Chapters 2 through 7 offer some remarkable meditations on the nature of reading, some direct, others mediated by Birkerts's autobiographical reflections. In this commentary I want to bring into a single framework several issues raised by Birkerts, notably the nature of the engagement with a literary text and its influence on the self of the reader. Birkerts's account shows some superficial inconsistencies, which are illuminating in themselves, as I will suggest: these relate to his notion of "transparency" during reading and to the different construals of the self he proposes.

It should be pointed out first that, since the days of I. A. Richards (e.g., Principles of Literary Criticism, 1924), interest in the psychology of literary reading was largely ruled out of order, first by virtue of the argument on the "affective fallacy" by Wimsatt and Beardsley (1949), then as a result of the paradigm shift represented by structuralist and post-structuralist approaches to language. Those who put forward models of reading, such as Jonathan Culler, Christine Brooke-Rose, Michael Riffaterre, or Stanley Fish, based their understanding on various theoretical accounts of textuality. Their readers, as a result, were theoretical entities, the creation of textual strategies or of the larger cultural formations supposed to determine literary response and the competencies on which it depends. Even Wolfgang Iser, probably the best known of the reader response theorists, based his work on phenomenology and showed no interest in checking his account against the behaviour of actual readers. More recently a new theoretical trend has appeared, inspired by cognitive linguistics (notably Van Dijk and Kintsch, 1983) and the analysis of metaphor in everyday language initiated by Lakoff and Johnson (1980), which has attempted to explain literary reading in terms of cognitive models (e.g., Hobbs, 1990; Turner, 1991). As I have pointed out in discussion of a book by Johnson (Miall, 1997), this approach still has little to say about feeling or the role of the self in literary response. Thus Birkerts's book, with its detailed quasi-phenomenological descriptions, offers an unusual opportunity to reflect on the nature of reading. Birkerts raises issues which for over forty years have been of little or no interest among mainstream literary theorists and which lie outside the reach of the cognitive school. At the same time, his discussion finds some interesting echoes in the, as yet, little known field of empirical studies of literary response.

Birkerts has much to say about the compelling nature of reading. At times he conceptualizes this as our absorption with a fictional world, seen through the "transparency" of the printed page. The issue of transparency is also prominent in the account of Richard Lanham, which I have discussed elsewhere. In both cases, I will suggest, the metaphor is misleading. While in Lanham's case transparency is inappropriately contrasted with the self-consciousness he prizes in reading, in Birkerts's case transparency conflicts with the more perceptive reports he makes about the effort required to read.

As a child, says Birkerts, he lived within the books he read: "This notion of hiding, secreting myself in a text was important to me -- it underlies to this day my sense of a book as a refuge." During his time in first grade he knew that lying beyond the words on the page "was a complete other world." Then one day, "I went over and around and suddenly through the enormous letter shapes of Kipling's Jungle Book," and from that moment "a word became a window onto its meaningful depths." At the same time, he characterizes his relation to books as a "Dreamy sensuousness" (p. 35), which hints at the altered bodily state in which reading can place us.

Although Birkerts refers to his notion of transparency several times (cf. pp. 100, 101), it is belied by comments that show other aspects of the reading process. He notes that his adolescent conception of a writer was of one positioned "independently, at an angle to society; it was to live in a different and possibly dangerous way in the service of a vision" (p. 41), and as readers we search for that independence, "this outsider's perspective" (p. 42). To the extent that the vision of such literature is not in coincidence with our normal, everyday perspective, it will take effort to understand it. Birkerts points this out later in his description of the "depth" experienced during reading or in other aesthetic experiences. "Depth survives, condensed and enfolded, in authentic works of art." But we shatter the horizontal of the everyday "Not without some expense of energy," an experience that can be "disorienting" (p. 76). Reading literature is thus not transparent, at least not in those contexts where it continues to require "energy."

In his appeal to "depth" Birkerts argues for reading as a holistic experience, an "illusion" that Roland Barthes, among others, wanted to strip away. The contrast is between a psychological and a semiotic understanding. Where Birkerts sees reading as a dynamic process at the level of feelings and judgements within an evolving self, Barthes sees it as a ceaseless interchange and relationship of codes. In Barthes's view "Subjectivity is a plenary image, with which I may be thought to encumber the text, but whose deceptive plenitude is merely the wake of all the codes which constitute me, so that my subjectivity has ultimately the generality of stereotypes" (S/Z, p. 10). Barthes's model of a text is of a network, a lateral chaining of codes, not the vertical sense of depth espoused by Birkerts.

The issue here, beyond the spatial metaphors used by both writers, concerns the nature of the codes deployed in literary reading. While Birkerts, when he asserts transparency, fails to recognize the codes that shape literary reading, Barthes writes as though nothing is to be explained except codes (in S/Z he proposes five). The reality is perhaps rather more complex. Birkerts provides a glimpse of what the codes might be when he turns to discussing the effort required for engaging with the world created by a writer. He quotes the opening paragraph of McGrath's novel Dr. Haggard's Disease (1993). He points out how he has to read the paragraph several times before he can accustom himself "to the rhythm and voice of the work. . . . I cannot flip a switch and be there. . . . The words have to come alive in the ear -- I have to hear them and hear them deeply" (p. 99). Only after expending effort, and perhaps reading several pages into the novel, does he really begin to inhabit the world it represents: by degrees, he says, the words "become transparent, a reverberation of sense" (p. 100). Yet his allusion to "sense" here, or his earlier phrase "dreamy sensuousness" (p. 35), suggests that more is at stake than looking through the words onto that other world.

McGrath, notes Birkerts, is a stylist; the writing is skilled. Yet "I find breaking in as arduous as ever." To "hear" the words, as Birkerts puts it, is to engage with their aural, their material reality, and to hear them "deeply" suggests attending to their strangeness, their unfamiliarity in this context. When a word or a phrase is defamiliarizing, reading must occur a little more slowly. In our empirical studies (Miall & Kuiken, 1994) we have found that response to such words is usually coloured by feeling: the extra time allows the meaning of that feeling to be developed by the reader. In the aftermath of such a passage, the feeling appears to evolve and to make contact with appropriate concepts or memories that provide a first interpretive framework for the reader's understanding (Miall & Kuiken, 1995). It is this kind of work that Birkerts appears to report as he embarks on his reading of McGrath.

The linguistic code, in this perspective, is not semiotic but stylistic, a deliberate deviation from the common use of words by the author for an aesthetic purpose. It is not semiotic insofar as it evokes through feeling what is distinctive and individual in the reader, rather than constituting the reader as the site of a common stock of codes inherent in the reader's culture. In this respect reading is not transparent, since literary style (not continuously, except in the case of some poetry, but in periodic fashion) presents obstructions to immediate understanding. We are continually alert to the shifts in tone and meaning that a writer achieves through alliteration, ellipsis, metaphor, or other devices.

Once reading is underway, however, the world of the novel may at times seem as real as the world we are living in. Birkerts observes that there are now two worlds, there around me, and here in the book, such that "the printed page, unread, marks the line between them. As a material thing, a paper surface, the page has its place in the former realm; as it is read, and rendered transparent, it becomes an entryway to the latter" (p. 101). Yet transparency is still a misleading term, on Birkerts's own evidence. It is still harder to go from the present world to that of the book than the reverse: taking up a novel again midstream, Birkerts notes, "I still find that reentry takes effort" (p. 103). Traces remain of the effort required to hear the words, to respond to them richly and sensuously.

Birkerts suggests one of the reasons that makes sustained reading possible, despite the effort that at times may be required. Through imagery we establish a framework within which reading continues to unfold:

Having brought a setting to life in our imaginations, and having invested it with the tones and shadings that are uniquely our own, we sustain it -- and trust the author not to frustrate us. . . . We do not at every moment remember the setting afresh, any more than when we sit in a restaurant we keep recalling that we are in the middle of a city. We work hard to establish the image, and then we move our attention elsewhere . . . (p. 97)

The setting, once established, is something taken for granted, and this is one source of the illusion of transparency. In fact, the setting is composed entirely of the power of the words on the page to evoke appropriate concepts and memories from our own experience, and unless the setting describes a place with which we are actually familiar, the image is necessarily a constructed, composite one. Transparency implies the gaze though a window on an actual scene, or a replica such as we watch on film or on the television screen. But imagery during reading is probably rather unlike such compelling visual perceptions, as psychological studies of imagery have suggested: the image is a specification for a scene rather than a portrayal of one, and it is made up as much of implicit meaning and feeling as it is of explicit visual elements (cf. Nell, 1988, pp. 215-221). In this way it is likely to draw more deeply and effectively on the reader's sense of self, as encoded in memory and feelings: the reader will have constructed a "uniquely" personal image (as Birkerts notes). This provides a matrix within which the issues played out by the literary text are given the most chance to interact with the reader's self concept, perhaps to modify or transform that concept in some significant way. The term "transparent" is thus misleading on two counts: it misrepresents the nature of the reader's engagement with literary language, and it implies that what the reader "sees" is as real as a film screen rather than a construct of the reader's mind.

The imagery of the reader provides the conditions for what Birkerts calls a "change of state," a reorientation of the self. Understanding how that comes about, however, will not be easy: "Isn't this the most elusive and private of all conditions, that of the self suspended in the medium of language, the particles of identity wavering in the magnetic current of another's expression? How are we to talk about it?" (p. 78). Such reading is "an inner orientation," a "shift," "a change analogous to, but not as totally affecting as, the change from wakefulness to sleep" (p. 80).

Discussion of the image already suggests that reading is far from passive, as gazing through a window into another room might be. The more appropriate scenario, says Birkerts, is that of "collaboration." As readers we bring our own substance to the words on the page, "We are actively present at every moment, scripting and constructing" (p. 83). Our own resource of images, memories, and feelings is drawn upon. The reality that we inhabit is created out of our own lives: a novel "smelts its reader, extracting responsive emotions and apprehensions and then showing them forth in an aesthetic frame." As a result, the boundaries of the self become clearer and more permeable, Birkerts argues (p. 93).

At its most involving, reading recreates the self: "I feel as if the whole of my life -- past as well as unknown future -- were somehow available to me . . . as an object of contemplation" (p. 84). We are called, says Birkerts, to "a larger conception of the meaningful" whose "implicit injunction . . . is that we change our lives, that we strive to live them in the light of meaning" (p. 85; cf. p. 87).

Self-realization of this kind comes most immediately from identification with a character in a novel or a play. "These recognitions are eventually externalized as ideals and in that form guide the behavior long after the spell of the reading passes" (p.90). This is a transformation, to use Birkerts's term, towards "what we might call, grandly, existential self-formation." Our lives "feel pointed toward significance and resolution" (p. 91): in reading we anticipate the shape of the self to come. In his reading, Birkerts says, "the pages of novels began to lay down the traces of an expectation in me" (p. 94). Identification allows us to recognize and bring into being an aspect of the self that might have remained unnoticed, and then to begin to live within it or through it. Reading about certain characters is an inductive process: rather like the action of a magnet on iron filings, it aligns previously disparate tendencies in the self and enables us to realize a new dimension of experience.

There are also ethical insights to be gained from the situations depicted by literature. After reading Naipaul's A Bend in the River Birkerts describes how "I found myself brooding for days on the ways in which cultures and value systems come into collision," and looking at the people in the streets around him with a new eye (p. 103); such reading quickens "the empathic nerve" (p. 106). This might be considered a kind of decentering of the self, a rather different process from the affirmation and development of the self emphasized elsewhere by Birkerts. The process, however, appears to unfold in the same way: reading enables us to test or question the boundaries of the self. Empathy with a character whose described experience is quite different from our own must still draw upon aspects of our existing knowledge: our feelings, memories, and concepts still provide the basic template on which that alien experience is elaborated. But the literary text in turn may react back in unexpected ways on that experience, casting our feelings in a new light, modifying or questioning some unconsidered concept. While the process may lead in a different direction, away from self-affirmation, the underlying mechanisms remain the same.

A related issue concerns the bond formed by the reader with the author, a profoundly important bond, if we accept Birkerts's claim. The world we find in the book, he says, is "held fully in the suspension of a single sensibility -- the author's." That world is "irradiated through every part with the intended coherence of its conception. The fictional world is a world with a sponsoring god -- or creator . . . " (p. 81). Birkerts's discussion does not go on to consider the various possible types of narrator, or their different impacts on the "implied" author described by Wayne Booth (1961). Clearly, our relation to the author is influenced by the narrative voice, whether omniscient, limited, or first-person; and in each case we may infer an author who never speaks in his or her own right. But the empathy principle seems relevant here too, if in a more comprehensive sense.

We make the assumption that what is shown by the author has a purpose in the larger structure, that every detail, to the last word, has a necessary function in the whole, even if we cannot at first make out what that function might be. Thus we extend to the implied author the same courtesy we extend to people, until the evidence shows otherwise: we trust the validity of the experience of the author and the author's sincerity in conveying it to us. For example, I believe that Jane Austen created an Elizabeth Bennet because such a character accurately represented important perceptions and judgements that Austen was able to make about her everyday experience, and that as an author she has used her skills to convey this perception as effectively as she can. As a reader I must empathise with the author's vision in order to understand it, and I must certainly do so before dissenting from it (for example, on ethical grounds, as when I would differ from Milton on the position he assigns women in Paradise Lost).

This bond or contract we make with the author, however, begins to be violated in various postmodern fictions, from Pynchon to Coover, and is effaced by hypertext fictions such as those of Moulthrop. The author in such cases is not dead, as Barthes or Foucault would have it, but variously a truant, a showman, or a confidence trickster. Unlike James's apparently solid house of fiction, the hypertext author is a realtor with only pictures of prospective houses to show you, where the pictures appear to be of rooms in so many different houses that you can never figure out how the kitchen relates to the living room, or exactly how many bedrooms there are. If there is a house at all, it is one that you can only theorize about inhabiting, since the author cannot let you forget the artifice of his showmanship long enough for that engaging of the self with the fiction to get underway. The image, discussed earlier, constructed from our feelings and memories, is not allowed to sustain reading beyond one or two narrative nodes before a jump to another scene requires the creation of a new image. One of the purposes of such fiction, of course, is to alert us to the fictionality of our reading processes, to disable the partly automatic role they play in normal literary reading, the disabling that Moulthrop has signalled in his term "breakdown." While this can be an entertaining process, it remains questionable whether the self-conscious, ironic stance it imposes can be maintained for long, and whether such reading will come to replace the continuous "linear" narratives of book culture.

Without the "image" in place, and a continuing resonance of issues raised by our pursuit of characters in recognizable situations, the fiction remains fragmented. It will fail to absorb our interest, and the experience of reading beyond the interaction with the page celebrated by Birkerts will never develop: "When I am away from the book it lives its shadow life, its afterlife, and that, as the believers have always insisted, is the only life that matters" (p. 95). Perhaps it is not the only life, but it is a life that hypertext fiction, as theorized by Moulthrop, seems designed not to evoke.