“Kubla Khan” and the Embodied Mind

William L. Benzon

Coleridge's "Kubla Khan" is one of the most important lyric poems in the English language. When Coleridge published the poem in 1816 he prefaced it with the story that it was but an incomplete record of an opium-induced vision. Tossed and torn between that preface and the poem itself, scholars have written numerous articles around and about the poem, and even a monograph and a book or two. Of equal significance, "Kubla Khan" is frequently anthologized, is taught in secondary and post-secondary schools, and has managed to secure a tenuous foothold in the popular culture of the English-speaking world. Orson Welles placed the poem's first five lines on the screen in the early moments of his classic "Citizen Kane," a film about a news baron who died at an estate called Xanadu. More recently Ted Nelson has been exploring hypertext in his Xanadu Project, Olivia Newton-John sang a hit song entitled "Xanadu," the rock group Rush reworked Coleridge's imagery into another "Xanadu,"¹ and "Kubla Khan" was presented in a pop-up book (Bantok 1994).

     Meanwhile, back at the academy, the cognitive sciences have been remaking psychology through the metaphor of computing and neuroscientists have developed techniques allowing us miraculous visions of the living brain. As these scientists began to think about the arts - Herbert Simon (1994) on the logic of literary criticism, Semir Zeki on Inner Vision: An Exploration of Art and the Brain, Gordon Shaw (2000) on the "Mozart effect" - literary critics have turned toward these newer psychologies. Mark Turner's work is perhaps the best known, but Reuven Tsur has been cultivating his own brand of cognitive poetics, Norman Holland has interrogated The Brain of Robert Frost and other thinkers have made contributions as well. My own interest in cognition, the brain, and literature was prompted by my early graduate studies when I decided to take the methods of classical structuralism of Levi-Strauss, Roman Jakobson, and Jean Piaget and apply them to "Kubla Khan." I was surprised by the results of that analysis, which revealed an elaborate and highly symmetrical structure in this poem that Coleridge had presented to the world as a "fragment...a psychological curiosity" of doubtful aesthetic value.²

     The purpose of this essay is to bring that analysis up to date by incorporating some recent intellectual developments, particularly in cognitive linguistics, and then to consider some of the implications of that analysis. In the next section I present my methodological tool kit, followed by three sections in which I apply those tools to the text: "Xanadu," "Enchantment," and "Sound and Sense." Then I switch into a more speculative mode and think about how a brain might produce or make sense of the poem. I conclude with some general observations about the implications of these methods for literary studies.

Methodological Preliminaries

     Given that language prompts us to construct meanings, each of us in accordance with our own history and experience, the purpose of my analysis of "Kubla Khan" is to characterize what is there, either in the text, the words themselves, or tightly bound to it by linguistic and cultural conventions. Whether or not any given person perceives the prompts is a different question, and one well beyond the reach of this essay.

     My first task is to state a standard of objectivity. After I have done that I briefly characterize the various concepts and techniques I have employed in analyzing the poem. I conclude with an explanation of diagramming conventions.

     Some things are as uninteresting as they are obvious, for example, that the poem exists in two sections, one 36 lines long and the other 18 lines long. Nor is the rhyme scheme terribly problematic. Moreover, one can, by such simple means as counting periods and commas and noting whether imagery is visual or oral, discern patterns in the text that are not so obvious. Yet I will argue that those patterns are there, in the linguistic material. But once I begin to describe patterns of meaning -- whether or words, phrases, sentences, or groups of sentences -- I move into dangerous territory. As a way of characterizing my sense of how to draw the line between what is really there and what is not I offer a statement from Reuven Tsur's (1987) monograph on "Kubla Khan":

     That the caverns are there and that they are opposed to the "sunny pleasure-dome" is given in the poem: but that . . . this opposition reflects the conflict of heaven and hell, is hypothesized by the critic. 

     The opposition between the caverns - which contain the sunless sea - and the pleasure-dome is in the ordinary meanings of those words; it is part of the language system. However each of us may construe "sunless sea" and "pleasure-dome," all of us construe them in semantically opposite ways. The language system does not, so far as I know, have a conventionalized link between heaven and sunny pleasure domes.

     This standard is, admittedly, a vague one. I expect that my judgments on some matters will be contested. I believe, however, that the analysis I propose demonstrates an overall pattern of relationships between sound and sense that is undeniably there.

     I am thus claiming that my analysis is an objective one. Given that objectivity has been under strenuous intellectual attack for the past quarter of a century - at least - that is a tendentious claim, especially in a domain so notorious for its subjectivity as is literary studies. I am not being deliberately obtuse in making this claim, but this is not the place for epistemological debate. I would, however, like to clarify a few points.

     In claiming objectivity I am not claiming to have either an innocent ear, an open mind, nor a pure heart when analyzing "Kubla Khan." I do not believe that objectivity can be obtained by generous applications of good will and hard work. Objectivity is obtained by explicit and sharable methods. Only experience can tell whether any particular tool kit yields objective results. Most of the tools in my kit have been around for awhile, though the kit as a whole is novel. When I claim that I am showing what "is given in the poem" I am claim only that a group of skilled people using these intellectual tools will be able to reach substantial agreement. That agreement may well differ from what I set forth here. As my method is novel and untried, it is possible that others may have deeper insights into its use than I do. I invite their arguments.

     This claim to objectivity, however, does not extend to the more speculative work that follows my analysis. Whether or not those ideas are valid has little bearing on the validity of this analysis. The purpose of those speculations is to help explain the analytic results, not to justify or support them. Those who, for whatever reason, reject the assumptions and methods I outline in the rest of this section effectively reject my claim that the analysis is an objective one. While such a person might, in principle, be more favorably disposed to my speculative essay into psychological explanation, that would be a most peculiar, if logically possible, position.

     Here then, in brief, are my assumptions and my tools. While my major objective is to provide insight into "Kubla Khan," I am also interested in knitting these various ideas together into a single fabric. That job is so taxing that I have little choice but to assume that the reader already has sufficient familiarity with these notions that he or she does not need a full exposition either here or in the main text.

     From cognitivism in general, and computational linguistics in particular, I adopt the view that time is intrinsic to our understanding of linguistic and psychological processes. I learned this from the late David Hays, who learned it while working on machine translation at the Rand Corporation during the late 1950s and into the 1960s. All computation takes place in time. This is obviously true of real computation, whether it is a child doing sums or a supercomputer simulating an atomic explosion. But this is also true of the abstract computing processes examined by mathematicians through formal axiomatic methods. Time is intrinsic to this analysis. For example, the computational equivalent to Gödel's incompleteness theorem concerns whether or not a certain computation will, in principle, come to a halt (cf. Arbib 1964).

     The idea that time be taken into account is not new to literary studies. Stanley Fish (1980) advocated it in his well-known essay on "Affective Stylistics." But the idea has not, with few exceptions (e.g. Cureton 1992), been systematically exploited.

     The temporality of literary experience has one consequence which I think is quite important, but will not attempt to utilize in my analysis. "Kubla Khan" is only 54 lines long and can be recited aloud, at a reasonable pace, in two to two-and-a-half minutes.³ Whatever meanings it has must do work their magic in that period of time. One may practice the poem as a musician practices the score, one may contemplate it, read it many times, scour the critical literature, and so forth. At best, that only prepares the mind to be stimulated by the text. The practice and rumination do not constitute that experience. Whatever meanings "Kubla Khan" may have, they are, as Coleridge asserted in his preface to the poem "like the images on the surface of a stream into which a stone has been cast, but, alas! without the after restoration of the latter!" This is, of course, true of all poems.

     As I have already indicated, my early training was in classical structuralism and semiotics. That is a huge and diverse territory. This is certainly not the place to review, let alone justify, those methods. I have been influenced by that general approach and, for better or worse, have an affinity for the attention to detail and system required by it.

     More specifically, I have been very influenced by Roman Jakobson's conception of language and, in particular, the analysis of language functions in his 1960 essay on "Linguistics and Poetics." I also owe a debt to Levi-Straus; in particular, to his techniques of analyzing myths in terms of binary oppositions (nature vs. culture, male vs. female) and so forth (1967, pp. 202-228, 1976, pp. 146-197). Whatever one may think about the theorizing that takes place around and about the ethnographic analysis - I have little interest, for example, in Levi-Strauss's assertions that his myth analyses are further tellings of those tales - it is the analyses themselves that are important. One does not learn how to analyze myths from reading theory, one learns it from seeing it done. My impression is that, in such matters, theory generally lags behind analysis.

     Cognitive linguistics is a collection of various ideas having a loose affinity with one another. I am generally sympathetic to the affinity though I would not count myself as a cognitive linguist. Thus I believe that the structures and processes of language are to be accounted for by the cognitive and communicative work they do, not by reference to abstract formal systems. The meaning of a word is in what it prompts you to notice or to do.

     Among these ideas, I have found that of mental spaces and conceptual blending particularly useful. Both ideas are inherently temporal. A mental space (Fauconnier 1994) is a temporary assemblage of perceptual, cognitive, emotive, and motor elements that arise to support some current activity. The individual elements in mental spaces are all part of a person's more-or-less permanent repertoire, but the combination employed at any moment may be novel. Mental spaces thus represent more or less temporary conjunctions of perceptions, ideas, feelings, motivations, and so forth.

     Mental spaces can give rise to conceptual blends (Fauconnier and Turner 2002). The idea descends, I believe, from Lakoff and Johnson's (1980) account of cognitive metaphor, but is more general in that cognitive metaphor employs mappings between only two domains at a time, where blending can involve more than two domains. But it is also a different notion in that the blend is a temporary construction while a metaphor mapping is a more-or-less permanent part of one's mental machinery. Metaphor mappings can provide content for mental spaces and, conversely, a frequently repeated conceptual blend can become entrenched as a metaphor mapping.

     Given that cognitive metaphor theory may be the aspect of cognitive linguistics best known to literary critics, one might reasonably expect it to play a prominent role in the analysis of such a highly allusive and richly figured poem as "Kubla Khan." Cognitive metaphor does not, however, seem to underlie the poem's most salient features and so the critic expecting its use will find my analysis disappointing. I may, of course, be wrong in my judgment on this matter.

     When a music analyst says that a blues consists of three segments of four-bars each, or that a so-called popular "standard" consists of four segments of eight-bars each, she is talking about the constituent structure of these musical forms. Constituents, of course, may themselves have constituents. In popular music eight-bar phrases are likely to consist of two four-bar phrases, which, in turn, will consist of two two-bar phrases.

     Similarly, linguists analyze sentences into constituent parts, such things as noun phrases, verb phrases, prepositional phrases and so forth, down to the level of individual nouns, adjectives, prepositions, and so forth. The emergence of transformational grammar in the late 1950s and early 1960s inspired psychologists to see whether or not abstract grammatical structures affected the language comprehension process (Neisser 1966: 259 ff., Taylor 1976: 105 ff.). Psychologists of music have undertaken similar studies (see Sloboda 1985). In both cases, constituent structure was shown to have psychological effects. Constituent structure is thus no mere analytic fiction, but reflects real mental processes underlying the temporal structure of language and music.

     Thus my analysis of "Kubla Khan" starts with an analysis of its constituent structure, from the whole poem down to the level or two- or three-line segments. Much, but not all, of this structure can be determined by examining line-end punctuation. This is the most objective component of the analysis.

     Considered as a unit of analysis, the line is a conjunction of units of thought, or sense, and units of physical realization - speaking and hearing.

     The significance of the poetic line is easily demonstrated by the common experiment of taking some fragment of ordinary prose and breaking it into separate lines. The result is rarely good poetry, but the poetry-like presentation invites one to consider each line both as a unit by itself in addition to its connections with the lines before and after. The quasi-autonomy of the poetic line belongs to the cultural conventions governing how we read poetry. The psychological, not to mention the neural, underpinnings of this effect are, as far as I know, obscure.

     Nonetheless, the linguist Wallace Chafe has quite a bit to say about what he calls an intonation unit, and that seems germane to any consideration of the poetic line. In Discourse, Consciousness, and Time Chafe asserts that the intonation unit is "a unit of mental and linguistic processing" (Chafe 1994, pp. 55 ff. 290 ff.). He begins developing the notion by discussing breathing and speech (p. 57): "Anyone who listens objectively to speech will quickly notice that is not produced in a continuous, uninterrupted flow but in spurts. This quality of language is, among other things, a biological necessity." He goes on to observe that "this physiological requirement operates in happy synchrony with some basic functional segmentations of discourse," namely "that each intonation unit verbalizes the information active in the speaker's mind at its onset" (p. 63).

     While it is not obvious to me just what Chafe means here, I offer a crude analogy to indicate what I understand to be the case. Speaking is a bit like fishing, you toss the line in expectation of catching a fish. But you do not really know what you will hook. Sometimes you get a fish, but you may also get nothing, or an old rubber boot. In this analogy, syntax is like tossing the line while semantics is reeling in the fish, or the boot. The syntactic toss is made with respect to your current position in the discourse (i.e. the current state of the system). You are seeking a certain kind of meaning in relation to where you are now.

     Chafe identifies three different kinds of intonation units. Substantive units tend to be roughly five words long on average and, as the term suggests, present the substance of one's thought. Regulatory units are generally a word or so long (e.g. and then, maybe, mhm, oh, and so forth), and serve to regulate the flow of ideas, rather than to present their substance. Given these durations, a single line of poetry can readily encompass a substantive unit or both a substantive and a regulatory unit.

     The third kind of unit, fragmentary, results when one of the other types is aborted in mid-execution. That is to say, one is always listening to one's own speech and is never quite sure, at the outset of a phrase, whether or not one's toss of the syntactic line will reel-in the right fish. If things do not go as intended, the phrase may be aborted. Fragments do not concern us, as we are dealing with a text that has been thought-out and, presumably, edited, rather than with free speech, which is what Chafe studied.

     Chafe's notion is consistent with an observation made initially by Ernst Pöppel. After reviewing studies by others and offering some of his own, Pöppel concluded that our awareness of the present extends roughly three to four seconds. That suggested that lines of poetry last no longer than that and that, where written lines appeared to take longer to read, they have a strong break in the middle. Working with a poet and critic, Frederick Turner, Pöppel found evidence for these notions in the poetry of several cultures, thus showing how versification technique deals with this constraint (cf. Turner and Pöppel 1983, Pöppel 1985, pp. 75-82).

     Forgetting about meaning, now, and considering the line as a unit of organized sound, we find the line defined as a certain number of metric feet, or measures. These units are, in turn, defined by stressed and unstressed or long and short syllables according to the versification system. The number of feet that constitutes a line varies according to the formal specifications of a particular verse form. These details need not concern us.

     What does concern us, however, is that meter is periodic. Periodicity is fundamental to the operations of the human nervous system. I have argued, in Beethoven's Anvil: Music in Mind and Culture, that music involves a linkage between two streams a neural activity, a periodic one, and an aperiodic one that evolves from an initial to a final state. So it is with poetry as well, and the line is the focal point of the coordination between these two streams, a coordination with Coleridge manipulates in critical ways in the "Kubla Khan."

     Finally, we should remember that, however deeply the sign may be split between signified and signifier, in the brain, it is all electrochemical activity in neural tissue. The tissue that serves speaking sounds and hearing them is much like the tissue that serves ideas, perceptions, and feelings. In the brain, the signifier and the signified dance to rhythms beaten out by the same neural drummers.

     Finally, I assume that we are to understand poetry as the creation of a human nervous system. If one believes, as I do, that "the mind is what the brain does"(Kosslyn and Koenig, 1995, p. 4) this is not much of an assumption to make. If it isn't the brain, than what could it possibly be? No, it could not be otherwise, not to thinkers at the beginning of the 21st century.

     But, given that the neurosciences have amassed more data and observations than knowledge, it does not seem that we gain much, if anything, by making this assumption explicit. We are certainly not in a position to describe, in any illuminating detail, how poetic meanings evolve in brains. Further, I am tempted to assert that the analytical portion of my argument does not invoke the nervous system in any fundamental way. That is not, however, quite the case. For my analysis does depend on the fact that poems unfold in time, especially when I consider the poem's second movement. The stress I place on the poetic line, that too depends on temporality. Pöppel's four-second limit for immediate awareness is a fact of how the brain operates. We may not know exactly what it is about the brain that results in this limit, but the limit is surely neural. A commitment to the temporal nature of poetic experience implies an acceptance of the material substrate of that experience, the dynamics of the human nervous system.

     Once I move beyond analysis to some speculation about some underlying psychological mechanisms I do explicitly talk about the nervous system. While my argument does not depend on highly specific details, it certainly does depend on the fact that we have a nervous system, that is it part of our body, and that it regulates the actions of that body. That is not much, but my speculations about the semantics of Xanadu depend on just that. Finally, I offer some more specific speculation that invokes the limbic system.

     Rather than explain my diagramming conventions here and there in the main body of the essay I will explain the conventions for the most complex kinds of diagrams here. Where a diagram is of a different sort, I will provide further explanations as necessary. Figure 1 illustrates the conventions of cognitive diagrams:

Figure 1: Cognitive Diagram

     A rectangle with a single narrow border is a mental space in the sense of Fauconnier and Turner. Mental spaces may be nested. The nodes in a semantic field represent concepts while the links between the nodes represent relationships between concepts.⁴ A node's label names the concept it represents. Notice that one node is simply labeled with an "X." That node has an arrow extending from it to another mental space. That mental space provides the meaning of the node at the tail end of the arrow. As a matter of convenience I will sometimes use ordinary language phrases to represent the content of a mental space, as in the upper part of this diagram.

     A rectangle with a double border is a line, or a continuous group of lines. A line is necessarily a mental space, but not all mental spaces constitute lines. Lines have necessarily been rendered into language, but the contents of mental spaces are not necessarily linguistic. The semantic content of a line may include a conceptual blend. I list lines in the order they appear in the poem, from top to bottom and include a word or two to identify by the semantic content of the line. A dotted connection between lines or between a line and a mental space means that the connected elements share content from the same semantic domain, whatever that may be.

     Figure 2 illustrates a rather different kind of diagram, the intentional diagram: 

Figure 2: Intentional Diagram

     This diagram is a bit peculiar, and deliberately so. It depicts the central nervous system operating in two worlds: the body's interior milieu (right) and the external physical world (left). While the nervous system is physically of a piece with the body, it has a relationship with the body's interior that is similar to that which it has to the external world. Just as the nervous system is aware of the external world through a variety of sensors and operates in that world through the motor system, so it is aware of the interior milieu through sensors in the autonomic nervous system and operates on that milieu through the autonomic nervous system and the endocrine system.

     The middle box represents the Jill's central nervous system (CNS). The double arrow to the left indicates physical interactions between Jill's brain and the external world. Energies impinge on Jill's sense organs and are transduced into electrochemical signals in her nervous system; conversely, neural signals direct the muscles to tense or relax, and they allow Jill to affect the world. In a similar fashion the double arrow to the right indicates physical interactions between the CNS and the interior milieu. The CNS is aware of the interior milieu through sensors in the autonomic nervous system and it operates on that state through the autonomic nervous system and the endocrine system.

     Now, if that is Jill's nervous system at the center, then why is Jill also depicted there to the left, in the external world, and apparently on a par with Jack? From her brain's point of view, Jill herself is also a creature in the external world. She can see her body, touch it, hear her voice, and so forth. Her body exists out there in the world where it is on a par with other bodies, such as Jack's. Her body (and personal history as well) is represented within her nervous system by the neural self (NS) (Damasio 1994, 1999). The solid arrow from the interior milieu to Jill-in-the external world represents the perceptible effects that milieu has on Jill's body. Obviously Jill has a privileged relationship with her own body that is different from her relationship with Jack's. There is not going to be any solid arrow from Jill's interior to Jack's body nor is Jill's neural-self going to be moving Jack's body.

     Finally, the dotted lines indicate intentional relationships between Jill and objects and processes in the external world. These are not relationships of physical causality. Rather, they reflect the meaning that Jill confers on those things by virtue of her history and the current state of her needs and desires. I have the notions of intentionality and meaning from Walter Freeman's (1995, 1999) discussion of the nervous system, though both are consistent with at least some standard philosophical usage. The relationship between the neural self and Jill is one of identification. In pathological states that identification may be broken but it may also be deliberately set aside for aesthetic or ritual purposes (cf. Benzon 2000). The relationship between the external and internal rose and the internal and external Jack is one of reference. Physical processes allow Jill to see, smell, and touch the rose, but her intentional relationship to the rose is given only in the internal dynamics of her nervous system. It is that relationship that is indicated by the dotted line.

Xanadu

     First I briefly the poem's two movements, then I analyze the first movements into its three main constituents. After that I undertake a detailed analysis of the line structure of the first two sections and conclude with a similarly detailed analysis of the final section.

     "Kubla Khan" is 54 lines long and consists of two movements.⁵ The first movement presents us with the sights, sounds, and sensations of Xanadu, Kubla, Alph, and the fountain while the second movement presents us with the thoughts and speech of an unnamed poet and his audience. These are radically disjoint worlds whose only apparent connection is through the juxtaposition of the dome and the caves that occurs in lines 36 and 47.

     Beyond this, there is a very specific linguistic difference between the two movements in the use of pronouns (see Table 1). Movement 1 (lines 1-36) has only four pronouns while Movement 2 (lines 37-54) has sixteen. This suggests that the cognitive structures sustaining self-presence, which are linguistically realized through personal pronouns, are strongly operative in Movement 2, but not in Movement 1.  

Table 1: Pronouns in "Kubla Khan"

     I will demonstrate that the two movements of "Kubla Khan" each consist of three sections which are ontologically different. I do not, however, mean ontology in its normal philosophical sense of pertaining to the ultimate nature of the world. Rather, I mean it in the sense that has become common in cognitive science, particularly in the study of knowledge representation. In this sense an ontology is the set of basic objects and processes in some domain. One needs such an ontology either to program a computer to operate in some domain or to account for how people think about and act in a domain. The specification of such an ontology makes no commitment to what is ultimately real about some domain; hence it is not a metaphysical notion. Rather, it is a psychological (humans) or informatic (computers) notion.

     For example, in the early 1980s David Warren and Fernando Pereia created an experimental computer system for answering questions about geography. Their ontology included areas, points, and lines as kinds of geographical feature, block, terrain, country, wetland, and mountain as kinds of area, and appropriate subcategories for points and lines. These categories are not understood as being what is ultimately real about geography. Rather, they are simply categories useful in describing geography (Sowa 2000, pp. 52-53). Similarly, when Harry Jerison (1976, p. 99) considered the long-term evolution of vertebrate nervous systems he argued that

     Reality, or the real world we know intuitively, is a creation of the nervous system: a model of a possible world, which enables the nervous system to handle the enormous amount of information it receives and processes. . . . The "true" or "real" world is specific to a species and is dependent on how the brain of that species works; this is as true for our own world--the world as we know it--as it is for the world of any species.

      Jerison is not making assertions about how the world ultimately is. He is only noting the relationship between an animal's nervous system and the world they experience.

     That is the sense in which I refer to ontology in the world of "Kubla Khan." The text is structured into sections such that different kinds of things exist in different sections, different kinds of actions or processes are possible in different sections. These differences have to do with sensory properties and with fundamental possibilities for action and so are ontological in the sense that cognitive scientists use the term.

     The first movement of "Kubla Khan" is allocentric in that the narrative voice is aware only of Xanadu and events happening there. As it unfolds we move from a visual-volitional domain, through an auditory-expressive domain, to a domain of perceptions (that has critical links to the first two domains). The second movement of the poem is egocentric in that the narrative voice is self-aware. It begins in a domain of recollection, moves to one of supposition, and ends in one of command and assertion. In both movements of the poem units of rhyme structure are aligned with units of semantic structure in the first and third sections, but not the second section. In the second section rhyme and sense become desynchronized.

     My analytic method begins with an analysis of the two movements into the constituent parts I have just indicated, and their subdivisions down to the level of couplets, triplets or, in some cases, single lines. (This is comparable to the analysis of a piece of music in phrases and sub-phrases.) In doing this I use the most explicit criteria available. For Movement 1 line-end punctuation is almost completely sufficient to the task. Movement 2 is a bit trickier. It turns out that the middle sections of both movements of the poem are, in turn, divided into three subsections; and again, the middle divides into three. All other subdivisions are binary.

     The middlemost section of Movement 1 (ll. 20-22) describes the eruption of the fountain:

     Amid whose swift half-intermitted burst

     Huge fragments vaulted like rebounding hail.

     Or chaffy grain beneath the thresher s flail.

     The middlemost section of Movement 2 is line 47: "That sunny dome! those caves of ice!" This is practically a word-for-word repetition of the final line of Movement 1. For this reason I will refer to that conjunction as the Emblem. The Emblem occupies the same position in the combinatorial or syntagmatic structure (to use terms from structuralist linguistics) of Movement 2 that the fountain plays in Movement 1.

     Movement 1 is divided into three sections: Section 1.1 covers lines 1-11; Section 1.2: 12-30; Section 1.3: 31-36 (see Table 2). The primary textual index of the division between 1.1 and 1.2 is the space between lines 11 and 12. Such a space also exists between lines 30 and 31 in the manuscript, but is usually omitted when the poem is printed.⁶ In 1.1 & 1.2 we are presented with the interaction of man and nature in two different spaces. In 1.3 we move to a point in Xanadu from which everything else can be perceived.

Table 2: Constituent Structure, Movement 1

     Section 1.1 is dominated by Kubla and his decree. The emphasis is on order: precisely "twice five miles," the gardens are located there ("And there were gardens"), while the forests are here ("And here were forests ancient"). This order is primarily a spatial order and the imagery is primarily visual, creating the illusion of an objective perception of the geography of Xanadu, of its delineation in space. There is a sense of light; the gardens are bright and the forests enfold "sunny spots of greenery." Coleridge seems to be setting up the traditional association between the word (Kubla's decree) and light and order. Let us say that this segment of the poem employs a visual-volitional ontology.

     In contrast, Section 1.2 is dominated by aural and kinesthetic imagery: the woman wailing, "ceaseless turmoil seething," "And sank in tumult." Order seems violated where possible: a place so savage that it has elements of both the holy and the demonic, ceaseless turmoil, tumult, vaulting fragments, and "Five miles meandering"--as though the repetition of "five miles" in conjunction with" meandering" is intended to foreground the order/disorder contrast with 1.1 by contrasting the specificity of twice with the vagueness of meandering. Finally, there is a pronounced shift in tone, with line 12 opening with an exclamation ("But oh !") and the entire section having a feel of excitement and tension that contrasts with the controlled exposition of 1.1. We can think of this as an auditory-expressive ontology. (In the "Low Down and Abstract" section of the essay I will hazard an explanation of how both the visual-voluntary and auditory-expressive domains are constituted through the neural self.)

     Here the external world of Xanadu is likened to subjective states (the chasm is like a woman possessed by desire for her lover) and bodily processes (the eruption of the fountain is like breathing). The earth is, by implication, like the human body. The fountain dominates Section 1.2 while Kubla is passive. He does not act, but only hears ancestral voices foretelling war. Water and breath have taken the place light holds in 1.1; the emphasis is on energy, force, life, and not on reason, light, and order. This contrast between 1.1 and 1.2 is pointed up in the contrast between the phrases "sunless sea" and "lifeless ocean." These phrases designate the same object; but the concepts are different. As light and order dominate 1.1, the body of water is conceptualized in those terms; the sun does not shine on the subterranean sea. But 1.2 is dominated by life energy: sexual desire, water, the breath. Thus the terminal sea is designated in those terms; it is said to be lifeless. Coleridge is playing the same trick here that he did with "five miles." The qualifying term is chosen to be consistent with the governing ontological categories, light and order or life energy and disorder.

     The most striking feature of the semantics of Section 1.3 is that it consists of things perceived by some unnamed person. Nowhere is the earth directly present, but this person is able to see "the shadow of the dome of pleasure" floating on the river and to hear "the mingled measure/ From the fountain and the caves." This is a world is thus a world of percepts. When these percepts are traced to the things that generate them one finds Kubla's dome and the river stretching between fountain and caves. With its reference to the reflection (shadow) of Kubla's dome floating on the river's surface, the semantic space of 1.3 thus partakes of the contrasting spaces traversed in 1.1 and 1.2. Thus the first movement evolves through three domains: visual-volitional, auditory-expressive, and perceptual.⁷

     Further analysis of the constituent structure of Movement 1 proceeds by a scheme in which periods dominate colons, colons dominate semicolons, and semicolons dominate commas, with successive commas indicating coordination or serving some other purpose.⁸ Sections 1.1 and 1.3 consist of two sentences; 1.2 has three sentences. The full analysis is given in Table 2.

     Two minor notes: Section 1.21 is punctuated only with exclamation points; hence the punctuation scheme indicated in the previous paragraph cannot be applied here. Clearly the exclamation points early in lines 12 and 14 mark interjections and not substantive intonation units (in Chafe's terms). The division of 1.21 into two subsections should be obvious enough (note that each subsection begins with an exclamation). The subdivision of 1.231 into two couplets, though not clearly indicated by the punctuation (the comma at the end of line 27 could justify a division at that point in addition to the indicated division at the end of 26), is consistent with the sense of the two couplets: 1.2311 covers one phase of the river's course and 1.2312 covers another.

     The entire structure is represented graphically in Figure 3. The most striking aspect of this structure is its symmetry. And the most striking aspect of the symmetry is the interplay between ternary and binary branches. Once the analysis extends to Movement 2, it becomes possible to analyze the relationship between Movements 1 and 2 by asserting combinatorial equivalence between pieces of text having equivalent positions in the immediate constituent tree.  

Figure 3: Constituent Structure Tree, Movement 1

     Now we need to take a more detailed look at what is happening in sections 1.1 and 1.2. Kubla is the agent in the visual-volitional domain, while the fountain is the agent in the auditory-expressive domain. Each agent creates something, Kubla, the pleasure-dome; the fountain, Alph the sacred river. Just as the fountain is not present in 1.1, so Kubla is not present in 1.2. Neither agent is present in 1.3, but aspects of both of their domains are present. This section serves to mediate between those two domains. Thus the first movement exemplifies one of the key structuralist topes, mediation between binary opposites - though structuralists do not have the only franchise on this idea.

     While the fountain does not appear in 1.1, its major creation, the river Alph, does. The following elements in 1.1 show the fountain's influence: fertile ground (made so by Alph's waters), sinuous rills (again, Alph's waters), perhaps the incense-bearing trees, and perhaps the ancient forests (both depend on water for growth). But all are encompassed by Kubla's decree. He has created a garden, a space in which the forces of nature are ordered by human conventions. Again we have the structuralist theme of binary opposition. Section 1.1 plays it out in the visual-volitional domain while 1.2 will play it out in the auditory-expressive domain.

     Consider Figure 4. The orange rectangle represents a mental space for the visual-voluntary domain; the shaded areas represent sections 1.11 and 1.12. The lines for Kubla's decree are shaded in pink to represent their association with the human realm while the lines for Alph's journey to the caves are shaded in green to indicate their association with the natural world. The enclosure of the gardens (lines 8 and 9) is the immediate consequence of Kubla's decree and is similarly shaded pink while I have shaded the lines of 1.132 in green to indicate their affiliation with nature.

Figure 4: Gardens Bright

     Section 1.2 is dominated by the fountain and is ternary in structure. Section 1.21 concerns the chasm, 1.22 centers on the fountain itself, and 1.23 follows the river from the fountain to the caves. Interestingly enough 1.21 is, like 1.11, five lines long, divided into a couplet and a triplet, and rhyming A B A A B. The first section of Movement 2 ("A damsel with . . . singing of Mount Abora") is similarly five lines long with the same couplet/triplet structure and a similar rhyme scheme, A B C C B.

     In Section 1.1 things and are deployed in space while in 1.2 they are deployed in time. The emotive function of language (Jakobson 1960) is introduced in 1.2, though most of the language is referential, while 1.1 employs the referential function exclusively ("there were gardens," "here were forests"). (Kubla's decree would use the conative function; but we never hear the decree itself. We are just told about it.) In emotive speech the voice, as breath, comes into its own as the vehicle for the _expression of emotion in sound. Note that talking of emotion, is but a form of referential discourse and makes no immediate claim on the feelings of either addressor or addressee. The _expression of emotion resides in the tone of voice rather than in the reference of the words being intoned. When one exclaims (But oh!) one is carried away from controlled self-presence. One neither chooses nor willfully intends emotions, but emotions inevitably influence intentions.

     Human feeling and physiology are assigned to one semantic space that is initially distinct from the semantic space for natural life. Hence the two are analogically connected through similes--"as holy and enchanted/as . . . woman wailing," "As if this earth . . . were breathing," "like rebounding hail/Or chaffy grain." In the image of the dancing rocks the distance between these two semantic spaces vanishes; the vaulting fragments are poetically tamed by the harvest image, threshing grain. In line 23 the rocks are dancing; the dance comes from man, the rocks from nature.

     Figure 5 depicts this interleaving in the first two subsections of 1.2. The blue rectangle represents a mental space for the auditory-expressive domain as realized in section 1.2; the shaded rectangles represent sections 1.21 and 1.22. As before, the green line units represent the natural world while the pink ones represent the human.

Figure 5: Dancing Rocks

     Let us start from the top, with most developed of these similes, section 1.21. As one term we have the deep chasm (1.211). (Perhaps "romantic" functions as a contrast to "stately" in 1.111, setting up a contrast between the pleasure-dome and the chasm.) It is athwart, across, the cedarn cover - a characterization that serves as a link between the chasm and the incense-bearing trees in the garden. It is savage rather than stately. The simile follows immediately:

     . . . as holy and enchanted

     As e'er beneath a waning moon was haunted

     By woman wailing for her demon lover!

     Note that enchant is derived from cantare, to sing. Yet the only singing present here is the wailing of the woman, which contrasts with the sweet song of the damsel with a dulcimer (dulcis, sweet: melos, song) in Movement 2.

     Why is she wailing beneath a waning moon? Is this a particularly auspicious place from which to seek the return of her lover? Or perhaps we have an equivalence between the human order, the woman, whose lover is gone, and the natural order, the moon, which is moving toward absence? And, the waxing and waning of the moon is a cyclic process, as is sexual desire, as is menstruation. Whatever is the case--I don't think it matters that much--both the resurgence of sexual desire and the waxing and waning of the moon are beyond the call of human will. Notice also the contrast between the moon, which is seen at night, and the implicit presence of the sun (through sunny) in 1.1222. Finally, following Reuven Tsur (1987), I note that the poem does not assert that the woman is physically present in Xanadu but only uses her as a term of comparison. What is important are the qualities of sense, meaning, and emotion that enter the poem with her.

     From the chasm the poem shifts focus to the fountain in Section 1.22, which is divided into three sections, 1.221, 1.222, and 1.223. While 1.21 had two sections, one (1.211) focusing on the external appearance of the chasm and the other (1.212) on cyclic inner life in man and natural periodicity, each subsection of 1.22 has both a natural and a human component. In 1.221 the periodic (momently--moment by moment, in discrete repeated moments rather than continuously) eruption of the fountain from the earth is likened to breathing. In 1.222, the structural center of Movement 1, the fountain is compared to both to hail (natural) and grain (raised by man); thus the central element has yet another triple structure, a three-termed simile. Finally, in l .223 the rocks are said to be dancing.

     Notice the way in which the violence of 1.221 becomes poetically transformed into the orderly motion of dancing rocks. The transformation occurs with the comparison to "chaffy grain beneath the thresher's flail." While threshing may be forceful and destructive, the grain is beaten in rhythm and the act of threshing takes place within the orderly cycle of events relating to the planting and harvesting of grain--the human ordering of natural life for human ends. Nor should we miss the connection between the birth of the sacred river that nourishes the fertile ground and the reaping of the harvest. With the introduction of order and purpose, through a simile, the stage is set for the rocks to dance, to display a regular motion in time. (As I do not know how to parse these two lines into one natural and one human--perhaps that is the point--I have attributed both qualities to the section as a whole and assigned a pink-green gradient to the enclosing rectangle.)

     About 1.223 there is little more to say. "At once and ever" I take to be a statement of periodicity, "at once" being a single cycle, and "ever" indicating that all else is a repetition of that cycle.

     Before considering 1.23 a brief summary is in order. Section 1.21 focuses on the chasm, using a simile to link it to both human desire and the celestial cycles, thus linking external nature with inner life. Just as the fountain erupts periodically, so a person breaths periodically. With that link established, Coleridge can counterpoint the two in his development of the fountain's emergence from the chasm and the sacred river from the fountain. Note that sections 1.21 and 1.22 are the only sections in the poem which use similes. Once the discourse moves past the mid-point of the first movement, the similes cease.

     With 1.23 the poem returns to semantic territory familiar from Section 1.1. In 1.231 the course of the sacred river is followed to its end in the lifeless ocean. As noted above, five miles is repeated from 1.121, but with a contrasting qualifier meandering rather than twice. The contrast is further heightened by with a mazy motion. The general similarity between 1.112 and the three final lines of l .231 is remarkable:

     Where Alph, the sacred river, ran   1.112

     Through caverns measureless to man

     Down to a sunless sea. 

     Through wood and dale the sacred river ran,  1.231

     Then reached the caverns measureless to man,

     And sank in tumult to a lifeless ocean.

     The overall combinatorial frame is much the same for both passages; but the paradigmatic selections used to designate and characterize the ocean are specific to the local semantic framework.

     Finally: "And 'mid this tumult Kubla heard from far/Ancestral voices prophesying war!" That the prophecy is of war is most simply glossed as a statement of continued conflict and turmoil. Kubla is not the only agent operating in Xanadu and that situation is not going to change. On one domain Kubla utters a powerful decree, but in this domain he can only listen to the warning of others. Also, it is a human element to usher us from this domain as the wailing woman ushered us into it.

     Consider Figure 6. As previously, we move through the poem from top to bottom. At the top we have the visual-voluntary space, with Kubla, his decree, and the gardens. We see a link from Kubla to the gardens he decreed, and also to his hearing of the prophetic voices in 1.232.

Figure 6: Alph Meanders, Kubla Hears

     Similarly, Alph is linked to the gardens in the visual-voluntary space. We can see that 1.23 depicts the "projection" of the visual-voluntary world into the auditory-expressive as 1.21 ("So twice five miles . . . ) is the projection of the auditory-expressive into the visual-voluntary.

     Let us now step back a moment at review the poem to this point. In 1.1 the poem explores the relationship between man and nature through a visual-volitional ontology by using the image of the garden. In 1.2 we explore that relationship through an auditory-expressive ontology. Notice that when we are in the visual domain, the lines asserting the human term of the interaction precede those asserting the natural term (see Figure 4). In 1.11 we have Kubla's decree, then we are presented with Alph; in 1.12 we first see the ground being encircled by Kubla's walls, then we see the gardens and the ancient forests. In the auditory domain, however, nature comes first (see Figure 5). We are told of the chasm, then the wailing woman; of the ceaseless turmoil, then the panting earth, and so forth. Once this interleaving has run its course, ending in the dancing rocks, section 1.23 (Figure 6) runs through elements of 1.1, thereby placing them in an auditory-expressive context, and does so in reverse order (Figure 7):⁹

Figure 7: Reversal

     Thus 1.2311 ("meandering with a mazy") deals with elements from 1.121 ("twice five miles"); 1.2312 deals with the river's descent into the caves, echoing 1.112; while, at last, 1.232 puts Kubla himself in his place, thus setting limits to his decree (1.111). It is as though this sequence were a single "chunk" in Coleridge's mind which he could traverse in either order depending on the context.

     The most salient aspects of the analysis to this point are summarized in Table 3:  

Table 3: Summary Analysis of 1.1 and 1.2

     Before going on to see how Coleridge explores the relationship between those domains in a world of perceptions, one further point must be made. In "Kubla Khan" the oppositions nature/man (culture) and visual-voluntary/auditory-expressive are orthogonal to one another rather than one being a mediated form of the other, as in Levi-Strauss' (1967) early theory of myth. The nature/man opposition is explored in contexts set by the terms of our two ontological domains. We thus have oppositions operating on oppositions rather than a series of ever decreasing oppositions in one dimension.

     We are almost ready to consider the realm of perceptions, Section 1.3. The transition to 1.3 is prepared in line 29, where we are told that Kubla heard the voices. This is the first place in the poem where we have a verb of sensation or perception. Kubla made a decree, the woman wailed, the fountain made a mighty racket. But we are not told that anyone saw or heard anything. That has now changed. Kubla heard the ancestors and, in 1.3, someone will hear mingled measure.

     Let us first examine Figure 8, depicting a bit of cognitive structure:

Figure 8: See and hear

     The acts of seeing and hearing are in one mental space. As indicated in the methodological notes, the nodes within the space are concepts, while the links between the nodes are relations between the concepts. This diagram thus shows that some unnamed person is seeing and hearing something. Each of those somethings is a "dummy" node that is then connected to some other mental space which contains whatever it is that is seen and heard. While this person is not explicitly indicated in the text, his or her presence is logically implied by the passive form of "hear" in line 33; something cannot be heard unless someone is hearing it.

     This person's perceptions characterize this section of the poem. Reuven Tsur (1987) has characterized this it as follows:

     First of all, we are confronted here, not with the solid dome of pleasure, but with its shadow, which, though it may have a clear-cut gestalt, can also be regarded as a most typical instance of thing-free entity. However, being buoyant on the surface of the waves, it is continuously modifying its stable shape, thus becoming a most typical image of an ever-changing, shifting physical reality, that has - in this way - a structural resemblance to emotional processes. . . . The shadow of the dome floated midway - midway between the fountain and the caves, I assume. This suggests a symmetrical disposition of the perceptual space, that "counts toward" a strong gestalt. At the same time, this is the spot where the inarticulate noises and tumult from the fountain and the caves (mentioned in the two preceding sections of the stanza) meet and mingle. Now, the relatively stable objects (the fountain and the caves) are far away, and only a thing-free and gestalt-free entity, the mingled sounds emitted by them is perceived. Upon this thing-free and gestalt-free entity a symmetrical orientation-scheme (suggested by "midway") is superimposed.

     With this in mind we can now examine how Coleridge weaves the visual and auditory worlds together, line by line and then, at the end, almost word by word. In Figures 9, 10, 11 and 12 the smaller rectangles each represent a line of poetry and are presented in order, from top to bottom. Recall Chafe's assertion that a line represents a single "atomic" thought. Coleridge's objective in section 1.3 seems to be to encompass, or at least touch, both of these realms within a single line, a single atomic thought. He prepares for this in section 1.31 and then does it in each of the lines in section 1.32. Section 1.311 (the shadow floating on the waves) is closely related to the ontology of 1.1, while 1.312 (the mingled measure) is related to that of 1.2. 1.311 and 1.1 are visual and spatial while 1.312 and 1.2 are aural and temporal.

     The image developed in 1.311 would thus seem to assert some relationship between the visual-volitional world of 1.1 and the auditory-expressive world of 1.2. Kubla is the agent of visual-volitional processes. He creates the dome and its reflection (shadow) appears on the waves. The fountain the agent of auditory-expressive processes and creates the river, whose surface reflects the dome. That reflection thus embodies a relationship between those two realms; the content of the refection derives from one realm and it floats on the surface of the river, which originates in the other realm.

     Figure 9 shows that line 31 is linked to the visual-voluntary realm while line 32 is linked to the auditory-expressive realm.

Figure 9: Shadow Floating, 1.311

     In section 1.312 the tumultuous sounds of 1.2 are ordered in measure in the world of perceptions. If we place the element measure in the visual-volitional space we get a diagram like Figure 10.

Figure 10: Hear Measure, 1.312

     By combining Figures 9 and 10 we obtain Figure 11. I have quoted "point" because it is a term in my analysis of the poem, not a term in the poem itself. Yet Coleridge is clearly defining a specific point, one from which one can hear the sounds from both the fountain and the caves and from which one can see the reflection of the dome in the river. This point would seem to be at the center of the semantic universe Coleridge has been exploring in Movement 1 of "Kubla Khan." More prosaically, we might think of this is the spatiotemporal point with respect to which linguistic deixis is determined (Lyons 1977, pp. 677-703).

Figure 11: A Point Defined

     This point is nowhere explicitly designated in the text; it is only implied by the text, by the acts of seeing and hearing that are in the text. This point emerges into our consciousness only as we imagine the scene Coleridge is placing before our mind's eye - and in our inner ear. It is thus akin to the point in the story of the Buddhist's walk that is one of the canonical examples in the conceptual blending theory of Fauconnier and Turner (2002, p. 39 ff.).¹⁰

     Section 1.32 tightens the links between the visual-volitional and auditory-expressive realms. Section 1.321 (line 35) characterizes this conjunction with respect to the creative forces behind it. That it is a miracle links it to auditory-expressive realm, since miracles are beyond man's capacity to enforce his voluntary purposes on the world. That it is also of rare device brings it within the scope of the will, characterizing it as a contrivance of great subtlety.

     In 1.322 this conjunction is revealed to be a visible structure: "A sunny pleasure dome with caves of ice." The sunny pleasure-dome is linked to one realm while the caves of ice are linked to the other. Ice, or course, is a crystalline form of water, which, in the compass of this poem, comes from the fountain; in its crystalline form, water can glitter like the sun.¹¹ Putting this analysis in diagrammatic form yields Figure 12. The parenthesized it derives from the opening pronoun of line 35. The object in question is not simply the dome, the fruit of Kubla's will; it embodies the relationship between the dome and the caves, the fruit of miracle and artifice.

Figure 12: The Emblem

     Note that in both of these lines the conjunction between these two realms has become as intimate as Coleridge is able to make it.

     It will not do, however, to think too much about these geographical and architectural arrangements. Are the caves beneath or beside the pleasure-dome? Is the pleasure-dome thus at the same location in Xanadu where the river disappears into the earth? If this is so, then the dome must be considerably closer to the caves than to the fountain. So, wouldn't the tumult at the caves drown out the tumult from the fountain? What happened to the mingled measure? The more you think about, the less sensible this marvelous conjunction becomes. But then, one isn't supposed to think about it, certainly not in the seconds that transpire as one is reading through these lines on the way to the Abyssinian maid. Whether or not this Emblem is physically realizable is irrelevant. What is important are the associations it knots together, associations that span the entire semantic universe of the first movement of "Kubla Khan."

     The final two lines in effect "concentrate" the percepts and the perceiving consciousness into an emblematic thing. The subjective aspect is concentrated in the conjunction of "miracle" and "rare device" while the objective aspect is concentrated in the image of the "sunny pleasure-dome with caves of ice." Section 1.1 of the poem projected one aspect of subjective experience and self-perception onto Xanadu while section 1.2 projected a contrasting aspect. The contrast between the semantic spaces of sections 1.1 and 1.2 has become projected into a contrast between entities within the mental space of section 1.3. This is a type of compression, as Fauconnier and Turner (2002, pp. 312 ff.) have discussed; call it semantic compression. Everything in Xanadu is linked closely to this Emblem.

     Finally, we should note that it is one thing for Coleridge to explicitly assert the existence of "A light in sound, a sound-like power in light," as he does in "The Eolian Harp" (line 28) and, beyond that, to write philosophical prose about such things. It is quite a different matter to embody that principle in patterns of imagery, where one section of a poem uses visuo-spatial imagery, the next section uses auditory-temporal imagery, and a third section uses both types of imagery--all without there being any explicit assertion of union between sound and light. What was, in "The Eolian Harp," a consciously contemplated trope appears to be, in "Kubla Khan," an unconscious structuring principle. It is this unconscious structuring that makes "Kubla Khan" so remarkable, and so difficult to gloss.

Enchantment

     First I provide an overview of the second movement. Then I discuss the use of personal pronouns, followed by a discussion of how Coleridge manipulates the reality status of successive lines. I conclude with some remarks spanning both movements.

     As Movement 1 ends with an act of perception, so Movement 2 starts off with an act of a memory. The apparent break between line 36 and in 37, between the Emblem and the damsel with a dulcimer, conceals a deeper continuity. We have but moved beyond the indirection of 1.3 and gone deeper into the mind. We now face a semantic space whose explicit content consists of mental acts. The constituent structure of Movement 2 is given in Table 4. 

Table 4. Constituent Structure, Movement 2

     This movement begins with the focus on the damsel with a dulcimer. I have already remarked on the structural similarity between sections 2.1 and 1.21 (wailing woman). Recalling the etymological significance of dulcimer, we can see that the damsel contrasts with the woman wailing of 1.21: a discordant wail versus sweet music. And the woman is associated with a chasm, while the damsel sings of a mountain - my inner psychoanalytic voice tells me these are two aspects of the same creature, though I will leave that thought alone until I reach the more speculative part of this essay. Neither female is directly present in the poem. The woman appears in one of the terms of a simile while the damsel is an element in a memory. The memory is of a vision, a link to the visual-volitional realm (1.1); and within the vision there is a song, a link to the auditory-expressive realm (1.2). The conjunction that the poem worked so hard to create in Section 1.3 is thus taken as an accomplished fact in Movement 2 of the poem.

     We must also attend to the Coleridge's deployment of the various functions of language, in Jakobson's (1960) sense, particularly the referential, emotive, and conative functions. Except for section 1.21 (the wailing woman) and line 36, all of Movement 1 employs the referential function. Movement 2 opens in the referential function, but, as we have seen, with an entirely different domain of reference: thoughts and speech acts.

     The identification of the initial boundary of section 2.2 is obvious; line 41 comes to a full stop and the mood changes dramatically at the beginning of line 42. The language is still referential, but of a different kind. The speaker is no longer reporting, he is supposing; he is no longer oriented toward the past, he is oriented toward the future; and the focus has shifted from the damsel to the speaker himself.

     The section's ending boundary, however, is not so obvious. The analysis internal structure of 2.2 is also tricky. The punctuation markers that served in the first movement of the poem are not so indicative in the second.

     If the poem from "Beware! Beware!" (line 49) had been enclosed with quotation marks (as is customary for direct quotation), then I would be inclined to treat that final section of the poem as one major unit. But, though Coleridge knew the conventions of quotation, he chose not to use them here. The effect is to minimize the distinction between asserting that "all would cry" and the actual content of their cry. In Chafe's terms (1994, p. 195 ff.) the assertion of the cry and the cry itself have the same immediacy. It is as though our focus had shifted from the mind of the narrator to the mind of the auditors. And that, I suggest, is exactly what has happened. We have also shifted from highly energized referential speech to emotive speech, and we continue in that mode through the end of line 50.

     Line 50 functions in the conative mode - orientation toward the addressee - and that is one thing that differentiates 2.2 from 2.3. The rhyme scheme also gives us strong supporting cues. Lines 28, 49, and 50 rhyme (there, beware, hair), giving them a strong affinity. Line 51 does not rhyme with those lines. It is part of an unrhymed coupled that is paired with another unrhymed couplet in a chiasmatic rhyme figure that ends the poem (thrice, dread, fed, Paradise). Thus we can fix the final boundary 2.2 at the end of line 50.

     Section 2.3 then opens with two lines in the conative function (2.31) and concludes with two in the referential (2.32). The final couplet states the reason for the poet's power, and the reason he must be feared: he has been to Paradise.

     An immediate constituent tree for Movement 2 is given in Figure 13. The most striking thing about this diagram is its resemblance to the tree for Movement 1 (Figure 3). We have a central core of ternary structures flanked by a mantle of binary structures. At the center, line 47 repeats the Emblem that ends Movement 1. The tree diagrams for Movements 1 and 2 represent the overall syntagmatic or combinatorial structure of the poem. The final analysis of the relationship between the two movements depends on comparing fragments of text having equivalent positions within these trees--but that comparison must wait. 
 

Figure 13: Constituent Structure Tree, Movement 2

     Where Movement 1 depicted the projection of human forces into nature, Movement 2 is about the intersubjective processes of poetic communication--the relationship between the roles of poet and ordinary citizen within one person and the relationship between the poet and others. The conditional of 2.2 is the device through which we move from the speaker's mind to the minds of others observing that speaker and his creation - we will examine this in considerable detail later. Section 2.23 moves to the actions the auditors takes in response to the poet's activity. Having heard, then seen, they are now going to cry out (2.231). and that cry (2.232) begins with a warning: "Beware! Beware!" Just as Section 1.2 ended with a warning--the ancestral voices warning of coming war--so the analogous section in Movement 2 ends with a warning. This analogy is enforced by the sound structure of the poem. There are only two places in the poem where consecutive lines begin with the same words: lines 28 and 29: and; and lines 48 and 49: and all. These two places are the loci of warnings. Sound structure is clearly being used to indicate the functional equivalence of these warnings, which would be no surprise to Roman Jakobson and his students.

     The warning in 2.232 is completed with "His flashing eyes, his floating hair!" Flashing eyes picks up the imagery of the dome, light and the organs of vision; while floating hair is in the territory of the fountain and the river. The connection is through floating, meaning here "floating in the air, flying in the wind," but pointing back to the dome's reflection floating on the waves. In this context the inspired poet poses a threat to the citizen as the tumultuous fountain and caves posed a threat to Kubla in Movement 1. We have here the magnetic gaze of the ancient mariner who, with his "long grey beard and glittering eye," captivated the reluctant will of the wedding guest.

     In section 2.3 members of the poet's audience try to contain (2.31) and explain (2.32) him. The containment proceeds through encircling the poet and denying the vision he offers -- suggesting that it is possible to deny what one sees but harder to close one's ears. Vision is more subject to the will than hearing. In the final words of the poem the auditors affirm the poet's connection with a deeper reality, with Paradise. That is what they call whatever it is that enables his inspired speech.

     One need not undertake any sophisticated analysis to see that the second movement of "Kubla Khan" differs from the first in two ways. It is: half as long, and set in a different domain. The analysis has revealed, however, that the two movements of the poem have very similar forms. What that implies is not at all clear. Further, the first movement relies heavily on semantic blending through interleaving lines related to different semantic spaces. This all but disappears in the second movement where, instead, we have clever manipulation of the relationship between the speaking voice and the speech itself.

     We are thus dealing with the cognitive machinery for managing the discourse. I want to start with the fact that Movement 2 has many personal pronouns while Movement 1 has few. Personal pronouns "obtain" their referents from the immediate speech situation and thus depend on the relationship between that situation and the content of the speech.

     Let us begin, however, by considering the use of third-person pronouns, which do not necessarily depend on the immediate situation for there referents. Consider the phrase "on her dulcimer she played" (l. 40). In order to understand that phrase one needs to determine the antecedent for "she." How is it that one does this? It cannot be done at all if an appropriate antecedent has not been introduced prior to the occurrence of the pronoun. Given an appropriate antecedent, the fact that one can identify the pronoun with that antecedent implies that the unconscious mental machinery that manages the speech situation has been keeping a "list" of topics of conversation and refers to that list whenever a pronoun is used. If the pronoun is feminine, it looks for the most salient item on the list of feminine gender; if the pronoun is plural, it looks for an item of plural number, and so forth. If the list contains two or more items that have the required gender and number and are equally salient, then there is no way to resolve the pronominal reference and one is confused.

     In face-to-face interaction one is always aware both of oneself and of ones speech partner. When it is necessary to refer to either of these parties the speaker will use some form of the first person or second person pronoun, depending on the reference. The meaning could be conveyed perfectly adequately by using the person's name, but doing so is strange (see Benzon 2000 for a detailed discussion of this point).

     Figure 14 shows Jack and Jill conversing with one another, with Jack to the left and Jill to the right. The rest of the diagram should be self-explanatory (given the explanation in the section on diagramming conventions); it simply depicts the ordinary usage of first and second person pronouns from Jack's point of view (on the left) and from Jill's (on the right). The intentional links are labeled with the pronouns Jack or Jill would ordinarily use when referring to one another in speech. 

Figure 14: I and You

     With this in mind, now consider Figure 15, depicting the nervous system of a person who is reading "Kubla Khan"-- note that it is the poem that dominates the external world. In 15A the reader identifies with the poet (P); in 15B the reader identifies with a typical auditor (A). That shift in focus happened in the middle of line 49, "And all should cry, Beware! Beware!"

Figure 15: Shift of Identification

     Just as the first movement first considers how various actors operate in Xanadu, and then moves into a perceiving mind to gauge those interactions, so the second movement starts in the mind of a poet and, in the course of following his thoughts, somehow ends up in the minds of auditors listening to the poet. This is an important point and requires careful demonstration. We need to look carefully at how this section of the poem flows from one mental space to the next, and how Coleridge manipulates the poetic line to interrupt the flow of these mental spaces.

     Pronoun management is only one aspect, among many, of regulating one's speech. One must also keep track of the reality status of one's assertions.¹² A proposition such as "John fell ill" has the same meaning regardless of whether or not one has seen John's illness oneself or only heard about it. But the value that assertion has for some third party may vary according just how the speaker came to that knowledge. Accordingly, some languages, but not English, have verb affixes that serve to indicate the reality status of a proposition. Such affixes are called evidentials (Willett 1988, Bybee, Perkins, and Pagliuca 1994, pp. 95 ff., Curnow 2001). Such evidentials indicate a relationship between the speaker and the content of her speech. Most commonly, evidentials differentiate between having direct knowledge of the event in an assertion, knowing about it through hearsay, or inferring it on some basis (which may include having dreamt about it).

     In contrast to the first Movement--where everything presented to us on the same footing--it is just there; the materials in Movement 2 are given various kinds of reality status. We begin with a statement that is presented as being true in the here and now: "I had a vision." That assertion is made in one mental space. We are then given the content of the vision. The vision itself belongs in a different mental space, one that has whatever reality status one assigns to visions.

     Let us begin by diagramming the mental spaces for 2.1. In doing so let us adopt the convention that the base space, in which the poet asserts that he had a vision, is to considered real. As the poet is a creature of the poem that base space cannot, in fact, be real. But the precondition for entering into literature is that we treat the text's fictional base space as being real. In Coleridge's own phrase, we enter into a "willing suspension of disbelief for the moment." Here then are the mental spaces for 2.1:

Figure 16: Real and Hypothetical

     Space one is real and is the parent space of two (Fauconnier 1994, p. 115 ff.; Lakoff 1996, pp. 95 ff..), which is imagined relative to space one. Space two is, in turn, the parent of space three, which is also imagined relative to space 1. The "poet" in space one is, of course, the person who is (imputed to be) speaking this poem.

     Coleridge then employs the same distinction, between real and imaginary spaces, in a considerably more complex construction in 2.2, which is a counter factual conditional having an elaborate set of consequents: "If A, then B, C, D, and E." Let us diagram this as follows:

Figure 17: Counterfactual Conditional

     Note that we have the same overall configuration of mental spaces here as we did for 2.1, one real parent space (1) connected to an imaginary space (2) which is, in turn, parent to another imaginary space (3). In this case the first imaginary space contains a conditional rather than a vision; and that conditional is rather more elaborate than the vision was. Similarly, in both cases the first imaginary space contains a person making an utterance, whose content is contained in another imaginary space. In the case of 2.1 it is the maid whose utterance is a song while in the case of 2.2 the utterance comes from an unspecified group of people who have rather quite a bit to say.

     Something strange, however, happens as we continue reading through this conditional, line by line. There is a major structural break after line 50 (" . . . his floating hair") and then we have the final quatrain, "Weave a circle round him thrice." Following Wallace Chafe 1994, pp. 241-242) I suggest that the lack of an explicit attribution of this speech to anyone at this point (l. 51) heightens its immediacy, thus making it relatively easy for the reader to "lose track" of her identification with the poet and to establish an identification with the auditors. To be sure, there is no doubt who these people are, they were introduced three lines previously, but the nature of the utterance has changed and the poem has entered a new phase. The reader is now fully identified with those amazed bystanders, making them, their perceptions, and their fears fully real - at least within the confines of this poem.

     Consider the following diagram: 

Figure 18: Protective Incantation

     The first space, the one that established this line of thought, has all but "faded away." The elements in the second space (expanded from node E in the previous figure), "people say X," are not actually expressed in the final quatrain. Rather, they have been inherited from a previous point in the poem, line 48. The problem, of course, is line 48 was staged from within a mental space in imaginary mode (space 2 of Figure 17). Nowhere are we explicitly told that those people have become real. How did this happen?

     What happened is the Emblem re-entered the poem at line 47: "That sunny dome! those caves of ice!" We have, of course, seen it before. Within the context of the entire poem, the object designated by those words is as real as anything in the poem. It was carefully created in Movement 1. Its invocation in Movement 2 thus means that others can see, be amazed, and be fearful. The real existence of that object obviously implies that our speaker did revive the visionary damsel's song and that, as we are told in the last line, he has been to Paradise. It is as though the reality of the dome-and-caves has retroactively changed the reality status of poet's counterfactual statement. Now that an element in that counterfactual has become fact, the entire statement has become fact as well.

     In making this argument I am making a number of assumptions about the interaction of lines and mental spaces. In the long-term these assumptions will have to be justified by empirical evidence about the operations of the nervous system. For the moment, however, that is beyond our reach and so I must be content simply to state my assumptions.

     I am assuming that each line will, by default, inherit the reality status of the previous line unless that status is contravened. Reality status can be conveyed by explicit statement--such as "he imagined . . .", "she saw . . " and so forth--or it can be indicated by a shift from one type of speech to another. The latter is what happens in line forty-seven. The previous lines had been in the referential mode, but 47 is in the expressive mode. In effect, this shift from the referential to the expressive mode "breaks" the mental space containing the complex conditional begun back in line 42 and "resets" the reality status from that point to the end of the poem.

     Notice that Coleridge introduces the Emblem by inserting it into the middle of an otherwise complete construction. Without the Emblem those lines would read: 

. . . with music loud and long,

I would build that dome in air,

And all who heard should see . . . 

Instead, Coleridge gave us: 

. . . with music loud and long,

I would build that dome in air,

That sunny dome! those caves of ice!

And all who heard should see . . . 

     The Emblem is thrust into the middle of this movement as the fountain forced its way into Xanadu, thus emphasizing the break in reality status.

     Figure 19 contains a crude visualization of this. Each row represents a line or two from the poem, as appropriate. In the first row we have five mental spaces, each of which corresponds to one of the elements in the conditional statement depicted in Figure 17 and each has a reality status of imaginary, as is appropriate. The contents of the first mental space - "Could I revive . . ." - has been realized in two lines of poetry while the contents of the rest have not as yet been realized at all. They exist only in posse. As we move through the diagram from top to bottom, the contents of these mental spaces become realized in order. However, after lines 45 and 46 an intruder shows up, the Emblem. It is rendered into speech and its mental space is real. Hence, when the next term in the conditional is rendered into speech, its mental space is also real, and so the last element in the conditional becomes real as well.

Figure 19: Inserting the Emblem

     This is a trick, one that depends on the extreme focus on the present that poetry demands of us. We must attend, not only to the meanings of words, but to their sounds as well, and thus must track the evolution of the sonic line as well as of the semantic line. The effect of this trick is that, while the last four lines, starting with 51, remain in the consequent of the original counterfactual assertion, they read as though they are being uttered by real people in the here and now - at least that's how the read in the moment, as opposed to how they may read to someone who is detached from the immediate experience and examining the text at some remove through the lens of this or that methodology. The hypothetical nature of those listeners, and thus the hypothetical nature of their utterances, is simply lost in the rush of exclamation and assertion. Those listeners are now real and the reader is now identifying with their point of view rather than the poet's. Reality has been retroactively revised. The imaginary situation of Figures 17 and 18 has now, in effect, become real and the mental space in which the auditors make an assertion has become the baseline: 

Figure 20: Revisionism in Paradise

     If the auditors are real, their assertions true, and the domes and caves are really there, hanging in the air, then the poet really has been to Paradise, and Xanadu is it.

     We have a more details to examine. The concluding couplet asserts that "he on honey-dew hath fed/ And drunk the milk of Paradise." Consider that the spherical shape of the melon echoes that of the dome and of the poet's flashing eyes, and that the color of honey echoes that of the sunny dome. Similarly, the milk is an echo of the river Alph and the floating of the poet's hair. These echoes may not be powerful, but they are real and they are fully within the semantic terms established within this poem. We thus have the following progression, from the beginning of the poem to its end:

Figure 21: Milk and Honey-Dew

      In the first Movement we are in an allocentric space where the narrative voice is unaware of itself. This movement evolves through an opposition between two major forces--one visual-volitional, the other auditory-expressive--which is itself crossed by another opposition, between the human and the natural. These forces end up being yoked together within the scope of a single line through the poetic magic of semantic compression. That line reappears in the second Movement, which evolves through an egocentric space. The narrative voice is now self-aware. In this context those forces come to constitute the poet's physical character. Notice, too, that this blending of semantic elements of the first movement into the second movement is a further extension of that interleaving of domains that constitutes the major device of the first movement. Now, at the very end of the poem, we learn these opposing forces, volition and _expression, constitute the poet's character because he fed on them, and in Paradise at that. At this point - feeding in Paradise - I believe we are once again in psychoanalytic territory, a discussion I want to defer until later.

     I would, however, like to conclude the analysis with a few more remarks that span both movements of the poem. Both movements are framed by similar actions (see Figure 22). 

Figure 22: Framing Verbs

     The first movement opens with Kubla's declaring the dome into being while the second movement opens with the poet's report of his vision. The first movement concludes with the unnamed narrator seeing and hear events in Xanadu while the second movement concludes with the auditors hearing and seeing the dome and the caves. In each case, something comes into being between the assertion and the perception. This, then, is but another piece of evidence that the two movements are built to the same underlying form.

     But what is it that comes into being, in effect, between the two movements? Given the way Coleridge blends the natural and the human in the first movement one suspects that the Xanadu imagery is being used to give verbal form to human will, feeling, and perception - as we will see shortly, that is what Kenneth Burke thought. If so, then those are the mental states of a person who is not self-aware. The second movement starts with self-awareness and then moves to awareness of the self that has been displaced into others. Thus, it must be that self-awareness is what emerges in the transition from the first to the second movement.

     Finally, I suggest that we declare an end to the question of whether or not "Kubla Khan" is a fragment of a complete poem. I have no trouble believing that Coleridge may have had an opium-induced vision that ranged far beyond the contents of this poem. But the text he finally published is a complete poem. It has a complex and elaborately worked structure that achieves semantic closure by the poem's end. Anyone who believes otherwise should present an argument that is comparable in detail to the one I have presented but that demonstrates that Coleridge started something somewhere in this poem that he then left unfinished.

     Now it is time complete the analysis with an examination of the "Kubla Khan's" sound structure. That analysis will reinforce the sense of the poem's completeness by showing that the two movements share what I call cardinal points of sound structure.

Sound and Sense

     The sound patterning of "Kubla Khan" is quite complex. I wish to consider only some of most obvious patterns; in particular, I want to examine the poem's rhyme structure.¹³ Rhyme foregrounds line ends and is the main meeting point between units of thought and sound.

     Consider Figures 23 and 24, which indicate the rhyme schemes for Movements 1 and 2 respectively. The middle column in each figure gives the final words in each line, with the constituent structure indicated to the left. All near rhymes have been treated as full rhymes for simplicity, and I have labeled the rhyming syllables without going to the beginning of the alphabet at the beginning of a new stanza. Thus the rhyme designations are continuous from the through the poem's 54 lines.

     The point of this analysis is that there is a scheme to Coleridge's rhymes in "Kubla Khan." And that scheme agrees with the constituent structure of the poem, as determined primarily by punctuation, except in two places, 1.22 and 2.22--analogous sections of Movements 1 and 2 respectively.

     The first step is to isolate the basic units. There are three in "Kubla Khan." Let us use these designations: 

     X: rhymed couplet

     Y: unrhymed couplet

     Z: triplet rhymed AAB 

     We can then group basic units into larger units primarily according to rhyme relationships extending across adjacent X, Y and Z units. Thus an AB couplet adjacent to an AAB triplet will be grouped into a higher level unit because the two units share line-end sounds. And thus with two adjacent AB couplets or with an AB couplet adjacent to a BA couplet.

     I have left the series of rhymed couplets in the center of Movement 1 unanalyzed because there is no method of analysis, other than treating them as a group, that would not be totally arbitrary. Much the same applies to lines 46 through 49 in Movement 2. As we will see, that is one of the major features of Coleridge's elaborate rhyming.

Figure 23: Rhyme Scheme, Movement 1

     A comparison of the rhyme structure and the tree structure of the poem reveals a most remarkable correspondence between the two, especially in Movement 1. Except for lines 17 through 24 (1.22), there is no violation of constituent semantic structure by the most elementary units of rhyme structure. No X, Y, or Z unit crosses a division in the constituent structure. Thus, with the exception noted, the most elementary units of the tree structure, which corresponds to the pause structure of the poem as indicated by punctuation, correspond to elementary units of rhyme structure. Nor is there any violation of higher levels of constituent structure by larger of rhyme groupings.

     There is less to say about the rhyme structure of Movement 2. Sections 2. l and 2.3 are quite straightforward and need no comment. Section 2.2 is rather indeterminate on the matter of correspondence between rhyme and semantic structure. As for the other five lines (46-50), there is not enough rhyme structure there to support any analysis of its correspondence with semantic structure--which is, I believe, precisely the point.

Figure 24: Rhyme Scheme, Movement 2

     For in both Movements 1 and 2 there is a section of seven or eight lines where there is no correspondence between rhyme and syntax. Reuven Tsur (1987) has remarked on this in his analysis. Of lines 17 through 24 he remarks:

     As for the eight-lines-section beginning with "And from this chasm" (quoted above), they are grouped according to two diverging grouping principles. In one respect, these lines are simple couplets, grouped by rhyme, accordingly. Syntactically, however, a second pattern is superimposed. The line "A mighty fountain momently was forced" (19) is grouped, rhyme-wise, with the next line. Syntactically, however, it is grouped with line 17. From the preposition from a verb is predicted; this prediction is fulfilled as late as the end of line 19, running on to the next couplet. Thus, the interpolated simile "As if this earth in fast thick pants were breathi