forth, it is imprecise by nature. Assuming that precision is a highly desirable theoretical feature of any philosophical framework, the discrepancy between a common and a theoretical notion of, e.g., "exists" does not have to be considered a conclusive argument against a particular theoretical framework. Thus the subject of the debate between the above mentioned philosophers might be considered the question of which theoretical characteristics of the notion of "existence," on the one hand, provide the refined meaning that is required for the precision of philosophical analysis and, on the other hand, is as intuitive as is possible at the same time.

The Relation Between PTL and TL is Unclear
Since this distinction between two types of language has been introduced, one may ask about the relation between them. Because of the role of TL, which is to provide a summarizing paraphrase of sentences in PTL, the relation between them can be considered akin to a metalanguage and an object language. After all, by virtue of the use of paraphrases, the former is often used to examine and clarify expressions of the latter. This analogy, however, is not so obvious. This is mainly because the subject of TL is not folk language per se, but rather the deep metaphysical structure of reality. As we have pointed out, this is indirectly achieved through the analysis of PTL; nevertheless, TL does not consider, for instance, the grammatical structures of the latter or names of sentences within it, etc. Thus, the interpretation of TL as the metalanguage of PTL would require enriching it with a capacity to explicitly describe the structure of the folk language in detail. 23 Another reason for skepticism regarding the plausibility of this analogy is based on the assumption that, while the subject of the object language is the world, the subject of the metalanguage is an object language. In contrast, both PTL and TL have a very similar subject-the world. Nevertheless, they play very different roles. While the former is used in day-to-day communication, the latter is intended to reveal and describe the deep metaphysical structure of reality. Because of this, even though the analogy between object language and metalanguage is somewhat limited, what is important is the very nature of the difference between these two languages. This difference lies in the roles of pre-theoretical and theoretical languages. What is important for our proposal is the claim that the aim of TL is to provide a description of deep metaphysical reality. This is achieved by the proper paraphrasing of expressions in the folk language, by (1) introducing new technical terms and/or clarifying the meanings of common language terms, and (2) the axioms of given metaphysical theories.

Restricted CP Results in Too Narrow a Notion of "Object"
Finally, one might claim that any kind of restriction on the CP contradicts the plausibility of the Principle of Intentionality, which states that every intentional act has its object (Meinong 1904(Meinong /1960. After all, some correlates of characterization made in terms of TL were analyzed in this very paper. "Object," which is supposed to correspond with the characterization "something that is not an object," is an example of this. This may suggest that by virtue of the Principle of Intentionality these items should be included in the domain of objects.
Notice, however, that these kinds of objects are usually described in terms of TL. As such, they do not have to be considered subjects of a metaphysical theory. After all, it seems unreasonable to expect a metaphysical theory to analyze objects that are not postulated by it, or those that are explicitly excluded from its domain.
Nevertheless, I have written about such "objects" as being those that are described in terms of the TL of Item Theory or modal Meinongianism. This has not, however, been done in the TL of those views, but rather in a language whose aim is to describe the consequences of accepting the TL of Item Theory or modal Meinongianism. As such, even though we referred to "object" in some way, the subjects of our analysis were those theoretical languages (or metaphysical theories) and not the metaphysical structure of reality. That makes our analysis the subject of metametaphysics rather than metaphysics. Since the subject of each and every Meinongian theory is the structure of reality, and not metaphysical theories of it, they do not have to be threatened by these kinds of "objects." That is because the analysis of metaphysical theories is not the aim of metaphysical theories.
The puzzle of the Principle of Characterization puts Meinongians in a highly problematic position. The default options were to accept the unrestricted principle of characterization or to propose some kind of restriction of it. The latter has been considered either to be based on ad hoc distinctions or as one that results in Clark's paradox. The former, unrestricted principle results in an inconsistent domain of objects.
Alternatively, we can place restrictions on both the Principle of Characterization as well as on the Principle of Intentionality. This would surely limit the notion of "object" to only those entities that are expressed in non-theoretical terms and, as a consequence, restrict the generality of the original Theory of Objects. Nonetheless, if one construes Meinongian theories as a kind of metaphysics, then this kind of restriction seems reasonable-it releases Meinongianism from paradoxical items and at the same time explains the phenomenon of referring to non-existent objects. The restriction is based on a distinction that applies not only to Meinongianism, but to metaphysics in general. This allows for believing that the imposed restriction has not been introduced ad hoc.