The Shishosetsu as Novel
An'ya koro (A dark night's passing, 1921-37), whose hero Tokito Kensaku discovers that he is the product of his mother's and the victim of his wife's sexual indiscretion, has been variously described as a brilliant romance, as a youth's long and arduous but ultimately successful search for a mate, as one man's quest for happiness in the face of bitter experience and the humanization he undergoes in the process, and as the affirmation of a personal reality over the reality of the family or larger social order.[37] These attempts to describe, as we might a conventional novel, what the text is "about" are misleading, however, for they imply the existence of a well-conceived plot to which all characters and events are related, when in fact nothing could be further from the truth. Both the author's inexperience at writing sustained narrative and the work's composition over a period of many years militate against the narrative consistency we expect in a novel, but there are other reasons as well. This "fictional" shishosetsu , as critics are wont to
[37] The characterizations are made, respectively, by Kobayashi Hideo, "Shiga Naoya ron," in Shintei Kobayashi Hideo zenshu 4: 114; Miyoshi Yukio, "Kako no 'watakushi,'" 147; Nakamura Mitsuo, Shiga Naoya ron , 116; and Edwin McClellan, in the preface to his translation of Shiga Naoya, A Dark Night's Passing , 10.
call it ("fictional" because a few events and characters have no counterparts in the author's own life, and shishosetsu because of the nonetheless complete identity of hero and author), is at bottom more shishosetsu than fiction in the novelistic sense, as we shall see below.
An'ya koro grew out of the author's speculation, during a self-imposed exile in western Japan in 1913, that he had been sired by his grandfather and not by his father, toward whom he had long felt great animosity (SNZ 8:18). But rather than universalize the hero's predicament through an exploration of the complex ramifications of illegitimacy and infidelity, Shiga insists on turning both text and hero in on themselves. The reader is lead through a series of claustrophobic emotional soundings by a narrator whose range of awareness is virtually coincident with his hero's; at the conclusion he still possesses no insights into the "themes" that superficially inform the text.
It is no wonder, then, that such a starkly self-centered work should have its detractors. Dazai Osamu, himself an egoist of no small proportions, is positively venomous in "Nyoze gamon":
A Dark Night's Passing : what an overblown title this book has! Shiga often accuses other writers of hamming it up, but he should know that he is as guilty as the next fellow. He really plays to the crowd here. This is a fine example of "stacking the cards" in your favor. Where is the Dark Night, anyway, in this insufferable song to the author's ego? What is there to commend in a style so awash in self-conceit? The hero catches a cold; the hero suffers an ear infection—is this the Dark Night? The book left me absolutely stupefied. A piece of juvenile literature this is, fit only for a composition class![38]
Dazai's description is not inaccurate. An'ya koro is indeed a narcissistic exercise brimming with elitism and adolescent conceit, as Nakamura Mitsuo has also argued in his important critical study of Shiga.[39] And yet, for all its eccentricities, it remains a strangely powerful work, containing in its pages some of the most extraordinary and moving passages in all of literature. It is also, ironically, the greatest work of an author who will be remembered primarily
[38] Dazai Osamu zenshu 10:321.
[39] Shiga Naoya ron , esp. 148-52. Nakamura also writes: "[Shiga's] literature is essentially a young man's literature. It is founded on the dreams and fastidiousness and carnal melancholy of youth" (ibid., 184).
as a short-story writer. Miyoshi Yukio, whose critique of An'ya koro is much more thoroughgoing than Dazai's casual remarks, nevertheless manages to touch on what makes it a classic: “An'ya koro is a truly remarkable shishosetsu ... in which its hero's originality comes alive as its fictionality withers away. This is not an irony but a measure of its success. As a novel, it is riddled with contradictions, and yet by that very failure it has achieved its just reputation as a landmark text in modern Japanese literature."[40]
Miyoshi, then, postulates an inversely proportional relationship, which Dazai understandably overlooks, between the text's fictionality and the vividness of the hero's presence. Dazai accuses Shiga of flagrant exhibitionism, but he never questions the propriety of Shiga's retreat from fictionality, as he is no less eager to produce stories that aim at collapsing the distinctions between author, narrator, and hero. Indeed, Dazai's remarks sound like the petulant grumblings of a writer unable to market his sincerity as successfully as Shiga did. Dazai's jealous wrath would have been mollified, one feels, not by the insertion of a greater distance between author and hero in An'ya koro but rather by a less blatant display of smug self-confidence, a quality with which Shiga was prodigiously, and Dazai poorly, endowed. The question of the hero's autonomy as a fictional character is never raised.
Precisely that question, on the other hand, informs Miyoshi's essay, which lengthily catalogs the contradictions making An'ya koro a failure as a novel. The catalog is worth reviewing, along with other commentaries; for the exact nature of the "failure" will help us determine the ways in which the modern Japanese shosetsu differs from what we would unhesitatingly label a novel, a comparison Miyoshi suggests but does not articulate in any detail.
An'ya koro is characterized by an identification between author and hero that is virtually total. It is not so much the text's personalism that undermines its fictionality, however, as its peculiar narrative stance. Shiga's admission that he modeled his hero after himself
[40] "Kako no 'watakushi,'" 154. One indication of An'ya koro's enormous reputation is a poll appearing in Bungei 9 (June 1952): 16-69. Of twenty-four writers and critics asked to name the five best and/or most important works in Japanese fiction written since the turn of the century, fourteen included it in their lists. Toson's Yoake mae (chosen by six writers), Tanizaki's Sasameyuki (four), and Soseki's Meian (three) finished a distant second, third, and fourth.
is couched in terms any novelist might use: "The hero Kensaku is myself. [An'ya koro ] depicts him doing what I would do, or what I would want to do, or what I have actually done, under the circumstances" (SNZ 8:20). But few can be as guilty of such complete confusion between author and hero as Shiga is in part 3, chapter 5, which describes the hero's anxiety about how the circumstances of his birth might affect the courtship of his future bride:
It occurred to Kensaku that one way of telling the other party about himself was to write a long autobiographical story. But the project came to a stop after he had published the prologue entitled "The Hero's Reminiscences." He decided not to show them even this short piece, so afraid was he of seeming to be inviting their sentimental pity.
(SNZ 5:312)[41]
This curious passage makes little sense at first reading. We learn nothing more here about this "long autobiographical story" than the title of its prologue. Its contents are not revealed. We do not find out when or where the prologue was published or the title of the work as a whole or anything else to suggest that Kensaku had actually written a story of this description. There is in fact no other textual information to verify the existence of such a story—none, that is, save the An'ya koro text itself. Nakano Shigeharu, in an extended critique of An'ya koro , puts it this way:
People will puzzle over this passage—and then again, maybe they will not.... They apparently read the phrase "he had published the prologue entitled 'The Hero's Reminiscences'" and link it to An'ya koro itself without giving the matter another thought. An'ya koro , after all, has a "prologue entitled 'The Hero's Reminiscences.'" The prologue to the "long autobiographical story," then, is the prologue to An'ya koro .[42]
[41] Shiga Naoya, A Dark Night's Passing , 222. McClellan's translation, on which I have relied exclusively for my citations from An'ya koro , actually reads: "It occurred to Kensaku that one way of telling the other party about himself was to write an autobiographical story. But the project came to a stop after he had written the preface entitled 'The Hero's Reminiscences'" (emphases added). However, the original clearly indicates that the story in question was a full-length narrative (chohen shosetsu ) and that the prologue was not only written but saw the light of print (kakagerareta ). These are important distinctions in view of the following argument. Further, McClellan's choice of the word "preface" in the above passage and "Prologue" for An'ya koro's opening section appears to be an attempt to transform this self-reflexive passage into something more readily assimilable as fiction to a western audience.
[42] “An'ya koro zodan" (1944), 548.
In other words, the prologue authored by Shiga Naoya meets all the specifications for the one allegedly authored by Tokito Kensaku, with the result that the reader is encouraged to view the prologue's "I" as synonymous with both the narrated "he" and the narrating "I" of the main text. As Nakano notes, the reader was in effect reading An'ya koro as a text by Shiga and by its hero, Tokito Kensaku.
This identification of author with hero is of a very different order than of, say, Flaubert with his celebrated heroine ("Madame Bo-vary—c'est moi"). Nakano emphasizes that Shiga did not become Kensaku—that is, make an emotional investment in a character he could identify with; rather, he made Kensaku over into himself , mindless of the implications for the text as an autonomous narrative. "It is understandable that some critics would mistake the hero for his author in this particular text," Nakano writes. "The text itself is written and structured so as to create just such confusion."[43]
Shiga alerts the reader to the author-hero equation at the very beginning. The first chapter opens with Kensaku's reflections on a story just published by a cohort named Sakaguchi that describes a man's affair with his housemaid, which Kensaku believes to be based on Sakaguchi's own experience. Kensaku, exercising the Shiga hero's now-familiar prerogative as moral arbiter, finds the author Sakaguchi and his hero equally guilty of indiscretion: "[Kensaku] could have forgiven the facts if he had been allowed to feel some sympathy for the protagonist; but the flippancy, the superciliousness of the protagonist (and of Sakaguchi ) left no room for such sympathy" (SNZ 5:20; trans., 26; emphasis added).
Kensaku's analysis of Sakaguchi's story is a primer for the reading of any shishosetsu (including, of course, An'ya koro ), for it at once defines and problematizes the relationship of author to hero. The author is completely accessible through the text, we are made to understand. Once this convention is accepted, hero and author—the former being the equivalent and not simply a creation of the latter—become interchangeable; thus, Kensaku cannot talk about Sakaguchi's protagonist without talking about Sakaguchi himself, and vice versa. Because Kensaku's equation applies not only to his adversary in An'ya koro but also to the author himself, Shiga can
[43] Ibid., 551.
think of a prologue to a text he wrote himself as part of the text, a creation of its hero.
Just as conspicuous as the symbiosis of author and hero in An'ya koro is the tenuousness of plot. This derives, no doubt, from the telosless epistemology in the Japanese literary and intellectual tradition, which regarded life as an experience without goal or denouement. But another factor, namely, the circumstances of its composition, is also at work. Twenty-five of the twenty-six installments of An'ya koro were published over the seven-year period from 1921 to 1928, with the final installment (part 4, chapters 16-20) appearing nine years later. Although an extended composition time need not change the basic plan of a fictional narrative—one thinks, for example, of À la recherche or Finnegans Wake —it has an unquestionable impact on a shishosetsu , as we noted in the case of Shimazaki Toson's Shinsei in Chapter 5. Embedded in fictional time, a novel easily takes on the quality of a seamless narrative regardless of interruptions in the actual composition. We read it as a unified, monolithic artifact, playing the fictional time scheme against the conventional, temporal organization of the referential world we call chronological time. We read a shishosetsu , however, in a quite different way, for it is very much a product of the writing time—that is, the period or process of textual production and its temporal relationship to the events recorded . Any shishosetsu , but in particular one written over a long period, is shaped by the events it describes as much as it gives shape to them. This can be true even of shorter works when events are coincident with the time of writing, as we observed in the case of the "Yamashina" cycle. The composition process's potential for becoming a source of narrative disruption, along with the system of solicitations and deadlines that pressure many writers to rely on (usually) recent experience for textual production, is a principal reason for the form's brevity. In the face of narrative disintegration, short-windedness becomes a mechanism of textual self-defense.
When An'ya koro first appeared in the pages of Kaizo, readers were conditioned to regard each installment by this writer, famed for his short stories, as a separate unit. They were not able to examine the installments together as a whole until 1937, sixteen years after the initial publication. Nakano Shigeharu recalls that the individual installments gave the impression of finished stories in the
magazine, but they did not give one of completion when they appeared together between the covers of a book.[44] Nakamura Mitsuo likens An'ya koro to a loose set of exquisitely formed pearls that were never strung together into a necklace.[45] Shiga himself had originally thought of writing a series of short stories and later arranging them into a longer narrative. Such was the rationale, he later recalls, behind publishing the prologue and the final chapter of part 2 as separate pieces (SNZ 8:19). The discontinuities and inconsistencies between episodes and especially between the first and second halves[46] are legion. The animosity of the hero toward his father and the mystery of his birth, for example, which figure so prominently in the first half, fade into insignificance in the second. Although there are no appreciable gaps in fictional time, the numerous interruptions in publication, especially of the second half, have attenuated the work's thematic and psychological unity, as Miyoshi, Nakano, and a host of other critics have observed.[47]
Perhaps the main reason for this discontinuity is the existence of the unfinished manuscript Tokito Kensaku , which we know from Shiga's own words provided the author with a working draft of An'ya koro's first half (SNZ 8:19). Shiga's initial difficulties with Tokito Kensaku were rooted in his reluctance, which we noted in our discussion of Wakai , to express fully the ill feelings he harbored against his father, even though those feelings had played a central role in his writing career. His dilemma as an author could be re-
[44] Ibid., 539. The first half of An'ya koro appeared in hardcover as early as 1922 and is generally more familiar to readers than the second half.
[45] Shiga Naoya ton, 89. An'ya koro might also be compared to an emakimono , or picture scroll, in which there is not one but several points of interest that appear and then disappear from view as the scroll is unrolled and rerolled. Each scene and each small, faceless figure in the landscape is important in establishing the mood of the whole without necessarily being related dynamically to other scenes and figures. Linearity in such a scroll, and in texts like An'ya koro , is necessitated by the physical arrangement of scenes/words on the paper/page, but it is not structurally supported by a series of causal relationships.
[46] An'ya koro is presently divided into four parts, but it was originally divided into two halves (parts 1 and 2, and parts 3 and 4). Most critical discussions treat An'ya koro as a two-part text.
[47] See Miyoshi Yukio, "Kako no 'watakushi,'" esp. 98-105; Nakano Shigeharu, “An'ya koro zodan," 550 and passim; and Hirano Ken, “An'ya koro ron," a chapter in Hirano Ken zenshu , vol. 4, Bungaku: Showa junen zengo , 236-64, esp. 254-57. See also Akiyama Shun, "Shiga Naoya no 'watakushi' ni tsuite." Akiyama argues that the first half, a tale of "frustrated existence," and the second half, a tale of "fulfillment," are practically unlinkable and that the many interruptions in writing the second half reveal that Shiga himself was aware of the difficulties in linking them.
solved in only one of three ways: he could throw caution to the winds and write about the matter that concerned him most, regardless of the repercussions in private life; he could ignore it and write about something else; or he could stop writing altogether until the matter had resolved itself in such a way that he could resume his career without compunction.
Shiga chose the last, and the result was a silence spanning the years 1914-17. Interruptions in Shiga's writing career sometimes affected the content and even the very existence of a story, as in the case of "Kinosaki nite." Had Shiga successfully completed Tokito Kensaku , he might very well have inserted the incidents "about the bee's death and about the rat that had its neck run through with a bamboo skewer and was then thrown into the stream" that he had witnessed at a Japan Sea spa into the section of this story set on the Inland Sea, as his diary tells us he once planned (SNZ 10: 721). It was his failure to do so (despite the appeals of Natsume Soseki, on whose literary page in the Asahi the story was to have been serialized after Kokoro in 1914) that made it possible for Shiga's famous meditation on death to take form as a separate story, ending the three-year hiatus. The appearance of the bee, the rat, and the water lizard episodes in a context other than the one Shiga had originally intended signaled the ultimate dismantling of Tokito Kensaku . Then, less than half a year later, came the celebrated reconciliation. The "fundamental discord" between father and son having been resolved in private life, Shiga no longer felt the need to deal with it in his professional life. The father-son relationship as a universal question had never interested him; it was enough that his own predicament be resolved. And once he had written what was to be his most thorough account of the relationship with his father in Aru otoko, sono ane no shi (1920), which is itself a product of several earlier manuscripts that were probably related to Tokito Kensaku , he "completely lost interest," as he put it, "in themes that had concerned me up to then" (SNZ 8:18).[48]
It is thus surely no coincidence that the rift between father and son even in An'ya koro's first half is depicted only in a most understated way. The real-life clash of wills is transformed into a kind of
[48] An'ya koro manuscripts nos. 2, 14, 33 (6:28-33, 248-50, 359) all contain material used in Aru otoko . See "Koki," in SNZ 6:373, 404, 425.
natural catastrophe beyond the control of either party: the hero is the illegitimate child of his supposed grandfather. This tragedy, seemingly ordained by fate, effectively neutralizes the hostility of both parties by removing all the blame. The fiction of the hero's birth, as presented in An'ya koro , serves to absolve both father and son of any responsibility for their mutual hatred. And since even these remnants of violent emotion, along with the general atmosphere of discord, are a legacy of the prereconciliation Tokito Kensaku manuscript, they virtually disappear in the second hall which has no antecedent text and was of course begun after the reconciliation.[49]
The hero's attitude toward his father is only one of several problems of discontinuity for which the text seemingly offers no adequate solutions. It is odd, for example, that Shiga would let an earache (the same one that Dazai Osamu satirizes with such venom) determine his hero's movement from Onomichi to Tokyo and even odder that he should dwell at such length on the operation to clear the ear of fluid, until one realizes that Shiga himself suffered an earache while in Onomichi, which forced his return to Tokyo for treatment. The makings of dramatic narrative are here: Kensaku is on the verge of ending his solitary life at Onomichi and proposing to Oei, his maid and grandfather's former mistress. But here, too, because Shiga did not become Kensaku but instead made Kensaku over into himself he restricts Kensaku's perimeter of actions. And so Kensaku, rather than take destiny into his own hands once he has resolved to propose to Oei, merely tells himself lamely that "even if his ear had not given him trouble, he would probably have not stayed in Onomichi much longer" (SNZ 5:217; trans., 158).
Although it is of course hardly unusual for a writer to incorporate his personal life into, or even to make it the exclusive subject of, his work, few writers have done it so seemingly without regard for an overall narrative plan. Shiga appears reluctant to edit out any impressionable personal experience coincident with the hero's actions, regardless of its relevance to the text as a whole. Thus the inclusion of many extraneous scenes throughout An'ya koro , such as the extended description of Kensaku's train journey from Onomichi back to Tokyo in part 2, his visit to a Kyoto museum in
[49] Miyoshi Yukio, "Kako no 'watakushi,'" 80-81. Nakamura Mitsuo makes similar observations in his chapter on An'ya koro in Shiga Naoya ron .
part 3 and to a temple on the way to Mount Daisen in part 4, and the rather tedious episode in part 1 of Kensaku's vain attempt to nab his sister's would-be suitor. They are examples of the "finished stories," as Nakano calls them, which end up detracting from the completed text. As with the earache episode, the best explanation for their presence in the text is their chance occurrence in the author's life, as we know through Shiga's diary.[50]
Shiga attaches little importance to Kensaku's supporting cast. Characters like Oei, Kensaku's brother Nobuyuki, and Kensaku's grandfather lack the substance to interact with the hero in any significant way. True, these characters have no real-life antecedents: Oei and Nobuyuki are entirely imaginary figures and Shiga was at pains to create a character who bore no resemblance to his real grandfather, for whom he felt great respect (SNZ 8:19). But Shiga also fails to bring to life even characters that are clearly modeled on his own acquaintances. They flit across the pages like clouds scudding across the sky. The use of initials for several characters, moreover, seems to confirm Shiga's lack of deep interest in anyone but his hero.
Even Naoko, the most important person in Kensaku's life, is overshadowed by the introspective hero until the curious passage at the end, which views Kensaku from her point of view (see below). The nature of the hero's relationship with Naoko is presaged in Kensaku's first glimpses of his future bride. Unseen by her, Kensaku spies Naoko working gaily in a house facing the street in Kyoto where he takes his morning walks. It is not until much later, when he has already asked his friend Ishimoto to act as go-between and proposed marriage that he exchanges his first words with her. He sees Naoko's seduction by her cousin not as a marital problem that concerns them both but as a fiddle he poses to himself in isolation. It is the "emotional reverberations caused by Naoko's indiscretion," as Miyoshi Yukio puts it, rather than the affair itself, that are at issue for Kensaku.[51] Kensaku's reflections, as transmitted by a narrator whose sensibility overlaps his hero's, extend no further than his own mental state. Nowhere is this fact demonstrated more
[50] Endo Tasuku makes note of this in his annotations of the An'ya koro text. See Shindo Junko and Endo Tasuku, eds., Shiga Naoya shu , 232 n. 8.
[51] "Kako no 'watakushi,'" 123.
forcefully than in a scene in part 4, chapter so, following the episode in which Kensaku, in a "reflex action," pushes his wife Naoko off a train moving slowly away from the station as she tries to step aboard. Later, husband and wife sit alone, face to face. Never before in the work do we encounter such an opportunity for dramatic revelation of character. But when Naoko pressures Kensaku to explain his irrational act, he replies tersely:
What you have to understand is that, for me, everything is my problem, mine alone, to solve.... It's an egotistic way of looking at things, but given my nature, it may also be the most practical for me. You don't have a place in it, I know.
(SNZ 5: 514; trans., 356)
Here is perhaps the most startling in a text filled with startling illustrations of a character so completely absorbed in his own sensibility that he interacts only very awkwardly with his fellow human beings. No one, not even his wife, has a "place" in his way of looking at things.
Despite such attenuated plotting and characterization, we must not overlook the fact that An'ya koro does possess a certain continuity, as evidenced in the prologue that serves as an overture for the entire narrative and in the numerous foreshadowings of later events. The depiction of Oei in the prologue as a sensual being adumbrates Kensaku's later sexual attraction to her. The mysterious rejection by a prospective bride's family of Kensaku's marriage proposal hints at a blot in the hero's past. The soaring airplane, the object of admiration at the beginning, becomes a symbol of disillusionment at the end of An'ya koro . Kensaku's interest early on in the theatrical confessions of "Omasa the Viper" and in the geisha Eihana's stubborn silence is later linked to his wife's confessions of adultery. The nameless mountain in Hoki casually mentioned during Kensaku's trip to Shikoku in part 2 turns out to be Daisen, on whose slopes he attains a kind of spiritual release in part 4. Ishimoto, a friend of Kensaku's briefly introduced in part 1, chapter 3, reappears in part 3, chapter 5, as the go-between who arranges Kensaku's marriage with Naoko. The seeds of Naoko's adultery—a lewd game called "turtle-snapping turtle" she played with her cousin Kaname in her youth—are metaphorically planted in the very beginning of Kensaku's marriage discussions (part 3, chapter
5), which are held at a restaurant that specializes in, of all things, snapping turtle cuisine. But surely the most impressive evidence of topical continuity are the striking rites of passage in part 2, chapter 1, and part 4, chapter 19. Aboard a steamer bound for Onomichi—and on the slopes of Mount Daisen years later, in the hero's and in the author's life—Kensaku feels himself being absorbed into nature. The imagery in the two scenes, which is unmistakably parallel yet effectively contrasted, signals a crucial transformation in the hero's consciousness. Both episodes commence with the hero's physical debilitation: a spell of seasickness on board ship and a much more serious bout of food poisoning just before the mountain climb. Kensaku compares the steamer to "an enormous, silent monster" and the mountain to "the back of some huge beast" (SNZ 5:154, 579; trans., 114, 401). Most significantly, he yields completely to nature's embrace on the mountain slope, whereas before, aboard the steamer, he resisted the feelings of being "swallowed up by the great darkness surrounding him" (SNZ 5:154; trans., 115). These and other foreshadowings suggest an overall narrative plan, however loose, that was carried out in spite of the sixteen years it took to complete the text and serves to bind the seemingly "unlinkable" two halves.[52]
Shiga's own insistence in "Zoku sosaku yodan" (SNZ 8:22) that he did not deviate greatly from his initial conception is borne out by an outline (included in the collection of draft manuscripts and documents relating to An'ya koro in SNZ , vol. 6) that introduces many episodes that figure prominently in the text: the hero's wrestling match with his "father," his infatuation with his "grandfather's" mistress, his marriage, his journey to Onomichi and efforts to write, his discovery of the secret of his origins, the direction of his attention away from his father to his wife and children, his wife's adultery, his trip to Mount Daisen, his fight for his life at the story's end (SNZ 6:558-59). True, the final product diverges in
[52] The characterization is Akiyama Shun's. See n. 47. Hiraoka Tokuyoshi is one of the few commentators who stresses the works continuity. See “An'ya koro no sakusha," in Shiga Naoya zenshu geppo , no. 12:5-6. He cautions elsewhere, however, that the several "epiphanies" (he borrows the term employed by Saeki [Nihon no "watakushi" o motomete , 209-16] to describe a Joyce-like aspect of Shiga's fiction) dotting the text with memorable instants lack the Joycean resonances that light up an entire text. Bungaku no doki , 59. Takahashi Hideo also contests the dualistic interpretations of the two halves in his Shiga Naoya .
places from the outline. Kensaku marries a woman from a well-to-do provincial family rather than one from a boardinghouse family. He learns the truth about his birth from his brother Nobuyuki rather than his grandfather's former mistress. His last illness is the result of food poisoning rather than an accidental fall. And An'ya koro ends with his life hanging in the balance rather than with his eventual recovery. Several key motifs are missing as well, as Kono Toshiro, editor of the draft manuscripts, notes. There is no mention of the hero's cohort in dissipation, Sakaguchi, or of the physical and spiritual exhaustion that accompanies his life of debauchery, although both play important roles in the text's first half.[53] Yet it is precisely those unexpected gaps that make this outline so remarkable. Kono Toshiro dates it sometime between late 1918 and early 1920, or roughly contemporary with the two independently published stories that were later incorporated into An'ya koro : "Aware na otoko" (A man to be pitied, April 1919) and "Kensaku no tsuioku" (Kensaku's reminiscences, January 1920).[54] This means that even though the vast majority of the unpublished material treats incidents (particularly Kensaku's life of dissipation and his sojourn in Onomichi) that appear in the first half of An'ya koro and only the last three unpublished pieces (nos. 34-36) treat incidents appearing in the second half, Shiga clearly knew the direction his story would take long before he published its first installment.
Reading the draft manuscripts as a whole, however, one is left with an overriding impression of the author's uncertainty about the text's composition. Several manuscripts are little more than false starts. It is not simply a problem of the author's treating a great many sporadic (and at the manuscript stage, seemingly unrelated) incidents. He is clearly unsure of how to present his story. Interspersed with manuscripts written in third-person narration are those with a first-person narrator-hero (watakushi ), as well as a first-person narrator (jibun ) who chronicles the story of an acquaintance identified either by name or simply as "he" (kare ). The problem of continuity or narrative unity would seem, then, to revolve less around thematic content, as so many critics have suggested, than
[53] "Koki," in SNZ 6:426.
[54] Ibid. Kuribayashi Hideo dates the outline even earlier, from 1917 or 1918. See “An'ya koro soko no kento," 135
around the mode of presentation. And it is surely this mode, far more than any mere similarities between author and hero, that invites people to read the text as a shishosetsu despite its several clearly identifiable "fictions."
We have already touched on one crucial problem in presentation that bears further examination: the episode in part 3, chapter 5, describing Kensaku's publication of a first-person prologue ("Shujinko no tsuioku" [The hero's reminiscences]) to "a long autobiographical story" that we take to be none other than An'ya koro itself. Shiga published the prologue, we also noted, as an independent story one year prior to its inclusion in An'ya koro —except that as a short story it is a third person narrative entitled “Kensaku no tsuioku" (Kensaku's reminiscences). That is, what was written about Kensaku in the short story becomes Kensaku's own writing in the prologue to An'ya koro . But here we encounter yet another problem. The hero in the prologue can no longer be Kensaku, because it is Kensaku himself, we are told in part 3, chapter 5, who authors the prologue and creates a character who reminisces at the beginning of An'ya koro about that memorable series of events occurring in the hero's distant childhood. Just who is writing what? the reader wants to know. A piece composed by Kensaku has somehow been incorporated into a third-person narration. Does Kensaku exist separately from the narrator, or is he one and the same person?
Shiga apparently agonized over his very point in the manuscript stage—thus, his many experiments with both first- and third-person narration. His decision to dismantle Tokito Kensaku seems to have been motivated, perhaps even more than by the reconciliation with his father, by a desire to apply a new, more distant perspective to his material.[55] In draft manuscripts nos. 21 and 29-32, for example, he attempts to separate narrator and hero by making the former an acquaintance of the hero (already dead in nos. 29-30) who chronicles the latter's life. Shiga ultimately felt too constricted by this approach, however. Using this format to distance himself from his hero, he was forced to describe his hero only from
[55] Such is the argument of Hiraoka Tokuyoshi, “An'ya koro no sakusha." This incisive article, tucked away in an otherwise nondescript collection of memoirs that commonly make up the geppo genre, has been the springboard for much of the argument on An'ya koro .
the outside. The narrator could read what the hero had written, but not chronicle his day-to-day thoughts. So great a distance was, in the end, unworkable for a writer like Shiga.
Faced with the contradictory desires of creating an autonomous hero and representing his own mental state, Shiga arrived at a compromise commonly elected by shishosetsu writers. As in the "Yamashina" cycle, he simply transformed the narrating "I" (watakushi ) into a narrated "he" (kare ), thus "objectifying" his hero without, however, granting him the true autonomy found in a conventional third-person narrative. How was this possible? The answer lies in the character of the Japanese language, analyzed in Chapters 1 and 2 and again in the discussion of Shuko in Chapter 7. One cannot narrate the experience of another person in Japanese without fictionalizing or "lying"—that is, without assuming a pose of omniscience that allows access to the minds of others. Telling the "truth" is in effect telling only of one's own direct experience. What we have called the "written reportive style" differentiates between the relating of one's own and the surmising of another's experience by grammatical distinctions based on this epistemology. True, the narrating "I" as authorial persona is itself a kind of fiction, as we observed in Chapter 3, but no less powerful as a benchmark of authenticity. For a narrator to describe a character other than himself and still use the markers of direct knowledge (e.g., pure adjectives and verbs instead of adjectival and verbal suffixes like -rashii and -garu ) found in the written reportive style means either that he adopt the pose of omniscience and forfeit all claims to personally authenticated truth or that he present an authorial persona clearly understood as such by the reader. The shishosetsu writer, needless to say, consistently chooses the latter alternative. Thus, he has nothing to lose in the way of authenticity by narrating in the third person. When the conventions of the written reportive style are observed, a kare shosetsu is in the end identical to a watakushi shosetsu . The first-person / third-person distinction being negligible in Japanese, Shiga has no need to "become" his protagonist in order to describe the workings of his mind after the manner of Flaubert; he simply makes the latter over into himself, as Nakano Shigeharu suggests.
Nowhere is this interchangeability more clearly demonstrated than in the curious passage in part 3, chapter 5. The prologue itself, as mentioned earlier, was originally published separately as a third-
person narrative? Shiga could have avoided altogether the overlap of authors (Shiga/Kensaku) and the resulting circularity had he retained the prologue in its original form, since "Kensaku's Reminiscences" would then not need to be identified strictly with "The Hero's Reminiscences" in part 3, chapter 5. What, then, motivated Shiga to make the switch in narrative person? His aim, it seems clear, was to lend his fiction an otherwise impossible authenticity. The hero's first-person voice in the prologue reverberates throughout the entire text. Even after the narrative switches from the first-person watakushi to the third-person kare in the main text, the hero continues to monopolize the point of view, and his sensibility infuses the narrative consciousness. We cannot see around him.[57]
Shiga's apparent slip in part 3, chapter 5, then, actually reveals his narrative intent: to lend a fictional character the authenticity of his own consciousness. So absorbed is Shiga in his hero that his own consciousness at the time of writing seems to displace that of Kensaku and determine Kensaku's actions and thinking, even when they contradict the work's internal fictive design. To say that Shiga is his own hero is to express all too inadequately the radical nature of this profoundly antirepresentational act. Skeptical of the power of language to represent another world, he refuses to invest his
[56] Other than the transformation of kare into watakushi , the prologue appears virtually unchanged from its original form, with only a few minor revisions, all of a semantical and not a grammatical nature, except for one change of tense. (The prologue does not include the last section of "Kensaku's Reminiscences," which describes a frog-killing expedition.) Compare "Kensaku no tsuioku" in 5:604-11 with the textual revisions to the original magazine version of "Shujinko no tsuioku," in "Koki," in SNZ 5:648-49. The parenthetical note originally prefacing "Kensaku no tsuikoku" ("Tokito Kensaku was the illegitimate child of his mother and grandfather. But he did not learn this until he had reached his mid-twenties" [ibid. 5:604].) is deleted in the prologue.
[57] One important exception to this monopoly of point of view will be discussed presently. McClellan's translation makes it appear as if the narrating consciousness is more independent of its hero than it actually is. In trans., 43, to take one of several examples, the narrator states flatly that Ishimoto “was only being polite" with respect to Kensaku's request that he attend a geisha party, when in fact the original says, "Ishimoto wa ... amari kyomi wa nai-rashikatta “ (SNZ 5:45). In trans., 93, the narrator notes, "[Kensaku] did not know it , but his face still looked cross," whereas the original makes no such assertion of knowledge about the hero: "Kensaku wa ... mada okotta yo na kao o shite-ita" (SNZ 5:121). In the prologue, McClellan has the first-person narrator make the kind of judgment about another's feelings that Japanese grammar will not allow. His translation of "Chichi wa itsu ni nai aiso-rashii koto o watakushi ni itta" (SNZ 5:12) is a much more direct: "[Father] spoke to me kindly “ (trans., 20). (Emphases added.)
hero with a discrete set of emotions generated by the text itself. Shiga's insistence in this way on the equivalence of, rather than the mere correspondence between, author and hero practically eliminates the fictional autonomy of the text's time-space continuum; at the same time, however, it presents a powerfully persuasive world of higher authenticity, namely, the narrator's own consciousness. Hiraoka Tokuyoshi puts it this way: "It is precisely because the authorial persona is half 'third person' [kare ] and half 'first person' [ji-bun ] that Shiga is able to pass off part of the autobiographical story that his hero Kensaku writes as the prologue to his own text. This became possible, of course, only after he transformed the story 'Kensaku's Reminiscences' into a first-person narration."[58]
This merging of the author and hero's consciousness, which becomes explicit in part 3, chapter 5, is common in the shishosetsu and already familiar to readers of Shiga in such works as Wakai . Needless to say, it breaks the rules of conventional, "realistic" fiction. The author who breaks them must therefore ask himself, just where does his narrator fit in the narrative? "In order to see all," Hiraoka argues, "the writer must focus on himself as writer. To do so, however, is to tread the thin line between ... art and artifice. That Shiga succeeded was due to his discovery in Tokito Kensaku of a protagonist who enacted the role of both 'other' [kare ] and 'self' [jibun ] and yet who was neither a completely third-person 'he' or first-person 'I.'"[59]
So preoccupied is the An'ya koro narrator with the hero's sensibility to the exclusion of all others, so blatantly does this shosetsu call itself into question as a novel, that the western reader is led to reconsider altogether his strategy of classification. Merely to say, however, that An'ya koro differs radically from the fictional norm is, in the words of Kobayashi Hideo, like telling an elephant that its nose is too long—too long, that is, if one is comparing it with, say, a lion's.[60] We must, in other words, refrain from using the conventional novel as a standard of measurement if we are to understand the text's enduring place in modern Japanese literature.
[58] “An'ya koro no sakusha," in Zenshu geppo , no. 13:5.
[59] Ibid., 7.
[60] "Shiga Naoya" (1929), in Shintei Kobayashi Hideo zenshu 4:16. Kobayashi satirizes those who would compare Shiga with Chekov. Miyoshi Yukio borrows the comparison in his discussion of An'ya koro . See "Kako no 'watakushi,'" 128.
What, then, is the standard to which we can appeal for an informed assessment? How can we characterize the text in a way that accounts for its extraordinary power? Nearly every commentator has pointed to Shiga's inimitable ability to project his presence on paper. Miyoshi Yukio argues that An'ya koro "works" because it succeeds, as any good shishosetsu does, in communicating the quintessential individuality of its author-hero.[61] Nakano Shigeharu argues that the great attraction of the An'ya koro text lies in the author's fastidious concern with his persona.[62] Nakamura Mitsuo, however, attributes Shiga's appeal to his invention, or at least perfection, of a contentless self: a transparent hero with virtually no distinguishing traits of his own, who gives the reader the illusion that he has actually come in contact with the real author.[63] No matter that Kensaku is a mere "bundle of sensations" living in a hermetic world; it is his virtue that he is utterly reliant on the reader's knowledge of his creator for any specificity of character. The contentless hero becomes a fitting vessel for the author/narrator's presence.[64]
This does not mean, of course, that the reader comes face to face with the author. Yet it is in the nature of Japanese narrative in the written reportive style that the reader identifies with the narrator's voice (whether expressed through the first-person watakushi or third-person kare ) and no other; for it speaks with a directness that allows for no transindividual consciousness. This characteristic is common to first-person narration in any language. But it is common to "third-person" narration as well in Japanese, for as we have seen, the deployment of kare in a shosetsu does not itself result in an autonomous third-person narrative. The con-
[61] "Kako no 'watakushi,'" 129.
[62] “An'ya koro zodan," 560. One of the decorum-conscious author's techniques, Nakano suggests, is his use of euphemisms in describing the less savory actions of his hero. For example, he employs a phrase like "Shintai dake wa, kare wa masu masu hoto no fukami e otoshite itta no de aru" ("With every passing day, his self-indulgence became more intense"; SNZ 5:133; trans., 101) for what simply means "Kensaku visited the brothels even more frequently."
[63] Shiga Naoya ron , 20—22; 137ff.
[64] Ibid., 90. Saeki Shoichi emphasizes that the very refusal of a shishosetsu writer like Shiga to invest his hero with a plenitude of presence makes it possible for a reader to identify all the more closely with the character in the belief that he is the same authorial persona he has known all along. See Nihon no "watakushi" o motomete , 5-25, esp. 10-11.
tentless kare is actually the watakushi in a disguise every reader recognizes.
The penultimate chapter marks the fruition of Shiga's first-person / third-person collation. The Shiga hero loses his "self" on the slopes of Mount Daisen, holy to the Tendai school. It is here, Hiraoka con-dudes, that Kensaku ultimately attains to a transcendent, person-less (in both ontological and grammatical senses, we might add) world where the hero's consciousness is absorbed into nature's vastness.[65] As if to emphasize this melting into the void, the author deletes the hero's name during nearly the entire lengthy passage at the end of part 4, chapter 19, which describes the state of blissful rest on the mountain slope, and instead punctuates the narrative with an occasional kare .[66] Kensaku literally disappears during that fleeting, eternal moment of meditation at daybreak. The name reappears only when Kensaku returns to himself at chapter's end and descends the mountain at the beginning of chapter 20.
But what a different character he has become! In An'ya koro's final chapter, we can see for the first time around the hero—that is, view him from a perspective other than the hero's own. In fact, we see the gravely ill Kensaku, delirious from the food poisoning that afflicted him on Mount Daisen, through the eyes of each character appearing in the chapter: Oyoshi and her mother the priest's wife, the doctor, and finally Naoko. Here, at the very end, An'ya koro sheds its shishosetsu skin and emerges as the kind of "realistic" narrative we are accustomed to as readers of western fiction. The authorial persona whom we have come to know so intimately as the Shiga hero fades away and is superseded by a group of characters, with Naoko as the central figure, that comport themselves like characters in a conventional novel. We are party for the first time to information beyond the hero's ken and to conversations out of his earshot. The hero, at the threshold of death, yields his monopolistic point of view to a suddenly omniscient narrator intent on transforming his shosetsu into a novel.
[65] “An'ya koro no sakusha," in Zenshu geppo , no. 13:7.
[66] The passage begins:" ... finding a suitable resting place in the grass, he sat down with the mountain at his back." (SNZ 5:578; trans., 400.) McClellan appears to attach particular importance to this fact, for his translation duplicates Shiga's usage of the proper name in this section even though it varies considerably elsewhere.
With Kensaku no longer at the core of narrative consciousness, An'ya koro begins to operate in a radically new way.[67] Is the hero's sudden exit from center stage simply a fluke? All indications are that it is not. Shiga makes no definitive statement about the reasons behind the abrupt change of narrative focus, but certain peripheral evidence is too suggestive to ignore.
First, there is the rebuttal ("Nakamura Shin'ichiro-kun no gimon ni tsuite" [1948]) to a critique of An'ya koro's finale, in which Shiga emphasizes that the shift in narrative perspective was no accident and not even unique, referring to the earlier description of Naoko's indiscretion in part 4, chapter 5.[68] He then mounts an attack on those who would scruple over narrative method.
Fastidious readers may think it odd that the hero suddenly disappears from the scene, but such people are surely in the minority. I think that the general reader, going through the work rapidly, would not be bothered by the passage. It is in any event the product of a calculated risk and not of carelessness. I was fully conscious of what I was doing. I found that this approach best suited my purposes, but I was also motivated by the desire to take a certain risk.
(SNZ 8:159)
[67] The hero's displacement from the narrative center, and his metamorphosis from actor to acted upon, is nicely realized by the scene that concludes part 4, in which he finds himself in Woman's hands as he lies before Naoko, his hand in her lap. This contrasts vividly with the scene that concludes part 2, in which he literally takes Woman in his hands as he fondles a prostitute's "round, heavy breast" (SNZ 5:275; trans., 197).
[68] Part 4, chap. 5, and part 4, chap. 20, differ substantially in at least one sense, however. The final chapter is an attempt at truly omniscient narrative that adopts several points of view; whereas part 4, chap. 5, which appears in English as omniscient narrative, is best interpreted as hearsay reported through Naoko. (The English translation sometimes moves away from Naoko's point of view, but the original never wavers. For example, the sentence "Kaname suffered from extreme stiffness around the neck and shoulders, and complained constantly about it" [trans., 256] is a rendering of "Kaname wa kata ya kubi no hageshii kori de, hidoku kurushigatteita “ [SNZ 5:482]; emphasis added.) This is Shiga's not altogether felicitous solution to the problem of how to transmit a great deal of information, to which Kensaku has become privy, in a conciser form than dialogue. Evidence for such an interpretation is supplied by a section of the original text (coming between part 4, chap. 5, and part 4, chap. 6) that was deleted when the complete An'ya koro came out in book form. The section begins: "Kensaku did not of course hear the story precisely as given here" ("Kensaku wa mochiron, kono hanashi o sono mama kiita no de wa nakkata ga") and continues with Kensaku addressing Naoko: "You have done what you should by telling me about the incident" ("Omae wa sore o uchiaketa to iu tokoro de, suru koto wa sunda no da" [SNZ 6:614; emphasis added]). The overall impression received from reading this chapter and the one that follows is one of Kensaku remembering what Naoko has told him about the incident. He has internalized her point of view, and that is why he so boldly announces later, part 4, chap. 10, that "everything is my problem, mine alone, to solve."
Shiga never discloses the exact nature of this "risk." But his opinion that the litterateur would stumble over a passage to which the general reader would not give a second thought suggests that he attempted to create a new, truly novelistic, narrative voice even at the risk of distancing himself from his hero's sensibility, which any critic attuned to the world of junbungaku aesthetics would be loath to accept. That he made the attempt at all is remarkable, considering what he must have thought to be the limitations of the language, as we can surmise from an essay published two years prior to the one quoted above.
Scarcely six months after Japan surrendered to the Allied Powers, when people were suffering from starvation, disease, and economic collapse in addition to the agony of defeat, Shiga wrote, in an essay entitled "Kokugo mondai," of still another crisis he believed that the people faced. It was one that, although seemingly not as pressing, could prove to be the most perilous of all: the state of the Japanese language.
Although we may not be sensitive to the fact (since we are accustomed to our own language from childhood), there is in my opinion no language more imperfect or inconvenient than Japanese. Considering how much the language has impeded the nation's cultural advancement, I believe that we should deal at once with the onerous problems it poses.
(SNZ 7:339)
Shiga goes on to recount the plan by Mori Arinori, the Meiji government's first minister of education, to adopt English as Japan's national language, and he argues in essence that Japan would have been spared the tragedy of war had Mori's plan been carried out. He then suggests that the time is ripe to adopt French, known for its logical clarity, as the national language, in the same pragmatic spirit that the Japanese showed when they adopted the metric system of weights and measures.
Shiga's belief that the adoption of English in the Meiji period would have resulted in a more civilized and less bellicose nation is laughable, to be sure. It is hard to take seriously (especially in today's economic climate) Shiga's attack on the language as being an impediment to advancement. Japanese may be "imperfect" in that it is constantly undergoing change to meet new conditions, but it is inherently no less perfect than any other language. Yet it will
also not do merely to discount the essay as so many cranky remarks uttered in a period of lost national pride or as wholly representative of the stream of tirades against the language and culture that gushed forth at war's end, as some critics suggest.[69] Indeed, if we read this foray into cultural criticism in light of Shiga's calling as a writer of fiction, the words "imperfect" and "inconvenient" take on new meaning.[70] For what was Shiga, as the would-be producer of an omniscient but still "authentic" narrative, to call a language if not imperfect, if that language sounds unnatural when it attempts to incorporate more than a single point of view? And what, if not inconvenient, if that single point of view must merge with the narrator's own? Shiga may have shared his countrymen's postwar frustration when he penned his critique against the native tongue. But his frustration would also seem to stem from a creative malaise. Having exploited the language's genius as fully as any author, was he not perhaps looking wistfully to new horizons that he had glimpsed through translations from western literature?[71] To become an author with all the power that the word implies; to be the mover of his hero, and not be bound by him—was this not the "risk" that Shiga took, the bold experiment he embarked on when penning An'ya koro's last chapter?
[69] Nakamura Mitsuo, for example, suggests the attack is a sign that Shiga's growth as an artist has stopped. See Shiga Naoya ron , 9 Kinda'ichi Haruhiko opens his well-known book, Nihongo , with a defense of the Japanese language against Shiga's attack. Suzuki Takao refutes Shiga's essay in the first chapter of his Tozasareta gengo and insists that the problem of "imperfection" lies in Shiga's imperfect use of the language and not in the language itself (25-26).
[70] For a dissenting interpretation, see Roy Andrew Miller, Japan's Modern Myth , 109-15. Miller's book, which shows how the Japanese have placed the language at the center of their cultural identity, correctly debunks the notion that any single language is unique by noting that all languages, by their very existence, are "unique" in some way; but it dwells rather excessively on the masochistic pleasure he claims Japanese take in the difficulty, oddity, and "illogicality" of their language. We have noted throughout, meanwhile, not the ontological uniqueness of Japanese but rather certain features that distinguish it in particular from western languages, in the belief that uniqueness and particularity are not the same thing and that the latter can be studied profitably without conceding the former.
[71] The question of how a western language translates into Japanese, important though it is, is too complex to go into here. Suffice it to say that it requires a radical transformation, as suggested in our discussion of the Japanese-to-English translation of Mon in Chapter 2. To the extent that its omniscient voice prevails in the translation and is not assimilated into the Japanese epistemology, however, a western novel inevitably comes off as a gross "fabrication," as Kume Masao and others have suggested apropos of Balzac, Tolstoy, et al.
This experiment, however, triggers the text's final crisis. Kensaku's critical illness embroils the narrative itself in a life-and-death struggle for survival and eventually deals it a mortal blow. The narrator's particularized voice, wedded to the written reportive style, finally cannot survive, or transcend, the hero's consciousness. When An'ya koro closes with Naoko's silent soliloquy—the first and last uttered by any character other than Kensaku—what we hear is an uncharacteristic, sexually neutral language that sounds uncannily like Kensaku's own speech: "Naoko wa 'tasukaru ni shiro, tasukaranu ni shiro, tonikaku, jibun wa kono hito o hanarezu, doko made mo kono hito ni tsuite iku no da' to iu yo na koto o shikiri ni omoitsuzuketa" ("[Naoko] kept on thinking, 'Whether he lives or not, I shall never leave him, I shall go wherever he goes'" [SNZ 5: 589-90; trans., 408]).[72] It is as though Kensaku's soul has taken flight and migrated to the body of Naoko. Or, on second thought, are those words not more appropriately spoken by the narrator of his hero? For when the former parts from the latter, it means the end of them both.[73]
Shiga's narrative experiment signaled not only the end of An'ya koro but for all intents and purposes the end of his writing career. His last three and a half decades, by all accounts, were spent living out contentedly his hyphenated role of author-sage. It was a role more securely played in private life than on paper and one that he played with apparent ease. For Shiga's goal in writing, as Hiraoka Tokuyoshi observes, was always a more satisfactory personal life.[74] Writing, in other words, was a means, not an end—a form of auto-therapy, as we noted at the beginning of this chapter, that led to a
[72] The only commentator to make note of Naoko's decidedly nonfeminine language is Donald Keene. See Dawn to the West 1:467.
[73] There was of course a more "practical" reason for the text's completion: a plea from Shiga's publisher to finish writing in time for its inclusion in a newly edited zenshu . Thus the appearance of the final and longest (part 4, chaps. 16-20) installment in Kaizo (Apr. 1937), after a nine-year hiatus. Such pressure (to say nothing of the fact that the work was already quite long) no doubt encouraged Shiga to end his narrative where he did, but that alone does not explain the motivation for his narrative experimentation—unless , knowing that he had to finish writing soon, Shiga did not feel obliged to extend his novelistic project. At any rate, it would have been impossible for him to pursue this project and still remain true to the traditional linguistic and narrative rules governing the production of "authentic" fiction in Japanese.
[74] “An'ya koro no sakusha," in Zenshu geppo , no. 13:6.
fulfilled silence. Dazai's inference in "Nyoze gamon" that Shiga had not written anything significant for two decades (excepting, we would protest, the final chapters of An'ya koro ) is correct;[75] but such is hardly grounds for condemnation, since Shiga had passed on to a new stage: from writing to living. Shiga was always more interested in attaining peace with himself than in searching at all costs for the meaning of "self." This very limited and concrete notion of selfhood—true to the traditional epistemology and the product of a particularized narrative voice—is of course what characterizes the shishosetsu in general and so much of Japanese literature. Shiga saw his task as one of resolving the contradiction within himself as observer and as actor, as writer and as private man. "The part of me that observes is completely divorced from the part that acts," Shiga writes angrily in a short piece included in the collection of An'ya koro manuscripts. "I must make myself whole. I must hold onto something, something very basic, something that will form a world all my own.... Without it I cannot confront existence, cannot survive. I must find it. But I don't know how. I will go to the mountains and think" (SNZ 6:60-61).
The logical conclusion of this train of thought is that literature is subordinate to life, that the goal is to be a content person, not a prolific writer. If we accept Shiga's motive for writing—to bring an emotional crisis to conscious light and liberate oneself from its burden—then we can see that this is precisely the goal that Shiga pursued. One might argue that this is the reason all writers write; but few writers have recognized, as Shiga did, its ramifications: if realized, it leads the writer to silence. For Shiga, literature was useful only insofar as it fulfilled a particular need in his life; once the need was met, he could discard his art as he would an outgrown pair of shoes. For Shiga, silence in the end signified a healthy reintegration into life. Granted, he was a fastidious stylist, intensely concerned with imagery and tone and profoundly aware of the genius—and the limitations—of his language. But looking at matters in the above light, perhaps we can say that his ultimate aim as author-sage was to do what most private citizens do, and that is not to write at all.
[75] Dazai Osamu zenshu 10:325.