The homogeneous nature of the theatre medium ELI ROZIK Abstract To date, traditional theatre semiotics has not succeeded in creating an intelligible and effective method of performance analysis. I believe, nonetheless, that a semiotic method that aims at revealing the mechanisms of generating theatre meaning is viable and even obligatory. I contend that the problem stems from the erroneous theoretical assumption that theatre draws upon different component codes and is, therefore, a heterogeneous medium. I suggest, rather, that the theatre medium is homogeneous; that 'iconicity' should be redefined in terms of imagistic thinking, in the sense of operating images imprinted on matter, and that this definition accounts for its homogeneity. I show that this definition also leads to the following conclusions: (a) iconicity transforms units 'borrowed' from other systems of signification and communication into descriptions of such units, thus bestowing homogeneity upon theatre performance-texts; (b) the notion of 'iconicity,' in the above sense, also applies to imprinted images of speech acts that combine verbal and nonverbal elements into indivisible nonverbal units; and consequently, (c) a performance-text is an iconic description of a (usually fictional) world, generated by the homogeneous theatre medium. Although reflecting criticism of traditional approaches, this study reaffirms the charter of theatre semiotics. Keywords: iconicity; image; imprinting; mediation; homogeneity; theatre medium. 1. Introduction Until 1982, Patrice Pavis was one of the leading scholars promoting a semiotic or, rather, a 'semiologic' approach to theatre performance-texts. However, his Languages of the Stage (1982) and Dictionary of the Theatre Semiotica 168-1/4 (2008), 169-190 0037-1998/08/0168-0169 DOI 10.1515/SEM.2008.009 © Walter de Gmyter 170 E. Rozik (1998, first published in French in 1996), which best reflect the state of disorientation during the last decades of the twentieth century, convey a sense of despair with regard to the possible creation of a methodology of performance analysis (Pavis 1998: 255). Whereas '[ťjheater semiology was born of a desire to avoid impressionistic discourse on performance' (Pavis 1998: 258), he asserts that 'the illusion of scientific analysis' should be abandoned (Pavis 1998: 255). I agree with Pavis that traditional theatre semiotics has not succeeded in creating an intelligible and effective method of performance analysis. I believe, nonetheless, that a semiotic method that aims at revealing the mechanisms of generating theatre meaning is viable and even obligatory. I suggest that the problem stems from the erroneous theoretical assumption that theatre draws upon different component codes and is, therefore a heterogeneous medium. Following Otakar Zich, Petr Boga-tyrev contends that the stage presents not only linguistic signs (speech), but also a range of various kinds of nonverbal signs, such as costume, scenery, gesture, movement, posture, miming, voice, and facial expression, whose heterogeneity he terms 'plurisignation' (Bogatyrev 1986: 43). Tadeusz Kowzan suggests thirteen component sign systems, including speech (Kowzan 1975: 182ff) and Martin Esslin expands the list to twenty-two such systems (Esslin 1992: 103-105). Other theories vary in between these boundaries. Erika Fischer-Lichte claims that since the theatrical code 'is constituted by a large number of different semiotic systems,' unlike language, it cannot be decomposed into a set of 'homogeneous smallest signifying units' (Fischer-Lichte 1992: 224). Although she assumes that further 'segmentation' is possible, this would result in leaving the orbit of the theatre code and entering that of its various constituent codes (Fischer-Lichte 1992: 218ff). Accordingly, she defines the 'theatrical text' as a 'multimedial text' (Fischer-Lichte 1992: 222; cf. Pavis 1982: 15; 1998: 58ff; De Marinis 1993: 1; and Ubersfeld 1999: 14). She fails to see, however, that this conclusion contradicts her own claim that 'at the systemic level, the theatrical code contains only iconic signs' (Fischer-Lichte 1992: 16), which, I suggest, bestow homogeneity on the entire medium. Moreover, a classification of all component signs cannot be accepted, because it is tantamount to the classification of everything in the world and thus worthless. It would appear, however, that possible heterogeneity on the grounds of the alleged impossible reduction of verbal and nonverbal codes to a common denominator poses a real problem; e.g., Jiří Veltruský considers that only two systems are essential, language and acting, endorsing thereby theatre's heterogeneity (Veltruský 1986a: 114-115). 'Speech' and 'iconicity' indeed seem to be incompatible categories, belonging in The homogeneity of theatre 111 different semiotic domains. In Charles S. Peirce's terms, whereas the verbal code is 'symbolic,' the nonverbal codes are predominantly 'iconic' or 'indexical,' with each code requiring different principles of decoding (CP 2.247, 2.274-2.308). Whereas 'speech' is usually conceived in terms of basically aural communication, 'iconicity' is in terms of both aural and, predominantly, visual communication. Furthermore, there seems to be a tacit agreement that an iconic replica of speech is impossible. I believe, however, that these observations are not valid for iconicity, but only for the domain of replicated signs. In particular, I intend to question the dual approach to heterogeneity and suggest that no theory presupposing the heterogeneity of the theatre medium would be able to produce a productive methodology of performance analysis. I claim, instead, that the theatre medium is homogeneous; and that 'iconicity' should be redefined in terms of imagistic thinking and communication, in the sense of operating images imprinted on matter, and that this definition accounts for its homogeneity. I show that such a definition also leads to the following conclusions: (a) iconicity transforms units 'borrowed' from other systems of signification into descriptions of such units, thus bestowing homogeneity upon theatre performance-texts; (b) this notion of 'iconicity' also applies to imprinted images of speech acts that combine verbal and nonverbal elements into indivisible nonverbal units of action; and therefore (c) the performance-text is an iconic description of a (usually fictional) world, generated by the homogeneous theatre medium. Although reflecting criticism of traditional approaches, this study reaffirms the charter of theatre semiotics. I support these theses by recent findings in neurobiology and speech act theory, and illustrate them by analysis of a scene from the Habimah production of Anton Chekhov's The Seagull. 2. The imagistic nature of iconicity 'Iconicity' is a crucial notion in any semiotic account of the theatre medium. In the context of a modern theory of signification and communication it was first employed by Peirce (CP 2.247, 2.274-2.308). From his seminal definitions we learn that 'icon' is defined in terms of 'motivation' through 'similarity': 'Anything whatever, be it quality, existent individual, or law, is an Icon of anything, in so far as it is like that thing and used as a sign of it' (CP 2.247). 'Motivation' means here that a sign can be decoded by natural inference, with no need to learn the system, through either similarity or contiguity. 'Icon,' in the sense of motivation by similarity (e.g., an iconic smile) belongs in a triad that also includes 'index,' in 172 E. Rozik the sense of motivation by contiguity (e.g., a real smile indicating a state of mind); and 'symbol,' in the sense of unmotivated (conventional) link between a 'signifier' (e.g., the sound of the word 'smile') and a 'signified' (the notion of 'smile').l Since 'similarity' applies to additional kinds of relationship, such as two peas, twin brothers, and two books of the same edition, an iconic unit is supposed to be first recognized as a sign, and only subsequently distinguished from other kinds of signs on the grounds of motivation through similarity (CP 2.247; cf. Sebeok 1975: 242). Following Jan Mukařovský, Bogatyrev implies that all things on stage are signs (Mukařovský 1986: 3-9; Bogatyrev 1986: 35ff), a semiotic principle explicitly articulated by Veltruský: 'All that is on stage is a sign' (Veltruský 1964: 84). Within the set of theatre signs, Bogatyrev distinguishes between a 'sign of a material thing' and a 'sign of a sign (of a material thing)'; e.g., a (real) 'national costume' is a sign of membership in a class, such as nationality and religion (Bogatyrev 1986: 33ff), and, when replicated on stage, it becomes a sign of a sign. In contrast, the curtain and the footlights are conceived as signs of the (material) stage (Bogatyrev 1986: 34). Signs and signs of signs thus coexist on stage. Fischer-Lichte accepts the notion of'sign of sign' (Fischer-Lichte 1992: 15), but prefers Peirce's notion of 'iconic sign' for characterizing this duality: 'Theatrical signs are ... not identical with the signs primarily generated by cultural systems, but rather portray these as iconic signs' (Fischer-Lichte 1992: 16). Indeed, iconic signs are signs of signs, because they replicate any kind of sign, including indexes (e.g., an actor enacting a laugh), symbols (e.g., a set enacting traffic lights), and even iconic signs (an actor enacting an actor on stage). Fisher-Lichte fails, however, to see that iconicity, which transforms 'borrowed' signs into descriptions of such signs, bestows homogeneity upon theatre-texts. The traditional definition of 'iconicity' in terms of 'motivation through similarity,' which underlies Fischer-Lichte's considerations, is widely accepted in theatre semiotics (Fischer-Lichte 1992: 16; cf. Carlson 1990: xiv), even by its critics on various grounds (e.g., Pavis 1998: 175). However, while providing a reasonable description of nonverbal units, it fails with regard to the verbal components of performance-texts, which are thought to be fundamentally symbolic in their 'borrowed' domain. / suggest instead a definition of 'iconicity' in terms of 'imagistic thinking' that, without contradicting the principles of 'motivation' and 'similarity,' solves the problem of the alleged heterogeneity of the theatre medium. Furthermore, the universality of the basic forms of theatre semiosis leads to the conjecture that their roots must lie in an elementary mental faculty, to which the traditional definition hardly relates: the natural ability of the brain to spontaneously produce images and employ them in an The homogeneity of theatre 173 alternative form of thinking. The redefinition of 'iconicity' in terms of imagistic thinking enables its connection to this vital faculty, plausibly of preverbal ancestry. Traces of primary imagistic thinking are found in dreaming, daydreaming, children's imaginative play and drawings, mythical creativity, and the iconic arts (Rozik 2002a: 247-313). Susanne Langer contends that images are 'our readiest instruments for abstracting concepts from the tumbling stream of actual impressions. They make our primitive abstractions for us, they are our spontaneous embodiments of general ideas' (Langer 1996: 145). Moreover, images are 'just as capable of articulation, i.e., of complex combination as words' (Langer 1996 [1942]: 93). Thinking is anchored in perception, the raw material of innate imagistic representation. She suggests the following definition of thinking: Man, unlike all other animals, uses 'signs' not only to indicate things, but also to represent them ... We use certain 'signs' among ourselves that do not point to anything in our actual surroundings. Most of our words ... are used to talk about things, not to direct our eyes and ears and noses toward them ... They serve, rather, to let us develop a characteristic attitude toward objects in absentia, which is called 'thinking of or 'referring to' what is not there. 'Signs' used in this capacity are not symptoms of things, but symbols. (Langer 1996 [1942]: 30-31)2 'Thinking' thus presupposes two main conditions: representation of things in the mind and manipulation of them in absentia; i.e., it takes place when such representations are disconnected from actual experience and applied to past or possible experiences. This definition of thinking suits imagistic representation too. Langer's prescientific intuitions are amply corroborated by recent findings in neuroscience. On the grounds of digital methodology, Stephen M. Kosslyn asserts that '[i]magery is a basic form of cognition, and plays a central role in many human activities — ranging from navigation to memory to creative problem solving' (Kosslyn 1995: 1). He distinguishes between 'propositionaľ and 'depictive' representation; the latter conveying 'meaning via their resemblance to an object, with parts of the representation corresponding to parts of the object' (Kosslyn 1995: 5). 'Depictive representation' is thus synonymous to 'imagistic representation.' Antonio R. Damasio suggests that having a mind means 'the ability to display images internally and to order those images in a process called thought' (Damasio 1994: 89). Moreover, whatever is not 'imageable,' including words and mathematical symbols, cannot be known and, therefore, cannot be manipulated by thought (Damasio 1994: 107). He distinguishes between 'perceptual images' (e.g., running your fingers over a 174 E. Rozik smooth metal surface) and 'recalled images' (Damasio 1994: 96-97), which should be conceived as images that have been disconnected from actual experience and become units of thought. Kosslyn characterizes thinking as hinging on two properties: 'first, information must be represented internally; and second, that information must be manipulated in order to draw inferences and conclusions' (Kosslyn 1996: 959). He reconfirms thereby Langer's claim that thinking is the manipulation of real objects 'in absentia,' i.e., by means of their representations, of which mental images are a special case. The redefinition of 'iconicity' in terms of imagistic thinking, in addition to connecting it to the natural faculty of the brain to think by means of images, expands its models not only to real models but also to imaginary ones, and suits all verbal and nonverbal forms of theatre semiosis.3 This definition of 'iconicity' poses two problems: first, spontaneous mental images are figments of the imagination, i.e., nonmaterial entities that cannot be perceived by the senses and, therefore, cannot be communicated. Mental images thus require material vehicles that enable their perception by others. I suggest, consequently, that it is the imprinting of an image on matter that creates an iconic unit. In Saussure's terms, both the imprinted image and the imprinted matter constitute the signifier of such a unit (Saussure 1972: 99). Each iconic medium is defined by the kind of imprinting matter that enables its images to be communicated. Whereas most iconic media use matters different from those of their models, underscoring thereby their signifying and communicative functions, the theatre medium is characterized by expanding the principle of similarity, including identity, to the material level; e.g., images of human behavior are imprinted on actors' bodies, images of clothes on real fabrics and images of light on real light. The signifier of the theatre iconic unit is then characterized by application of the principle of similarity on both its imagistic and material levels. Second, in contrast to words, spontaneous mental images are not signs, but clusters of signs and sentences, featuring spatial and temporal continuity, with no clear distinction between the various syntactic functions of its components. They also carry diffuse signifieds, in the sense of not determining clear boundaries between core senses and diffuse associative peripheries. These limitations make interpersonal communication problematic. Only the mediation of language, which is the main repository of relatively controlled abstractions, is capable of determining the core signifieds and syntactic functions conveyed by an imprinted image. The mediation of language is, therefore, a prerequisite for an iconic medium to become a cultural system. The homogeneity of theatre 175 I assume that such mediation happens spontaneously: a brain conditioned by a language naturally assigns senses to imprinted images, according to the words and syntactic patterns used for categorizing their models. It is precisely in this sense that an imprinted image and the verbal sentences used to describe it or its model are equivalent in different systems of signification. Moreover, in contrast to commonplace perception, I presuppose that a word is not connected primarily to a referent, but to its correlated image, with both being mentally represented in the brain, which explains the spontaneous use of one instead of the other. Consequently, an image becomes a cultural unit of thinking and communication of thinking under two conditions: imprinting on matter and mediation of language. If an iconic unit indeed conveys the set of abstractions culturally connected to the class of its real or mental model, it can be used as a univocal unit of description. Indeed, iconic media are used for describing (usually fictional) worlds. Such descriptions — by means of (imprinted and mediated) images — are made possible by actually disconnecting them from perception. Since performance-texts usually describe fictional worlds, with the text functioning as a clue for imagining them, such imprinted and mediated images function in an evocative capacity. The redefinition of 'iconicity' in terms of imagistic thinking supports the homogeneity of the theatre medium in logically opening the possibility of imprinted images of speech, refuting thereby the alleged heterogeneity of performance-texts, particularly on the grounds of verbal symbolism (in Peirce's sense). I conjecture, therefore, that the production of iconic replicas of speech acts is possible, as it is for any other concrete model in the world. In order to substantiate this thesis, in the following sections I address the question of segmentation with regard to real and iconic verbal interaction. 3. The segmentation of the performance-text Erika Fischer-Lichte appropriately assumes that the '[segmentation of the theatrical text is a fundamental operation in any analysis' (Fischer-Lichte 1992: 224). The notion of 'analysis' actually presupposes the notion of 'segmentation.' A crucial task of performance analysis is, consequently, to determine the elementary segments of a performance-text. 'Text' is defined here as the entire structured set of verbal and nonverbal signs/sentences, meant to be read (decoded), interpreted and experienced by a receiver. Accordingly, 'text' and 'reading' should be understood in a wide sense that applies to all systems of signification and communication. The notion of 'text' thus presupposes a language or medium that affords 176 E. Rozik discrete units and rules of combination capable of generating whole articulated texts. From the inception of structuralist and semiotic-oriented studies, attempts have been made to isolate and define the minimal units of theatre performance-texts. Already in 1942, Veltruský employed the notion of 'segmentation' (Veltruský 1986b: 131). Unfortunately, he approached the problem under the assumption that it is the play-script that, conceived as a literary work, is the object of segmentation (Veltruský 1986b: 130). This fallacy was overcome by later scholars who correctly presupposed that the performance on stage is the actual text of the theatre. However, Veltruský and most subsequent semiotic attempts have still presupposed that the (iconic) sign is its minimal unit. Pavis displays his disappointment with traditional semiotics, which in his view is plagued by 'dogmatic prejudices,' by its search for such 'minimal units' and attempt to 'reconstruct' performance-texts on their grounds (Pavis 1998: 255). Kowzan asserts that '[t]he application of the science of signs [semiotics] to the analysis of the spectacle [peformance-text] requires ... first of all the determination of the significative (or semiological) unit of the spectacle' (Kowzan 1975: 214, my translation). He contends that this unit 'is a slice containing all the signs emitted simultaneously, the duration of which equals that of the sign that lasts least' (Kowzan 1975: 215, my translation). Such a slice assumedly consists of several overlapping units, allegedly from different 'borrowed' codes, such as speech, intonation, facial expression, body posture, and gesture. As we shall see below, although on different grounds, this is quite a felicitous insight. However, it raises several problems: (1) Knowing which sign lasts least presupposes knowledge of principles of segmentation on a simpler level, which is the object of inquiry itself. (2) This principle, meant to segment a text into its significant units, paradoxically avoids signification as its criterion, and is thus merely mechanical. Keir Elam correctly remarks that 'Kow-zan's "slice" has not been found applicable and has produced no textual analyses of note' (Elam 1980: 48). In fact, descriptive texts are composed not by signs, but by sentences. Analogously to language, we may claim that the iconic sentence is both the most complex unit that the theatre medium can generate and, by the same token, the simplest unit of description of a (fictional) world. The notion of 'sentence' implies notions such as 'iconic syntax,' 'subject,' 'predicate,' and 'reference,' which are not pertinent on the level of sign. Unfortunately, no theory of theatre has isolated, described and defined iconic sentences. An imprinted image is never a single iconic sign, but at least an iconic sentence, and at most an 'icon'; i.e., a 'gestalť of iconic sentences (cf. The homogeneity of theatre 111 Pavis 1982: 15). An 'icon' is a complex iconic unit that can be analyzed into a cluster of simultaneous iconic sentences predicated on a single referent; e.g., 'the apple is red, is ripe, is fresh, and is tempting.' Each of these predicates may change on the time axis; e.g., from 'is ripe' to 'is rotten.' While reading these features on the level of their configuration (holistic reading) is equivalent to the subject of an iconic sentence, i.e., its referential function (e.g., the apple), reading them on the level of single qualities, states or acts is equivalent to its predicates, or categorizing function (e.g., 'is red'). All the verbal predicates that can describe such an iconic apple, and are true, can be said to be the icon's descriptive meaning. Similarly to verbal texts, no single iconic sign can describe any referential entity, unless it is included in an iconic sentence and its syntactic function determined. Since a performance-text is a description of characters (subjects) in interaction, it is sensible to conjecture that its typical minimal unit is an iconic sentence that describes a single action. Furthermore, due to the principle of motivation through similarity to real or mental models, an iconic sentence operates two principles of reading: the iconic, on the level of the replicating sign, and the (usually) index-ical, on the level of the replicated sign. Therefore, a preliminary question with regard to the basic iconic sentence that describes a single action is: what are the basic segments of real interaction? 3.1. Segmentation of real interaction Real interaction is an exchange of acts/actions. However, whereas for nonverbal acts this is self-understood, there is a problem with regard to verbal acts, because the commonplace perception is that words are used in a descriptive capacity. However, speech act theory has introduced a crucial conceptual change in this domain. In terms of speech act theory, a 'speech act' is a particular kind of 'act' that employs a verbal expression for performing an action; in John Austin's terms, a case in which '[t]he uttering of the words is, indeed, usually a, or even the, leading incident in the performance of the act' (Austin 1980: 8). Speech acts aim not at describing states of affairs, but at constituting or changing them; e.g., speech acts of command, apology and declaration (Austin 1980; cf. Searle 1979, 1985, 1986; Leech 1983; Levinson 1987; and Lyons 1981, 1988). Because of their performative nature, following Teun van Dijk, speech acts should be analyzed in terms of action theory (Dijk 1977: 167ff, 1980; cf. Rozik 1993). An 'action' is best defined as 'an event brought about by a human being' that aims at changing a state of affairs (Dijk 1977: 173). In Peirce's 178 E. Rozik terms, an act is an index, on the grounds of part/whole contiguity: an act is the perceptible part of an action, which is the whole unit, and reflects at least an 'intention' and a 'purpose.' In a verbal act, in contrast to a verbal description, the utterance of a sentence does not aim at categorizing and referring to a state of affairs, but at constituting or changing it. The performative nature of speech acts thus reveals a basic equivalence between them and nonverbal acts; e.g., between a verbal threat ('I warn you not to do x') and a nonverbal threat (T am wagging my index finger at 'you'). Speech acts are thus indexes that employ verbal means in a nonverbal capacity. This equivalence implies that real dialogue is not an exchange of verbal descriptions, but a specific form of interaction, and explains the natural interchange of verbal and nonverbal acts in a chain of interaction (Rozik 2001). Furthermore, the performance of a real act, and the action it indicates, is an event in the world; i.e., the potential referent of a verbal description. To give an extreme example, a declaration of war is a speech act that, although merely articulated in words, actually involves a country in war and, therefore, can be described as a 'declaration of war.' Less consequential examples materialize the same principle; e.g., a verbal order. An exchange of speech acts, i.e., a dialogue, is basically performed by two human beings who assume the functions of the agent and object of an action in turn, and are alternately referred by T and 'you,' following the flux of interaction. Such an interchange of functions, which is most evident in short and rapid dialogue (stichomythia), also applies to a string of speech acts with the very same orientation (the same T to the same 'you'), whether these are of the same or different kind. Since an action is defined by aiming at changing a state of affairs, 'in-tentionality' is part of its definition. Both speech act and action theories agree on the distinction, although in different terms, between 'intention' and 'purpose' (cf. Dijk 1977: 175ff; Elam 1980: 169).4 An intention is reflected in the nature of an act/action itself; e.g., a promise, which can be verbally articulated by means of a performative verb and/or conveyed by nonverbal indicators (cf. Austin 1980: 73ff). A purpose, in contrast, reflects an aim beyond the nature of an act/action (cf. Austin 1980: 101). Usually purposes are not articulated in words and have to be conjectured on contextual grounds. The deep structure of a speech act reflects three sets of functions: first, a set of performative functions: T (agent), 'you' (object), a 'performative verb' and nonverbal indicators of intention. A performative verb may function in two capacities: (a) for performing an act/action, e.g., a speech act that employs the verb 'promise' for making a promise; and (b) for The homogeneity of theatre 179 describing such an act/action; e.g., the use of the verb 'promise' for categorizing a speech act, whether it features this verb in a performative capacity or not, and whether it is a verbal or nonverbal act (e.g., a gesture of 'trust me') (cf. Austin 1980: 73-76; Lyons 1988: 743). It is the performative function (a) that defines a 'performative verb.' This set of performative functions indicates that an agent performs an act/action upon a (usually human) object by means of words. Second, a set of deictic functions: T and 'you' (their circumstantial aspects), the present indicative of the performative verb ('now') and the usually self-understood 'here.' These functions indicate that an extra-linguistic entity, categorized as T (the agent), performs an act upon another extra-linguistic entity, categorized as 'you' (the object) under the extra-linguistic circumstances of 'here' and 'now.' While T,' 'you,' 'here' and 'now' do not change their sig-nifieds, their referents interchange in the course of an interaction. Third, a descriptive function: although a speech act is not a description, it usually includes a genuine descriptive sentence ('p' or propositional content) whose ultimate meaning is determined by the intention of the action in which it is embedded. Despite the variety of functions, like a nonverbal act, a speech act is an elementary and unitary segment of interaction. Since the actual meaning of an embedded descriptive sentence depends on the underlying intention, it follows that the very same verbal sentence can be employed to perform different acts/actions; e.g., Til take you home.' may indicate a promise, if it is desired by 'you,' or a threat, if it is dreaded by 'you.' Consequently, nonverbal indicators and contextual clues are the only reliable indicators of intention, and their omission makes the embedded sentence ambiguous. In real interaction, most speech acts omit the performative and deictic elements, because of being nonverbally indicated or self-understood, and only the embedded sentences are verbally articulated. In Austin's terms, these are 'primary' speech acts, in contrast to 'explicit' ones that articulate all the components of their deep structure (Austin 1980: 7Iff), including performative verb, deictic elements and 'p.' However, if the nonverbal indicators are performed and taken into account, a speech act is always explicit (Lyons 1988: 743). The performative and deictic functions also characterize nonverbal act/actions, with the descriptive verbal sentence 'p' obviously missing. They are also performed by an agent upon an object, human or otherwise, while their deictic elements (the here and now circumstances of performance) are self-understood. A nonverbal act too is an index of an action, differing from a speech act only in not employing words to the effect of constituting or changing a state of affairs. Moreover, a nonverbal act is as ambiguous as a speech act, in the sense that the same act can 180 E. Rozik indicate different actions, depending on intention; e.g., 'holding somebody's face by the chin' may indicate a threat or an invitation to physical contact, and express either anger or love, with disambiguation relying on additional nonverbal indicators and/or contextual clues. In real interaction, therefore, verbal and nonverbal acts may interchange in any possible order. For both verbal and nonverbal acts, I suggest five nonverbal channels that may indicate the intention of an act: intonation, facial expression (Argyle 1984: 211-250; cf. Birdwhistell 1972), proxemic behavior (Hall 1963: 204-209; cf. Birdwhistell 1972), bodily posture (Argyle 1984: 251-271), and gesture. A gesture is a significant bodily configuration that combines facial expression, proxemic behavior and bodily posture, which usually converge in a specific hand gesture. Hand gestures are usually both culturally established and subliminal. The specific configuration of these channels is probably the best indicator of the nature of an action. I note that (a) different channels provide different means of indication and that there is a tendency to avoid redundancy; (b) some channels may be inactivated; and (c) most nonverbal indexes are culturally conditioned. All these channels generate what is usually called 'body language.' In addition, the very same channels may indicate other aspects of human interaction, such as characterization (e.g., social background, education, and temperament) and momentary attitudes, feelings, and emotions. These are reflected in each particular act/action, but do not affect its performative nature. Since the specific nature of an action cannot be determined on the grounds of its perceptible aspects, but only if its intention and purposes are adequately established, Dijk claims that interpretation is an indispensable phase in actual interaction: An essential component in the definition of action turned out to be the various mental structures 'underlying' the actual doing and its consequences. This means that actions cannot as such be observed, identified and described. We have access to them only by the interpretation of doings. Such observable parts of acts, however, may be highly 'ambiguous.' (Dijk 1977: 182) Interpretation or, rather, categorization of intentions and purposes of an already performed act/action, constitutes a genuine verbal description and a necessary step prior to reaction, if the addressee wishes his/her reaction to fit the intentions and purposes of the former act/action (cf. Leech 1983: 196). Accordingly, a response to an act features two phases: a categorization of the previous act/action and a subsequent reaction in response. The homogeneity of theatre 181 Erroneous interpretation is most plausible in the case of a missing performative verb ('primary' speech acts and nonverbal acts), but is possible even if the performative verb is explicit, because concomitant nonverbal elements may override it, and actually determine the nature of an act/ action. Furthermore, even an error indicates that the object of an action has engaged in interpreting a previous act. The basic structure of interaction is, therefore, a sequence of three basic units: an T action, a 'you' categorization and a 'you' re-action, which in turn becomes an T action, setting off such a sequence all over again. Because of the principle of similarity underlying iconicity it is sensible to conjecture that the iconic description of dramatic interaction, which shows a clear preference for verbal interaction, reflects both the basic structure of real interaction and the inner structure of real acts/actions. 3.2. Segmentation of iconic interaction Roman Ingarden's early attempt to characterize dramatic verbal interaction deserves special attention. He distinguishes four main functions of speech voiced by actors on stage, and attributed to 'represented persons' (characters) within the fictional world: (a) Representation of offstage objects ('represented objectivities'), including characters, by means of the evocative power of words, in addition to the objects represented on stage (Ingarden 1973: 380-381): 'Only when objects and events are only talked about and reported, objects that are situated or take place "off" stage, is the manner in which they are represented and depicted fully the same as in a purely literary work' (Ingarden 1973: 321). Indeed, absent objects or events (offstage temporally and/or spatially) are evoked by words just as in literary works, although the spectator's imagination is conditioned by the concrete reality of the stage, (b) Expression of experiences and psychic states of speaking characters by means of intonation, gesture and facial expression (Ingarden 1973: 321). (c) Communication between characters by means of what is actually said by them; i.e., by the descriptive function of language (Ingarden 1973: 382). (d) Interaction between characters, in which language is employed for affecting one another as in argument or conflict (Ingarden 1973: 321): 'the words spoken by a represented person in a situation signify an act and hence constitute a part of the action, in particular in the confrontations between represented persons. [And by means of such an act] a certain step is made in the development of the action of the "drama"' (Ingarden 1973: 386; cf. Austin 1980).5 He contends that without such verbal acts the fictional action would develop differently. The discovery of this function is indeed of crucial theoretical importance. 182 E. Rozik Ingarden also finds that the interactive function of speaking (d) is typical of both real and fictional dialogue (Ingarden 1973: 388). He suggests, therefore, that much can be learned from dramatic verbal interaction about its real forms (Ingarden 1973: 391). This is an additional felicitous insight. Indeed, there is a basic analogy between real and fictional verbal interaction, in the inner structure of speech acts, and the structure of interaction and its segmentation. The problem is, however, that his conclusions only apply to the use of language within fictional worlds, ignoring thereby the substantial difference between fictional and iconic forms of verbal interaction. For the segmentation of theatre-texts, Alessandro Serpieri et al. adopt a pragmatic approach. Their main claim is that in theatre speech is 'analogous to an utterance geared toward a speaking situation in everyday language, in that it produces meaning in relation to a pragmatic context' (Serpieri et al. 1981: 165). Moreover, 'one can claim that the theatre is entirely performative: indeed, performativity, which is realized in deixis, makes up what might be termed the specific theatrical language' (Serpieri et al. 1981: 168). The application of this approach, I believe, marks a crucial development in theatre theory, despite the following qualifications: First, instead of focusing on the performative elements of dialogue, Serpieri et al. stress its deictic ones and aim at identifying 'performative deictic units' as the basic segments of the text: 'The semiological unit of theatrical language seems to be a complex sign unit identifiable in a given performative-deictic orientation assumed by the speaker ... Thus the semiological unit can be defined as a unit of performative discourse simultaneous with its indexical axis' (Serpieri et al. 1981: 169). 'Deixis' is employed in the usual sense of a word (or morpheme) that refers to a speech act's circumstances of performance (e.g., T,' 'you,' 'here,' and 'now'); and 'orientation' refers to the axis created by a certain agent in acting upon a certain object. Their main thesis is that 'utterances can be segmented at every change of p er formative-deictic orientation by one speaker with regard to the other' (Serpieri et al. 1981: 169). Therefore, instead of looking for the boundaries between performative units in the changes of their performative nature, they opt for delimiting them according to the duration or cessation of a given deictic orientation. However, changes of deictic orientation do not necessarily coincide with the boundaries of single speech acts. A sequence of similar or different speech acts may take place without change of orientation. Therefore, while in some cases such a change is instrumental in detecting and isolating a single speech-act, it is misleading in others. Their fallacy lies in avoiding interpretation in the detection and segmentation of the units of interaction. The homogeneity of theatre 183 Second, instead of actually analyzing performance-texts, Serpieri et al. support their thesis through examples from play-scripts. Initially they reject both the play-script and the performance-text as possible objects of segmentation, the former 'to the extent that it remains "literary"' and the latter 'in so far as it is transient and unrepeatable' (Serpieri et al. 1981: 164). They eventually opt, however, for the play-script by negating its apparent literariness, and stressing its possible performance, which is already inscribed ('stamped') in it (Serpieri et al. 1981: 164ff). They actually illustrate their method of analysis through play-scripts such as Shakespeare's Macbeth and The Tempest. They ignore, however, that in these play-scripts wordings are inherently ambiguous, and that in different productions the same wording can be employed for different speech acts. The problem is that play-scripts usually articulate only primary speech acts, and not the nonverbal indexes of intention that can disambiguate them. In contrast, in performance-texts verbal sentences are always performed with their concomitant nonverbal indicators of intention and, therefore, are always explicit. In particular, performance-texts that reflect creative interpretation actually ascribe new pragmatic meanings to primary speech acts. In other words, Serpieri et al. erroneously presuppose the univocality of primary speech acts in play-scripts. Their fallacy thus lies in ignoring both the inherent ambiguity of speech and the evident reliability of nonverbal indicators. Moreover, their disqualification of performance-texts on the grounds of their ephemeral nature overlooks the use of video recording that, without ignoring deficiencies, overcomes this technical difficulty. I believe, however, that a play-script may be used for segmentation, only if the inherent ambiguity of its dialogue is taken into account. Third, instead of addressing the performative units on the textual level of a performance-text, Serpieri et al. actually segment verbal interaction in fictional worlds. Indeed, similarly to real dialogue, speech is used in such worlds for performing acts/actions, while reflecting the circumstances of their performance (deixis). In this sense, in addressing the analogy between real and fictional speech acts, they revert to Ingarden's fallacy. In contrast, I suggest that performance analysis should first of all address the iconic/descriptive level of a text, because the mental "existence" of fictional interaction depends on its evocative power. This does not preclude analysis on the fictional level. In this respect, their fallacy lies in not distinguishing between iconic and fictional speech acts. Henceforth, I suggest principles of segmentation of the iconic performance-text, while aiming at avoiding these fallacies (Rozik 2001). In the segmentation of stage interaction we face three different orders: (a) the histrionic order — the actors performing iconic verbal and 184 E. Rozik nonverbal acts that they imprint on their own bodies; (b) the textual order — the set of iconic acts inscribed on these bodies that describes acts/ actions in a (usually fictional) world; and (c) the fictional order — the set of fictional acts/actions evoked by a performance-text and assumedly performed by characters. These are the very same segments as perceived from different perspectives. Since segmentation of interaction in the fictional world depends on the decoding of the performance-text, the search for minimal units of description should initially focus on the textual order. As suggested above, whereas the iconic sentence is the maximal and most complex unit that the theatre medium can generate, from the viewpoint of the description of a fictional world such a sentence is the minimal and simplest unit. The basic segment of a description of a fictional interaction is, consequently, an iconic sentence that describes a single fictional verbal or nonverbal act. By sequential aggregation, such units constitute a description of a whole fictional interaction. Since the notion of 'description' presupposes a referent, an iconic descriptive sentence must feature both a subject, i.e., an iconic sign that identifies the referent (the character) that performs an act (holistic reading), and a predicate, i.e., an iconic sign that describes the performed act (partial reading). Due to its imagistic nature, an iconic description is limited to the replication of the sensory aspects of an action, i.e., of an act. Since the only intention is to evoke an action, an actor performs an iconic act with no need to perform the indicated action. In contrast to verbal descriptions, an action cannot be iconically described but, like in real life, has to be inferred from an iconic act. There are no iconic actions; but only iconic acts. I have suggested elsewhere that enacting a character implements the principle of 'deflection of reference,' in the sense of actors producing verbal and nonverbal indexes that, in contrast to the nature of indexes, do not refer to themselves, but to referential characters (Rozik 2002b). This is achieved by producing iconic indexes that evoke human entities (subjects) other than the actors themselves (characters), and iconic indexes of action (acts) that evocatively describe the characters' interacting behavior (predicates). The moment these acts are attributed to characters they reacquire their genuine indexical nature, as indexes of both the fictional actions and the characters that assumedly "perform" them. Actors are not 'pretending' or 'impersonating.' They are engaged in performing not acts/actions, but iconic descriptions of actions; and to be precise, clues for their evocation. A performing actor should be conceived as an icon, and be read on both the holistic and partial levels. An actor performs both a cluster of permanent imprinted images throughout a performance, which aim at The homogeneity of theatre 185 identifying a character, as the subject of all descriptions referring to it (referential function), and a series of interchanging imprinted images of fictional acts, which aim at describing its share in an interaction. A performance-text is predominantly a description of a complex fictional interaction by means of interchanging images of fictional acts imprinted on actors' bodies and deflected to characters. As suggested above, the principle of similarity underlying iconicity implies that some structural qualities are shared by real/fictional and iconic acts. First, the principle of segmentation that applies to real/fictional interaction basically applies to its imprinted image too; second, there is no difference between real/fictional and iconic acts in their inner structure: I + act + you (+optional p); and third, there is usually a total overlap between the boundaries of speech acts and the boundaries of the nonverbal indicators that disambiguate their wordings. Such an overlap corroborates, although on different grounds, Kowzan's intuition that the minimal unit of a performance-text is a slice containing all the signs emitted simultaneously. Despite these similarities, however, the descriptive nature of an iconic act determines a crucial difference. Whereas a real/fictional act is a real event and potentially a referential object of description in a world, an iconic act is a description of such an act in a (usually fictional) world. Therefore, whereas a real/fictional act is equivalent to an T sentence, as all indexes are, the descriptive nature of an iconic act implies that it is equivalent to a 's/he' sentence; e.g., whereas a real/fictional order is equivalent to the sentence T order you that p.,' the imprinted image of such an act is equivalent to 'S/hei (a character) orders s/he2 (another character) that p.' Whereas a real verbal or nonverbal act is transitive, in the sense of 'an I acting upon a you' (and vice versa), an iconic verbal or nonverbal act refers to two possible objects of description 'a s/hei and a s/he2.' The following are the alternative deep structures of real/fictional and iconic acts, whether verbal or nonverbal: A real or fictional nonverbal act: I + act + you. A real or fictional speech act: I + performative verb + you + that 'p' An iconic nonverbal act: S/hei + act + s/he2. An iconic speech act: S/hei + performative verb + s/he2 + that 'p.' These deep structures reflect the significant difference between being a real/fictional act/action and being a description of it. An imprinted image of a verbal or nonverbal act thus constitutes a discrete and distinct iconic sentence, which should be considered as an 186 E. Rozik elementary segment of iconic interaction. The notion of 'iconic sentence' equally applies to descriptions of fictional qualities. In addition to its own nonverbal quality, theatre iconicity also reflects the nonverbal nature of the described verbal and/or nonverbal acts/actions. The nonverbal nature of iconic speech acts contradicts the alleged duality of performance-texts, and substantiates the thesis that performance-texts are generated by a homogeneous theatre medium. 4. Segmentation of iconic interaction in Habimah's The Seagull I analyze here the dialogue between Arkadina and Trigorin, in act III of Anton Chekhov's The Seagull, in the Habimah 1992 production.6 A comparison of this scene in the play-script and the performance-text reveals that, in contrast to the former, in the latter the nature of each speech act is determined by means of nonverbal indicators as in real life. Although the director opted for minimal changes in the verbal script, and adopted the few original stage directions, he ascribed additional nonverbal indicators to most primary speech acts. All these choices were subordinated to a predominantly melodramatic mood. This scene reflects the two opposing motives of Trigorin and Arkadina, but only she accomplishes her purposes. She manipulates Trigorin's weak personality, with each change in her tactics affecting his behavior. This scene can, therefore, be divided into three parts. In the first, Trigorin confronts Arkadina in an attempt to persuade her to accept his attraction to Nina. He repeatedly asks her consent to remain with Nina after Arkadina leaves, while she, increasingly infuriated, absolutely refuses. In the second part, Arkadina changes her tactics. She flatters him until he renounces his request. In the third part, changing tactics again, Arkadina 'permits' him to stay, while he refuses. She 'allows' him to make the final decision, thus befooling his male self-image. The transitions from one part to the other are marked by extreme changes of conduct: e.g., from harsh blame to exaggerated flattery. In its performed version, this scene illustrates the main theses of this study, as in the following examples: 1) Trigorin: You must try to be sober, too — and sensible and reasonable. Do try to see all this like a true friend, I implore you ... [Presses her hand.] You are capable of sacrifice ... Be a friend to me, release me ... (Chekhov 1960: 160, my emphasis) The actor performed these speech acts while holding the actress' hands and pressing them against his chest, with an urging intonation. He also The homogeneity of theatre 187 repeated 'release me.' In this series of speech acts of entreaty the performative verb 'implore' was employed only once. Moreover, in the entire scene this is the only explicit performative verb. Being an exception, this example underlines the extensive use of primary speech acts on stage, making the use of nonverbal indicators indispensable. Although Chekhov's few stage directions illustrate this need, most underlying intentions were specified in the directing process. 2) Arkadina: Am I really so old and ugly that you can talk to me about other women without embarrassment? [Embraces and kisses him.] Oh, you must have gone mad! My beautiful, my wonderful ... You — the last page of my life! [Kneels before him.] My joy, my pride, my happiness!... [Embraces his knees.] If you leave me even for a single hour I shall never survive it, I shall go out of my mind — my wonderful, magnificent man, my master ... (Chekhov 1960: 161) In contrast to Serpieri et al., these speech acts of blame, flattery, entreaty, and warning/threat were performed without a change of orientation, requiring further segmentation on interpretive grounds. In addition, stressing Arkadina's change of tactics ('My beautiful...'), the actress performed a gesture of tapping her nose, while slightly smiling to herself, folding her sleeves and lifting Trigorin's head, thus indicating her ironic intention to deliberately manipulate him. 3) Trigorin: Take me, carry me off, but don't let me ever move a step away from you ... (Chekhov 1960: 162) This primary speech act, which was performed with an intonation of warning, could have equally been performed as an act of entreaty. This illustrates the inherent ambiguity of play-scripts: the same scripted words can be given different interpretations and it is the nonverbal elements that ultimately determine their nature. 4) Trigorin: No, we may as well go together. (Chekhov 1960: 162) Trigorin responds to Arkadina's last change of tactics, 'permitting' him to stay, and 'allowing' him to take the final decision. The actor accompanied this speech act by swallowing a glass of vodka in one gulp, creating an image of manliness, as if the final decision was his own. Although this verbal sentence could have been interpreted as a speech act of submission, this nonverbal act reflects Trigorin's intention to put a mannish end to the discussion, thus ironically marking her success. All these speech acts reflect a common purpose: to persuade one another. Trigorin strives to get Arkadina's consent and she to dissuade him. Eventually, at least momentarily, she succeeds in manipulating him to the 188 E. Rozik point of believing that it was his own decision. We know that subsequently Trigorin will join Nina, and eventually return to Arkadina. Whereas in the play-script most wordings would appear to be univocal and self-explanatory, in principle, the kinds of speech acts of which these could be part may vary, as demonstrated by this scene. In most of these examples the duration of the nonverbal indicators of intention overlap that of the embedded sentence 'p.' In a few cases, some nonverbal indicators were performed before the beginning of the verbal sentence, but there was total overlap at the end. This may imply that 'psychological' reaction begins before actual response. 5. Conclusions The notion of 'iconicity,' redefined in terms of imagistic thinking, in the sense of images of real or imagined models imprinted on (similar) matter and mediated by language, equally applies to images of both nonverbal and speech acts units, which combine verbal and nonverbal elements into indivisible nonverbal units. Iconicity thus transmutes units 'borrowed' from other systems of signification into descriptions of such units. The principles of similarity and mediation that underlie 'iconicity' also enable the segmentation of iconic and fictional characterization and interaction according to the verbal categories that segment their real models. A performance-text is, therefore, a nonverbal artifact on two accounts: the describing segments, due to their underlying iconic principle, and the described segments, due to the nonverbal nature of acts/actions, including speech acts. Consequently, in contrast to the traditional claim that performance-texts are generated by a (multiple or dual) set of heterogeneous codes, the principle of iconicity bestows homogeneity on all performance-texts generated by the nonverbal theatre medium. Notes 1. In Saussure's terms: 'signifianť and 'signifié' (1972: 101). 2. 'Symbol' is employed by Langer in the sense of 'sign.' 3. This principle also applies to stage conventions, which apparently abolish the similarity that characterizes iconicity. Spectators could not perceive them unless they were images imprinted on matter. 4. In Austin's terms: 'illocutionary force' and 'perlocutionary effects,' respectively. 5. Austin's performative theory of speech was published in 1962 (his lectures were delivered in 1955), and Ingarden's 'Appendix on the functions of language in the theater' was first published in 1958, and its English translation in 1973. The homogeneity of theatre 189 6. This production was directed by Boris Morozov, with Michail Kozákov as Trigorin, Shuli Rand as Konstantin, and Jetta Monte as Arkadina. References Argyle, Michael (1984). Bodily Communication. London: Methuen. Austin, J. L. (1980). How to Do Things with Words. London: Oxford University Press. Birdwhistell, R. L. (1972). Kinesics and Context. Philadelphia: University of Pennsylvania Press. Bogatyrev, Petr (1986). Semiotics in the folk theater. In Semiotics of Art, L. Matějka and I. R. Titunik (eds.), 33-49. Cambridge, MA: MIT Press. Carlson, Marvin (1990). Theatre Semiotics: Signs of Life. Bloomington: Indiana University Press. Chekhov, Anton (1960). The Seagull. In Chekhov's Plays, E. Fen (trans.). Harmondsworth: Penguin. Damasio, Antonio R. (1994). Descartes' Error. New York: Grosset/Putnam. De Marinis, Marco (1993). The Semiotics of Performance, Aine O'Healy (trans.). Bloomington: Indiana University Press. Dijk, Teun A. van (1977). Text and Context. London/New York: Longman. —(1980). Macrostructures. Hillsdale, NJ: Erlbaum. Elam, Keir (1980). The Semiotics of Theatre and Drama. London/New York: Methuen. Esslin, Martin (1992). The Field of Drama. London: Methuen. Fischer-Lichte, Erika (1992). The Semiotics of Theater, J. Gaines and D. L. Jones (trans.). Bloomington: Indiana University Press. Hall, Edward T. (1963). Proxemics: A Study of Man's Spatial Relations. New York: International University Press. Ingarden, Roman (1973). The Literary Work of Art, G. G. Grabowicz (trans.). Evanston: Northwestern University Press. Kosslyn, Stephen M. (1995). Image and Brain. Cambridge, MA: MIT Press. —(1996). Introduction. In The Cognitive Neurosciences, M. S. Gazzaniga (ed.), 959-961. Cambridge, MA: MIT Press. Kowzan, Tadeusz (1975). Littérature et Spectacle. The Hague/Paris: Mouton. Langer, Susanne K. (1996 [1942]). Philosophy in a New Key. Cambridge, MA: Harvard University Press. Leech, Geoffrey N. (1983). Principles of Pragmatics. London/New York: Longman. Levinson, Stephen C. (1987). Pragmatics. Cambridge: Cambridge University Press. Lyons, John (1981). Language Meaning and Context. Bungay: Fontana. —(1988). Semantics, vol. 2. Cambridge: Cambridge University Press. Mukařovský, Jan (1986). Art as semiotic fact. In Semiotics of Art, L. Matějka and I. R. Titunik (eds.), 3-9. Cambridge, MA: MIT Press. Pavis, Patrice (1982). The Languages of the Stage: Essays in the Semiology of the Theatre. New York: Performing Arts Journal. —(1998). Dictionary of the Theatre: Terms, Concepts and Analysis, C. Shantz (trans.). Toronto: University of Toronto Press. Peirce, Charles S. (1931-1966). The Collected Papers of Charles S. Peirce, 8 vols., C. Hartshorne, P. Weiss, and A. W. Burks (eds.), Cambridge: Harvard University Press. [Reference to Peirce's papers will be designated CP followed by volume and page number.] 190 E. Rozik Rozik, Eli (1993). Categorization of speech acts in play and performance analysis. Dramatic Theory and Criticism 8 (1), 117-132. —(2001). Segmentation of the theatrical performance-text. Semiotica 135 (1/4), 77-99. —(2002a). The Roots of Theatre: Rethinking Ritual and Other Theories of Origin. Iowa City: University of Iowa Press. —(2002b). Acting: The quintessence of theatricality. Substance 31 (2-3), 110-124. Saussure, Ferdinand de (1972). Cours de Linguistique Generale. Paris: Payot. Searle, John R. (1979). What is a speech act? In The Philosophy of Language, J. R. Searle (ed.), 39-53. London: Oxford University Press. —(1985). Speech Acts. Cambridge University Press. —(1986). Expression and Meaning. Cambridge University Press. Sebeok, Thomas A. (1975). Six species of signs: Some propositions and strictures. Semiotica 13 (3), 233-260. Serpieri, Alessandro, et al. (1981). Toward a segmentation of the dramatic text. Poetics Today 2 (3), 163-200. Ubersfeld, Anne (1999). Reading Theatre, F. Collins (trans.). Toronto: University of Toronto Press. Veltruský, Jiří (1964). Man and object in the theatre. In A Prague School Reader on Esthetics, Literary Structure, and Style, Paul L. Garvin (ed. and trans.), 83-91. Washington: Georgetown University Press. —(1986a). Dramatic text as a component of theater. In Semiotics of Art, L. Matějka and I. R. Titunik (eds.), 94-117. Cambridge, MA: MIT Press. —(1986b). Basic features of dramatic dialogue. In Semiotics of Art, L. Matějka and I. R. Titunik (eds.), 128-133. Cambridge, MA: MIT Press. Eli Rozik (b. 1932) is Professor Emeritus at Tel Aviv University . His research interests include theatre semiotics, nonverbal communication in theatre, performance analysis and metaphor in particular, and poetics. His recent publications include 'Acting: the quintessence of theatricality' (2002); 'Framing, decoding and interpretation: On the spectator's vital role in creating theatrical meaning' (2002); 'Origins of theatre' (2003); and 'Theatrical experience as metaphor' (2004).