吴琼 | 乔治·迪迪-于贝尔曼的症候图像学

 【摘 要】

乔治·迪迪-于贝尔曼是在“二战”之后法国理论语境中成长起来的一位激进的艺术史家，他一方面运用新的阅读技术对西方经典艺术史理论的核心概念和写作技艺进行批判性的考察；另一方面又在此基础上兼收并蓄从中世纪至当代的各家理论，尤其融合精神分析学的症候阐释学、瓦尔堡的间隙图像学和本雅明的时间辩证法，提出了自己的症候图像学。可以说，“症候”不仅是迪迪-于贝尔曼进行知识批判的工具，也是他开展图像阐释实践的方法论。

20世纪晚期，在法国巴黎，以社会科学高等研究院（École des Hautes Études en Sciences Sociales，以下简称“高院”）为中心，以路易·马林（Louis Marin）、达尼埃尔·阿拉斯（Daniel Arasse）、于贝尔·达弥施（Hubert Damisch）和乔治·迪迪-于贝尔曼（Georges Didi-Huberman）——不妨称之为高院“四剑客”——为代表，一种激进的或者说批判的艺术史写作悄然兴起。如今，“四剑客”中的前三位都已经去世，唯有迪迪-于贝尔曼仍笔耕不辍，新作迭出。而在写作姿态上，“四剑客”中又以迪迪-于贝尔曼最为激进 ，无论于古今理论间的纵横捭阖还是对历史文献的旁征博引，也无论在图像阐释时以“错时”（anachronism）论强力介入还是在展陈文化中对图像潜能蒙太奇式的激活，他的每一次出场都足以在建制化的艺术史和图像“世界”撕开一道症候性的“伤口”。因此，他注定成为近百年来最值得关注的艺术史家之一。

作为一位艺术史家，迪迪-于贝尔曼涉足的领域十分广阔，理论借用亦十分驳杂，以至于英语世界需要出版以他为主角的“词典”或“导论”一类的书籍，才能为读者清理进入其理论学说的路障，尽管用这些传统文类去处理迪迪-于贝尔曼的“实验性”写作，其适用性还有待商榷。正因其理论驳杂，加之受到21世纪初“瓦尔堡热”的影响，使得许多学人都试图从几个被视为理解迪迪-于贝尔曼的重要角度切入他的理论体系，比如症候/现象学、遗存/时间性及宁芙等，国内对他的研究也集中在关于遗存和宁芙的议题。的确，如果非要为迪迪-于贝尔曼的理论给出一个具有辨识度的命名，“症候图像学”无疑最为恰当。隐藏在症候背后的思维路径，一条是在“新理论”的情境中对经典艺术史写作及其“图腾式概念”的考古学批判，另一条则是对症候、时间性、蒙太奇等概念的方法论重启。也就是说，迪迪-于贝尔曼看似狂野的写作实则有其一以贯之的脉络，为人们所关注的各个角度在他那里都有着内在的关联。

一、直面艺术史话语：模仿与象征

1982年，迪迪-于贝尔曼出版了他的第一部专著《发明歇斯底里：沙科特和硝石库医院的摄影图像学》（La invención de la histeria: Charot y la iconograf ia fotográf ica de la Salpêtrière，以下简称《发明歇斯底里》），该书以巴黎著名的精神病医院硝石库医院（Salpêtrière Hospital）的摄影档案为素材，讨论医院院长让-马丁·沙科特（Jean-Martin Charcot，弗洛伊德曾求教于他）在对歇斯底里（症）的诊断中对影像的利用。迪迪-于贝尔曼在书的开篇说到，他感兴趣的是在摄影中“患者与医生之间非凡的共谋关系，以及欲望、凝视和知识的关系”，以及摄影术作为凝定“歇斯底里的妄想”与“知识妄想”二者关联的工具，又是如何帮助医生“发明”歇斯底里这个知识的。这个陈述颇似曾同样利用该医院档案研究疯癫史且最终病逝于该医院的米歇尔·福柯在《临床医学的诞生》中的开场白——“这是一部关于空间、语言和死亡的著作。它论述的是目视”。但又不仅仅是在向伟大的福柯致敬，实际上，《发明歇斯底里》更像是《临床医学的诞生》的续篇，都是利用医学档案对现代“目光”与现代身体“知识”之间的关系展开“知识考古学”式的审查。

但是，它又不止于此。迪迪-于贝尔曼所写的并不是一本医学史意义上的歇斯底里“诊断史”，他要借歇斯底里如何被“发明”的议题去讨论早期摄影术及医学摄影所涉及的面孔、姿势、时间和凝视的摆置技术/艺术。就像他说的那样，摄影术的曝光过程意味着这是一种“操纵时间”的技术，而它的可复制性又消除了这种操纵性，他把这一技术悖论转换为视觉悖论，并称之为瓦尔特·本雅明意义上的“灵晕”，“就这样，灵晕命名着时间燃烧、发出声响、令形象莫名哑然的方式……它召唤我们面对本雅明所谓‘光学无意识’：接触和距离在可见领域中的刺点、盲点”。从这个意义上说，《发明歇斯底里》又像是本雅明《摄影小史》的续集，是一本“灵晕”图像志。进而，如果看一下书中展示的照片及迪迪-于贝尔曼对照片的阅读，还会发现它们就像是“明室”的暗面，是一本罗兰·巴尔特意义上的“刺点”图像志——一本满是悲怆面孔和挣扎姿态的图像志，尽管巴尔特的名字只是频繁出现在注释中。

惊艳亮相的《发明歇斯底里》已然昭示迪迪-于贝尔曼在图像研究领域不走寻常路的野心，在1984至1988年，他在意大利访学，那里丰富的图像现场——那些关于面孔和姿势的视觉形式、关涉悲怆和死亡的视觉经验——无疑极大地激发了他的研究热情，直接的成果就是其于1990年出版的两本专著，分别为《直面图像：质疑艺术史的终结》（Confronting Image: Questioning the Ends of a Certain History of Art，以下简称《直面图像》）和《弗拉·安杰利科：异似与喻象》（Fra Angelico: Disguise and Symbol）。彼时，他的兴趣点已经转移到艺术史和艺术图像之上[实际上，1985年，他就在专著《图绘肉身》（Images in Body）中围绕巴尔扎克的小说《不为人知的杰作》讨论了艺术图像的问题]，更确切地说，转向自瓦萨里以来西方的艺术史写作实践和图像的“视觉”实践。尤其在《直面图像》中，迪迪-于贝尔曼对艺术史话语的批判性考察看似依然在延续福柯的考古学方法——他自己后来称之为“批判的艺术史考古学”，但实际是在延续《发明歇斯底里》中已然明确的论题：表征、可见性、时间、症候等。受到弗洛伊德的启示，迪迪-于贝尔曼将歇斯底里的症候视作能指之踪迹；这种只能像处理艺术品一样迂回接近的“症候阐释”技术，如今被作为处理艺术史写作和艺术图像中“症候”的基本方法。

与更倾向于设置意识形态化议题的英美“新艺术史”相比，迪迪-于贝尔曼“批判的艺术史”对艺术史写作本身有着更为明确的方法论自觉，他的考古学首先要揭示的是，作为一种历史书写行为和话语实践的传统艺术史，何为其理论的基本构型，以及这一构型内部的裂缝何在。迪迪-于贝尔曼说，为解决图像中那些令我们感到困惑的问题，历史上出现了一种有关艺术作品的特殊知识，进而发展成为一门学科——艺术史。这门学科以一种“确然的语气”告诉我们，可见物中的一切在这里都可以得到医学诊断一般的解释，因为这门学科确信“表征是统一地发挥功能的，是一面准确的镜子或透明的窗户”，不论在直接的或可见性的层面上，还是在间接的或可理解性的层面上，表征“都能将所有的概念转变为图像，将所有的图像转变为概念”。因此，我们观看一幅艺术图像的行为，就成为如何“命名”眼睛所见东西的过程，“看”屈从于“知”，“可见”屈从于“可读”。

然而，图像中总有一些不可阅读的纯粹视觉之物，会扰乱这一纯粹理性的理想主义情境。例如弗拉·安杰利科（Fra Angelico）在佛罗伦萨圣马可修道院3号冥想室墙上绘制的《圣母领报》（The Annunciation，图1），看似有着阿尔伯蒂（Alberti）意义上可资信服的宗教故事，或者潘诺夫斯基意义上清晰可读的主题或“知识单元”；然而画面中那些同样存在的、缺乏动态与色彩渲染、不同于同时代同主题画作中惯常出现的繁复细节与空间再现等表达处理方式，除了将其归因于艺术家想象力的贫乏，似乎很难找到其他合理解释。可是，当我们的眼睛逐渐适应密室的幽暗，当背光墙面上的画作逐渐清晰，我们就会发现，作为画作背景的那堵墙，那堵用石灰粉涂抹的白墙，除了一片纯粹的白色，什么也没有：没有图画，没有符号，没有意义。似乎除了纯粹的白色本身，它什么也不表达，它是不可读的，所有建立在再现或象征意义上的图像理论在它面前都无能为力。进而，一旦我们看到或关注到这片白色，它就会变成一道暧昧又刺眼的光，以其无意义的自身“迎向”我们，白色或它那如反光一般的“光晕”就开始对我们进行“凝视”，这如同来自实在界的凝视，将把我们拖向“非知”的深渊。就像《直面图像》的开篇所言，“时常，当我们将目光投向一件艺术图像，总有一种迎面而来的悖论感。那即时地和直接地抵达我们的东西，就像某个有点含混的自明性，总夹杂着某种困惑”。

图1  弗拉·安杰利科，《圣母领报》，1439—1443年，壁画，176厘米×148厘米，意大利佛罗伦萨圣马可修道院藏。

问题的症结在哪里？答案就在艺术史正典所确立的那些理想主义观念上。迪迪-于贝尔曼对艺术史话语的考古学批判就是从“直面”这些观念开始的。为此，他重点选择了两个人物进行讨论：作为“艺术史之父”的瓦萨里和确立现代图像学为一门人文科学的潘诺夫斯基，同时他着重考察了“模仿”和“象征”这两个概念。

瓦萨里的《艺苑名人传》不仅是“现代”（moderna）的第一部艺术史，还为现代艺术史写作建立了一个“复杂的、层叠的合法化程序系统”，借这一系统，该书确立了艺术家“新的主体地位”及艺术史家作为“新人文主义者”的身份。迪迪-于贝尔曼以“症候阅读”的方法对瓦萨里的合法化程序系统进行分析，逐一揭示结构化程序的操作效能及其逻辑“内爆”，其基本的结论可以归纳为：“瓦萨里确实有一个体系，但这是一个有裂隙的体系。”不过迪迪-于贝尔曼更为关注的是瓦萨里为“现代”艺术从生产到批评所建立的“艺术规则”，其中最重要的莫过于“模仿”这个“图腾式概念”，“文艺复兴所再生的就是对自然的模仿。这是伟大的图腾式概念，是所有母亲——艺术的大母神，是这种亚宗教的至高神”。然而，如同瓦萨里的艺术史体系是一个充满裂隙同时又不断缝合裂隙的体系一样，被文艺复兴人文主义者推崇为最高信条的“模仿”也是一个不确定的、混合的、什么都可以往里装的“魔法概念”，一个“漂浮的能指”。

比如这个概念以“相似性”为基本原则，瓦萨里在各艺术家的传记中使用频率最高的评价性术语皆与相似性有关：“栩栩如生”“逼真”“自然”等，某种意义上说，他的美学话语就是一个“相似性神话”的话语。在一篇专门讨论相似性神话的论文中，迪迪-于贝尔曼总结道：“在文艺复兴这个理想化的或理想主义的历史运动中，相似性扮演了魔法般的角色，成为一个本质性的术语，一个有关手段、源头和最终目的的术语：‘现代’意义上的艺术品生产必须掌握的手段；这一生产的真正源头；作为生产之目的而必须向其返回的原初意义，即一般视觉艺术的目的论。”

可这个原则并不如人们想象的那么稳固。为支撑相似性的神话学地位，瓦萨里不时地在传记的故事结构中插入一些模仿或相似性的神话，而以德里达的角度看，这其实就是逻各斯中心主义的“增补逻辑”，对此，迪迪-于贝尔曼称之为“叙事的症候”。其中最典型的莫过于为“现代”绘画之父乔托（Giotto）所铺陈的模仿神话——一个由上帝之手引导的超自然的、非自然的英雄“复活”神话。这个神话故事有一个奇迹结构——文艺复兴本身就是一个“复活”奇迹，乔托就是奇迹事件的开启者：还是一个10岁的孩童时，为父亲照看羊群的乔托出于天性，像一个原始艺术家一样在岩石和沙地上画下自己的奇思异想。到乡下办事的奇马布埃（Cimabue）恰好路过此处，他对这个没有文化的“自然人”源于自然的模仿才能惊叹不已，于是便将乔托带到了城里；如同经历从自然到文化的转变一般，乔托在城里完成了语言的转化，“发明”了艺术史起源的语言，故事的高潮便是乔托在执政官官邸的礼拜堂画了一幅伟大诗人但丁的肖像。一个文化场所和一种文化语言，一个享有社会声望的地方和一个声名显赫的人物，诸要素汇聚到一起，构成了一个“现代性”和现代性的“起源”。迪迪-于贝尔曼评论说，瓦萨里通过将一个故事的各个元素关联、安排和辩证化，并将它们“想象地投射到一个独特的、晶莹剔透的、神迹般的，或更简单地说声望颇高的事件中。这如果不是神话还能是什么？”也正因此，在瓦萨里的《艺苑名人传》中，肖像画的发展成为衡量文艺复兴艺术史的一条暗线，肖像画或肖像式再现是文艺复兴艺术的另一个“图腾式概念”，是凭借“生动自然地刻画真人”而享有优先地位的原始“文类”。可其中的悖论在于，这不过是“基于一个完美地虚构‘图腾’的视角”，因为乔托时代是不可能在佛罗伦萨出现流亡者但丁的肖像的，瓦萨里所指认的那件但丁像根本就不是乔托绘制的，或者说这个绘画事件根本就没有发生过。

如果说瓦萨里是将绘画归于再现、把再现归于相似性和可见性的表征，那么潘诺夫斯基的图像学就是在图像与象征、可见与可读或者说“看”和“知”之间建立了一种等效的联系。迪迪-于贝尔曼对潘诺夫斯基图像学思想的转变，以及在引出图像阐释三层次的理论时所使用的脱帽致意的例子，同样给予了详细讨论，但最终也落实到他的核心观念之上。迪迪-于贝尔曼指出，通过一种修正的康德主义，潘诺夫斯基明确了图像作为象征、图像学作为人文科学的地位，并将已然正典化的模仿观念汇入又一个“图腾式概念”即图像学中，“它告诉我们，艺术图像既模仿可见的，也模仿不可见的，它告诉我们，绘画、雕塑和建筑的可感‘形式’，就是为了移译理性‘形成的’那些不可见的概念或观念”。而这些又都是建立在对所谓综合、统一、结构主义式的符号学对应等观念的确信基础上，建立在对图像中“非知”的视觉元素的压抑和排除基础之上。

就这样，传统艺术史将所有的图像都交付给概念和定义、可见和可读的“暴政”。迪迪-于贝尔曼总结道：“我恰恰认为，艺术史是一种典型的‘现代’现象——它诞生于16世纪——它想通过赋予艺术图像新的目的来埋葬可视性（visual）和可喻象性（figurable）的古老问题，这些目的将可视性置于可见性（visible）（和模仿）的暴政之下，将可喻象性置于可读性（和图像学）的暴政之下。”而他所要做的，就是提供自己的替代方案，“针对那种图式化的和历史简约的统一模型，我们代之以喻象和症候的理论范式，我们认为这些范式可以更恰当地阐述图像深层的‘象征’效能这一始终悬而未决的问题”。

二、直面图像：喻象与症候

直面图像的“直面”，在法语中为devant，这个方位介词的弱意义就是“在……面前”“和……面对面”，但迪迪-于贝尔曼还赋予了它一种强意义即“朝向”“迎向”。同时，他沿着动词化的方向将它处理为观者与对象“面对面”的双向运动：直面图像既是图像“面前”的我“朝向”图像，也是我“面前”的图像“迎向”我；并且这一双向运动并不意味着最终会导向观者与对象的审美契合，比如通过某一意义或价值的指认达成主体与对象的统一，恰恰相反，如果观看仅限于对象的可见方面，那我们对图像的爱欲终归是一场错付，因为总有一些“非知的”或不可视的东西从目光的捕捉中逃逸，此时的“面对图像”，就是一次朝向失败的旅程，或者说就是见证失败的过程。也正因如此，在图像面前，唯有从失败开始，“直面”失败，“直面”图像对我们的质疑，才能开启另一个旅程——“直面图像”的旅程，迪迪-于贝尔曼称这个过程是一种“目光现象学”。

如何从理论的角度理解“对图像的爱欲终归是一场错付”？这涉及图像作为“对象”的弱意义和强意义。作为弱意义的对象就是传统艺术史所理解的图像：图像是象征，是能指和所指一一对应故而有着确定的意义或价值表达的符号，但如迪迪-于贝尔曼所言，这只是可见性层面的一个理想主义假定；作为强意义的对象根本不是居于客体之位的知识对象，而是一个运作，一个时间的“结”（knot），一个多元决定的“机器”，是可视性意义上的“喻象”（figure），在它的面前，可见和可读的失败正是阐释的开始。要明白这里所说的意思，就需要看一下《弗拉·安杰利科：异似与喻象》中的相关讨论。

“异似与喻象”两个概念均源于中世纪神秘主义神学家伪狄奥尼修斯：“dissimilitudo”原指世俗外表的欺骗性，但在否定神学的语境中，它还指原罪之后世间万物作为上帝形象的“残迹”在否定和肯定之间的辩证法；“f igura”原指眼睛所见之外的东西，而在神学语境中，又指对最本质的“所见之外”即上帝形象的表达。基于这样的理解，将这两个词分别译作“异似”和“喻象”似乎最为恰当。迪迪-于贝尔曼从否定神学借来这两个概念，也是要暗示“多明我会”僧侣画家弗拉·安杰利科对伪狄奥尼修斯及修会神学家讨论图像与不可见的奥秘的相关文本的“阅读”，这一阅读给予画家的并非直接的图像指导，而是否定神学的图像思维，其思维方式与后来被艺术史人文主义化的图腾式概念即模仿和象征的思维方式恰好相悖。

比如“异似”针对的就是追求相似性的“模仿”。按照伪狄奥尼修斯在《天阶秩序》中的观点，上帝作为绝对的超本质是不可见和不可表征的，为显示这个绝对，最恰当的办法就是以“不相似的相似”来进行表现，“神不是一种本质，甚至不是所有本质的完美结合。上帝是绝对的超本质。因此，怎么能以相似的方式将其诉诸形象”。迪迪-于贝尔曼甚至在激进的否定神学的意义上将异似同原罪，进而同谦卑关联到一起，在此意义上，一只蚯蚓远比一位戴着王冠的国王更适合上帝的形象，因为上帝越是为我们降低自己，愈显谦卑，我们得到提升或者说迎向上帝的机会就越大，“蚯蚓的形象就是这一过程的模仿：没有什么比蚯蚓更谦卑的了。它是所有动物中最没有身体形状的，它几乎与赖以为生的泥沙尘土融为一体”。

从圣像学的角度看，一方面，“不相似的相似”蕴含了一种反偶像/图像崇拜的图像知识，但从图像本身的角度看，“不相似的相似”其实就是一种喻象。所谓喻象，字面理解就是以此物喻彼物，在神学语境中，不可见的奥秘才是最本质的喻指对象，比如图像中圣母领报的主题，“如果说领报的神学真实就是于可见性之外抵达真实性，那一个画家该如何描绘可见性‘之外’？他该如何喻象一个预言，一个瞄准未来的神圣文本的记忆？或者说喻象一个奥秘，一个超自然？”另一方面，就喻象终归还是“象”而言，它指示的并非诉诸可见性的形象，而是一种构型（conf iguration），即“可见世界的构型”。进而，按15世纪在“多明我会”修道院盛行却未进入人文主义艺术理论的一种观念的理解，喻象不是可见性的呈现，也不是图像学意义上的符号对可读性的转译，相反，喻象是要引入异质性——“以喻象呈现某物，不在于呈现事物的某个方面，相反，它指的是赋予事物不同的方面，改变它的可见性，引入异质性、他在性。简言之，喻象某个事物，就是用一种方式而不是通过它的某个方面来指示它”。

就像前面提及的弗拉·安杰利科在修道院3号冥想室图绘的白色墙壁，它不是以可见、可读的方式呈现故事，但它也不是完全的不可见，它有其诉诸视觉的可见性，即物质性的白色。非常具体的白色，无可辩驳地在场的白色，在安杰利科那里，这个白色不再是能指和所指一一对应的表征性代码，相反，它是以异似性的“图像”对表征的打开，它是一个“事件”，以自身无可辩驳的在场亦即可视性扭结了众多可能的或异质的“意义”：修道院的墙壁、领报圣母房间的虚拟墙壁、物质性的白色颗粒、反射性的光源、什么都不是的“空无”等，由此使得“领报”这一事件及其神秘的奥义不可名状地得以涌现，“弗拉·安杰利科简化了模仿领报之表象的所有可见手段，为的是赋予自己一个适合‘模仿’领报‘过程’的视觉代理……在此，喻象的奇迹出现了，以其梦一般的自明性吞噬着我们的万有形象。有这个特殊的白色在那里就足够了。如光一般强烈……如岩石一般晦暗……其单纯的呈现使它成了一种充满障碍的、不可能的光的材料：带着自身神秘的挥发作用的墙面”。

再比如安杰利科在冥想室外过道的墙面上画的《阴影中的圣母》（Madonna delle Ombre，图2），为了给图像主体部分，即圣母子与圣徒在一起的“神圣的会话”中那些巧妙地模仿的面孔提供一个姿势和异似的对照，画家为“神圣的会话”提供了一个“基座”：四块仿彩色大理石的挡板，上面布满如同雨点一般喷洒的色斑，从某种角度看，这个“基座”不仅可见，而且有其神秘的仪式源头，它支撑着圣母子和圣徒的形象，如同祭坛支撑着它的祭坛画一样。在此意义上，不妨说它就是神圣者的喻象、基督墓地的喻象或神秘仪式的喻象，总之它被赋予了“丰沛的释经和沉思潜能”；同时，这个纯色彩的喷洒也是一个可视的症候，它一举摧毁或“撕裂”了那个时代的艺术所推崇的模仿，它通过对阿尔伯蒂式的窗户或镜子结构的拒绝，回到了绘画的“绝对拟古主义”，它似乎展示了“一种源头、一种原初的图画姿态”，它用一种异似展示了“用残余颜料再现某个神圣的、不可企及的、却又是激发画家整个创作欲望的对象的全部谦卑”，这串色斑虽然在外观上全无相似性的模仿，但它恰恰类似于一个过程，“一种涂油的姿态，甚至祝圣的姿态，它不是模仿，而是搬演（换句话说，重现，再次具体化）”。

图2  弗拉·安杰利科，《阴影中的圣母》局部，1450年，壁画与蛋彩，193厘米×273厘米，意大利佛罗伦萨圣马可修道院藏。

由此可见，喻象并不是直接再现或者象征地表达奥秘本身，喻象的本质是迂回（detour），就像本雅明的“辩证图像”或星座所形成的旋涡，“喻象总是为另一个喻象呈现奥秘，那另一个喻象又为再一个喻象呈现奥秘，如此等等，围绕奥秘就形成了一个旋涡”。因此喻象实质上或者说首要地是一种诉诸视觉或可视性的置换运作，是持续的“虚拟实践”，其目的就是要在时间和空间上分离的不同事件或人物之间建立一种联系，例如绘画侧面地描绘领报故事中的预言和盼望，就是在尝试这种虚拟实践，试图在自身之内生产某个类似于领报结构的东西。迪迪-于贝尔曼说：“通过建立两个事件、两个物体或两个存在之间的关系，以及由此而来的两个不同的‘类像’（similitude），喻象就可以包括或展示那使它们分离的异似（dissemblance）：这是成为纯粹的奥秘本身的一种方式，是避开故事的一种极端结果。因为一个避开故事的画家在他的喻象中生产的将不再是某个时刻的时间，不再是某个空间的场所。”

既然喻象的本义就是用一个东西来指代或置换另一个东西，那么这一工作方式与弗洛伊德描述的梦或无意识的工作机制恰好相似。实际上，喻象也是弗洛伊德理论的术语，指无意识以视觉化的方式表现自身，并且在弗洛伊德那里，当无意识以喻象的方式工作，所产生的就是“症候”（symptom）。迪迪-于贝尔曼由此将喻象从中世纪的神学语境引向弗洛伊德和拉康的精神分析学语境，这一引渡固然是基于二者貌似的一个交集点，即都指涉了某个不可表征之物，更重要的还在于它们的工作方式都具有“不相似的相似”或者说异似的特征，就此而言，迪迪-于贝尔曼所理解的症候似乎只是以精神分析学的方式对喻象概念的改造。不过，他的做法更为激进，症候的引入不只为了改造喻象，还为了解释喻象，是要给喻象/图像提供一个阐述技术、一个方法论工具。

迪迪-于贝尔曼特别强调，“症候”概念虽是借自精神分析学，但与临床应用没有任何关系。对于自己的症候概念，他有一段论述：

症候是一个关键事件、一个奇异点、一个侵入，但它同时也是一个意指结构的实施，一个责成事件涌现的系统的实施，但只能部分地、矛盾地涌现，其方式让意义仅仅表现为一个谜或“某种东西的表象”，而不是一组稳定的意义。这就是为什么症候同时以其视觉强度、其辐射价值及弗洛伊德的所谓其“隐藏”“正在起作用的无意识幻想”的适合程度为特征。因此，症候是一个两面的符号实体：介于辐射和掩饰之间、意外和自主权之间、事件和结构之间。

同时，迪迪-于贝尔曼强调，症候的两面性使得对它的运用必定涉及两个理论领域：现象学领域和符号学领域，而症候模型在处理艺术史问题时，关键就在于两个领域的“接合”（articulation），“有必要提出一种现象学，不仅涉及可见世界作为共情环境的关系，而且涉及意义作为结构和特殊工作的关系（这以一种符号学为前提）。因而还应当提出一种符号学，不仅涉及符号配置，还涉及事件、意外事件或绘画图像的奇异点（以一种现象学为前提）。这是有关绘画的一种症候美学或者说一种自主权意外事件的美学必须要做的”。

迪迪-于贝尔曼一方面把症候当作一个知识批判的范畴，去反思图像学所理解的象征；另一方面又将它作为一种阐释工具，去运作图像内部的能指滑动和时间间隙。针对图像学将图像视作能指和意义相嵌合的稳定结构，迪迪-于贝尔曼特别强调症候的两个特征：撕裂和多元决定。同样地，它们既是知识论的，也是方法论的。

在弗洛伊德看来，症候是由矛盾或冲突的本能产生和维持的，是被压抑的冲动的坚持和返回，但受压抑者并不直接表现为症候，而是以喻象式的置换将自身展示为冲突的结构，也就是说，只能以迂回的方式接近症候。症候作为表征根本上是冲突的表征，是表征内部的“撕裂”（rend）和对“撕裂”的维持，它抵抗综合，甚至它本身就是一种“撕裂的力量”，它向我们“讲述着结构自身的撕裂、平衡的断裂，讲述着一种新平衡、一种很快又会破裂的前所未见的平衡”。

进而，既然撕裂是表征结构的一部分，那就应当用撕裂来思考表征结构。在《不顾一切的图像》（Images malgré tout）中，迪迪-于贝尔曼说：相较于坚持模仿幻觉的人，他自己的著作“自一开始就站在相反的方向，那就是撕裂图像”。不妨说，撕裂就是“直面图像”的姿态，是迪迪-于贝尔曼自己“直面图像”的姿态的喻象，它构成了对潘诺夫斯基面对图像的那种“看即是知”的姿态的倾覆。撕裂是一种打开行动，它所要做的就是“用结构的裂隙来思考结构（表征的结构），用功能的断裂或机能失调来思考功能（象征性功能）”。

症候具有表征的功能，但它表征的是结构的不稳定、意义的不确定，因为它的符号只是漂浮的能指。就像安杰利科的白色，承载着多重可能的意义，它是意义的凝缩、置换和变形，并以“结”的方式形成了图像的一种症候，传达着迷宫般的意义轨迹。也正因此，症候是多元决定的，“多元决定‘打开’了症候的‘时间’。它只能通过冲突或模棱两可的元素进入现在，这些元素本身又指涉着其他冲突或模棱两可的元素，或者说过去的但仍在持续的记忆元素，它们通过赋予其症候一定的形式来扭曲主体的现在”。

同样地，症候的多元决定也预期了一种开放的、意义无限延宕的阐释过程，这尤其适合于对图像“细节”的阐释。不同于传统艺术史和艺术理论将细节理解为从图像整体中“截取”局部、再做放大式的“接近”、然后又通过“相加”将其还原或汇聚到整体之中的理想主义做法，迪迪-于贝尔曼强调，细节是能指之网编就的一个织体，是扭结了多重意义的“结”，“绘画的每一个细节都是多元决定的”。为什么如此强调“细节”的多元决定呢？实际上，迪迪-于贝尔曼反对把图像视作一个统一的再现体系和综合的意义系统，他的症候图像学，不论作为一种知识批判还是作为一种阐释工具，都是对作为完整系统的图像的“爆破”，症候就是那个引爆装置，而非表征性的细节就是症候藏身其中的矿藏，因而症候的抓取就是对这种细节的提取，可用它来完成对图像的撕裂。

例如老彼得·勃鲁盖尔（Pieter Bruegel the Elder）的《伊卡洛斯坠海》（The Fall of Icarus，图3）。这幅充满诸多令人难以理解的细节的画作，曾引起艺术史家的广泛讨论，后者无非都是试图从奥维德的《变形记》等假定的源头文本中寻求实证的解释，然而画作中有一个地方却很少被人作为需要解释的“疑难”而对其细节化，那就是伴随伊卡洛斯的落水而漫天飞舞的细小羽毛，因为人们很容易就可辨认出，这就是飞得太高的伊卡洛斯被高温熔化的蜡制羽毛脱落的细节再现。可迪迪-于贝尔曼没有止步于这种可见性的“看”和“知”，他认为如果仔细看一下那个被叫作羽毛的物质，就会发现没有明显的特征使其与身体落水所产生的泡沫区分开来；它们只是一些近似白色的笔触，既非单纯的描写，也不完全是叙事，它们既像羽毛又不完全像羽毛，既像泡沫又不完全像泡沫，亦即它们是漂浮的能指，“介乎一个所指即‘羽毛’和另一个所指即‘泡沫’之间，是纯图画性的，是纯白色”。正是细节的这一“既像……又不像”的表征悖论，正是这一介乎“之间”的“间隙”，让整个图像的寓意表达变得多义而含混。

图3  老彼得·勃鲁盖尔，《伊卡洛斯坠海》，约1560年，布上油画，73.5厘米×111厘米，比利时布鲁日画家艺术博物馆藏。

进而，为了让自己的细节理论与传统的细节美学相区隔，迪迪-于贝尔曼“发明”了一个术语：“色面”（pan）。“Pan”在法语中指布料、墙面、碎片、服装边饰等，迪迪-于贝尔曼将这些意义揉在一起，指示绘画中某个色块部分颜料的物质性、非表征性、随机性、结构的撕裂性等。“色面是画面内涂料的一种症候，‘涂料’在此应在质料因的意义上理解，‘质料’应在亚里士多德归于它的意义上理解，所关联的不是矛盾双方的逻辑，而是欲望和延展的逻辑。”与细节描写的物神化特征不同，色面是歇斯底里的，是细节的整饬性的扰乱，是图像内部结构撕裂的标记。其实，早在1985年讨论巴尔扎克的小说《不可见的杰作》时，迪迪-于贝尔曼就注意到了这种色面，他称那件“不可见的杰作”中对色块的使用——用混乱的色块再现“绝对之美”的身体——不过是“无对象的外观游戏，飘忽的游戏：一场几乎没有或什么都没有的游戏，显象和隐象的纯粹随机性。因此，这是一种症候——而且是绘画的纯粹症候”。弗拉·安杰利科的白墙和色彩喷射、勃鲁盖尔被标记为羽毛的白色，以及维米尔诸多作品中的颜料堆积，当然都是这样的色面。例如维米尔藏于卢浮宫的一幅小画《花边女工》（The Lacemaker），画中看似有着极度清晰的细节，有着阿尔珀斯的所谓“描绘的艺术”，但在画面左边前景的位置，会看到一团红色，传统观点认为那是一团线，但其中潦草的笔触和厚涂式的颜料堆积，与旁边花边女工手指间的两根线所展示出的艺术家精确的形式掌控能力完全不相称，它其实就是一团流动的红色颜料，“它的轮廓似乎游移不定；它的图式本身构成了一个斑点”。正是它的不可辨认，使得追求逼真性的表征幻觉主义在此遭遇“挫败”，但这不是艺术的失败，相反，它是艺术家精心设计的一个“乱源”——以色彩的侵入性和辐射能量彻底颠倒或扰乱了“部分与整体的传统关系”。

三、直面时间：错时/遗存

2000年，迪迪-于贝尔曼另一本重要的著作《直面时间：艺术史与图像的时过境迁》（Devant le temps: Histoire de l’art et anachronisme des images，以下简称《直面时间》）出版。开篇他就指出：

每当我们在图像面前，也就是在时间面前。像卡夫卡故事里的那个可怜的文盲一样，我们面对图像，如同面对《在法之前》：就像在一道敞开的大门面前……但这是什么样的时间？什么样的可塑性和断裂，什么样的时间节奏和时间震动，能在图像的这种敞开中岌岌可危？

这当然是艺术史意义上的时间，是隐身在图像内部的历史时间，也是艺术史本身的时间。迪迪-于贝尔曼认为，艺术史无所谓“诞生”，艺术史的每一次革新都是重新开始：瓦萨里将“文艺复兴”视作古代艺术死亡后的重新开始，他的“艺术史”亦是普林尼之后的重新开始，那里有一个历史死亡和复活的时间/时代神话；18世纪温克尔曼的《古代艺术史》则是瓦萨里之后的重新开始，该书按照风格的所谓“发生、发展和死亡”重建了一个现代意义上的历史范式或时间模型；而随着阿比·瓦尔堡对古代艺术及其时间的重述，对艺术和历史的思考又一次发生了根本性的转变，瓦萨里和温克尔曼所谓艺术的“生与死”“鼎盛与衰落”这类知识模型遭到解构，“自瓦尔堡之后，我们再也不会以过去同样的方式直面或站在图像的面前和时间的面前”。

图像的时间不是一般的历史时间，而是时间的间隙，是多元决定的时间，是异质的时间在某个动态相遇点的碰撞。在图像面前也即在多元决定的时间面前，是不同的历史时间的相遇，迪迪-于贝尔曼将此称为“错时”。错时不是时间或时代错误，而是不同时间的交织，在知识批判的意义上，错时论不仅是对传统艺术史进化论或循环论的时间观的重击，也是对迈克尔·巴克森德尔的所谓“时代之眼”的重击，是对艺术史书写建制的时间禁忌的刺穿，因为该禁忌有一个黄金法则：拒绝时代错误，其中首要的就是不要把我们自己的现实——概念、趣味和价值观——强加或“投射”到过去的现实或历史研究对象上。我们只能通过过去、通过对过去的共情来理解过去和过去的物品，就像弗拉·安杰利科的彩色表面，我们只能通过寻找那个时期的来源来解释这种绘画选择及其背后的美学、宗教和艺术意义，可在错时论的角度看，对时间一致性的这一寻求不过是时代和时间的理想化想象，同时代人不一定意味着他们必定有同时代性或共同的情感。就像安杰利科，他与同时代的阿尔伯蒂就不在同一时间，他的那些带有原始喷射痕迹的色彩火焰与阿尔伯蒂理解的作为“镜子”或“窗户”的绘画就属于完全不同的时间——除非把前者“仿”大理石挡板的幻觉视作阿尔伯蒂意义上的模仿——其作为一种时间症候，构成了时间多元性的表征。“错时是表达图像的丰富性、复杂性和多元决定性的时间模式，在仅存的弗拉·安杰利科彩色壁画的例子中，至少有三个时间——三个异质的时间，相互关联的错时性时间。”这三个时间分别是：同时代阿尔伯蒂式的代表模仿概念的错视画的时间；13至14世纪“多明我会”修士论述色彩纪念功能的记忆术的时间；更早的伪狄奥尼修斯的异似的时间。从这个意义上说，“图像的历史就是时间上不纯粹的、复杂的、多元决定的客体的历史”。

迪迪-于贝尔曼的错时思想深受20世纪初三位德国思想家阿比·瓦尔堡、瓦尔特·本雅明和卡尔·爱因斯坦的影响。例如在《直面时间》中，他就通过三位思想家讨论了抽象绘画中的时间辩证法，尤其是“现代”与“过去”在相遇中的相互质疑和相互重塑。

在一幅图像面前，无论它有多古老，现在永远不会停止重塑，只要目光的褫夺没有完全让位于“专家”的虚荣自满。在一幅图像面前，无论它有多新，无论它有多当代，过去永远不会停止重塑，因为图像只有在记忆的构造中——即便不是萦绕不去——才变得可以想象。最后，在一幅图像面前，我们必须谦虚地承认这个事实：它可能会比我们活得更久，在它面前我们是脆弱的元素，是过渡的元素，在我们面前它是未来的元素，是永恒的元素。图像往往比凝视它的人拥有更多的记忆和更多的未来。

而在《触感类像：痕迹的考古学、时代错置与现代性》（La ressemblance par contact: Archéologie, anachronisme et modernité de l'empreinte）中，他再次融汇三位思想家的观点，从错时角度考察了多样且异质的“远古”翻模技术（imprint）是如何在与当代艺术实践（如杜尚的现成品艺术）的相遇中形成“一道闪光、一个星座、一个错时物体的辩证图像”。他还表示：“错时的视角首先是一个时刻，是错时的一个考验，在历史缺席的时候将自身强加于历史，这不是要取代历史，而是要在以前被历史忽视的某个点让它重生。”

三个人中影响最大的当然还是阿比·瓦尔堡，更确切地说，迪迪-于贝尔曼是将瓦尔堡从被压抑的历史中解救出来的关键人物，这一解救行为从某种意义上说就是一次错时行动，是迪迪-于贝尔曼以其多元决定的症候美学从不同方面对瓦尔堡的时间的差异性进行重写，但其实也可以说是后者的历史时间对迪迪-于贝尔曼的症候的差异化，因为正是对瓦尔堡以“时间幽灵”的方式重新召回，才使迪迪-于贝尔曼摆置“幽灵时间”的症候找到了理想的道具。

在《遗存的图像：阿比·瓦尔堡眼中的艺术史以及种种幽灵的时间》（L’image survivante: Histoire de l’art et temps des fantômes selon Aby Warburg）中，迪迪-于贝尔曼对瓦尔堡的时间模式进行了多面向的描述：

历史的“幽灵模式”，在那里，对时间性的历史时代的确定不再按学院式的知识传承，而是由幽灵、“遗存”、残余来表达，并体现为形式的持续返回，也就是由不构成知识、未经思考的概念及时间的无意识方面来表达。而在最后的分析中，这所谓幽灵模式也是一种“心理模式”，是在心理角度构成返回的意义上说的，不过不是对理想角度的返回，而是对后者理论解体之可能性的返回。也因此我们在此拥有了一种“症候模式”，在那里，形式的出现和改变将作为张力所体现的过程的象征物而获得分析。

实际上，这三个称谓也可被视作迪迪-于贝尔曼对瓦尔堡的“古代遗存-情念范式-宁芙”三元组思想的症候化表述，在他看来，瓦尔堡的这些概念就是艺术史传统的时间范式的批判性替代，“如何超越瓦萨里连绵不断的传人所想象的历史观来描述时间的流动线团？如何超越那些被小心翼翼又层级森严地划分的活动亦即学院派所谓‘美术’来描述图像的流动线团？遗存和情念范式概念的引入就是为了回答此类问题。它们旨在让研究文艺复兴视觉文化的历史学家更好地理解多元决定的含义，以应对图像的多义性和可塑性，并在事物和符号的领域开展深入的工作。‘遗存’一词让他或她能够把握历史时间的多元决定；而‘情念范式’的表述让我们可以理解西方文化中常见的拟人化表征有意义的多元决定”。

瓦尔堡毕生都在关注一个问题：“古代遗存”（Nachleben der Antike）在文艺复兴艺术中的“返回”，瓦尔堡的“Antike”固然包含年代学的意义，但他关注的不是历史中存在过的“古代”及其物品，而是那个时候的文化因子甚或所谓“生命能量”（它们就寄存在古代的形象、神话、文学描述、占星术等媒介中）在后代经由各种媒介不断地传播、变形和重生的过程，所以“古代”在时间进程中总是以“貌似”的样子出现，“Nachleben”就是“貌似”的古代，是古代的“劫后余生”，或者说是现代的“拟古”形态，是古代在现代的“记忆痕迹”，是散发出古代风味的“遗韵”。迪迪-于贝尔曼认为，“遗存”就是瓦尔堡对历史时间的界定，遗存是一种症候，其所指示的道路是理解时间幽灵的最佳方式，因为历史时间是不纯粹的、是多元决定的，亦即错时是历史时间的基本情境，“最古老的事物时常出现在不那么古老的事物后面”，但它在“后来”出现时几乎摆脱了“以前”的样子，比如15世纪借某种形式的遗存在图像中返回的女酒神信徒，不再是古希腊的形象，而是标记了该形象的各种变形的幻影：先是古典的，然后是希腊化的，接着又是罗马的，最后在基督教的背景下被重新配置，它是一个离开又回来且在重复中不断差异化的类像。简而言之，“古代遗存”在艺术史经验的层面被称为古代的影响，在人类学的层面则是古代以“拟态”的形式在现代的“幽灵般”返回，这是时间性的位移，是寄存在时间性之上的生命经验和历史经验的象征化，因而也是过去的时间/生命体验和现在的时间/生命体验在某个“遗存-形式”上的相遇。这意味着，“古风”之为“遗韵”的价值，就在于它是时间的辩证法，迪迪-于贝尔曼评论说：“遗存只是症候，是时间失序的承载者。”

这种时间遗存的具体形式就是瓦尔堡所说的“情念范式”（Pathosformeln）。瓦尔堡认为，情念范式不是简单的图像形式，而是负载了生命能量的形式，是以象征的方式对生命能量的置换。人类生命能量的传递总是体现出沉静和骚动的两极性，而总是在两个极性之间来回摆动，则使得人类文化的情念范式在根本意义上或者说在最后的分析中总是体现为优雅和悲怆两种形式或形象构型。迪迪-于贝尔曼承认，瓦尔堡的情念范式是两极化生命能量的形式结构，“没有两极性，或者说没有‘能量张力’，就不可能有情念范式”。不过，相比瓦尔堡较为侧重情念范式的表现形式，迪迪-于贝尔曼则更为关注其中的“极性的动力学”，即遗存的极性结构在历史时间中的重复方式，遗存时代的某个极性在后来既可以发展为张力的最大化，也可能被去极性化，或者其“消极的”价值可能变成“积极的”价值，例如异教遗存中总是以充满动感的、悲怆性的舞蹈姿态出现的酒神女信徒，其形式结构到15世纪的绘画和雕塑中变成了天使的形象；一件表现尼俄柏的孩子们被杀害的古代雕塑中出现的恐惧姿态，到安德烈亚·德尔·卡斯塔格诺（Andrea del Castagno）的著名雕塑《大卫和歌利亚的头》（Davide con la testa di Golia）中变成了胜利的英雄姿态（图4），遗存的形式结构的这种变形恰好体现了时间能量的悖论式运作，迪迪-于贝尔曼又称之为弗洛伊德意义上的“构型”或“症候构型”，甚至称弗洛伊德的本我与自我之间的冲突构成了瓦尔堡的极性的“元心理学境域”。

图4  （左）《尼俄柏的孩子》，约2世纪，大理石雕塑，意大利佛罗伦萨乌菲齐美术馆藏；（右）安德烈亚·德尔·卡斯塔格诺，《大卫和歌利亚的头》，约1450年，皮革，115厘米×77厘米，美国华盛顿国家美术馆藏。

瓦尔堡认为，沉静的情念范式总是体现为女性形象那富于动感的衣饰、微风吹拂的秀发、轻盈灵动的步态等，而其中最能体现或负载这一形象结构的对象就是宁芙（nymph）。在他看来，图像充满动感的附饰（accessories in motion）主要专注于“捕捉头发和衣饰的瞬间运动”，例如波提切利在《春》中对迎接维纳斯的时序女神、播撒鲜花的春之女神、在百花盛开的花园中舞蹈的美惠三女神这些古典宁芙的表现，其中重要的不是被再现的对象（头发、衣饰）本身，而是它们稍纵即逝的运动或动态，因为这种尚未固化的状态，这种对“时间性”的形式化，恰恰是古代及其生命能量的象征化。对于瓦尔堡而言，宁芙作为情念范式的高级表达形式，本质上是一种辩证的形式：外部与内部、骚动与静观、幻象般的美和倏忽而至的死亡，都在宁芙的形象及其装饰中互为表里。迪迪-于贝尔曼对宁芙的幻象特质没有太多兴趣，他更重视绘画中宁芙的灵动姿态，认为那是异教舞蹈的姿势语言充满张力的变体；他甚至认为瓦尔堡在博士论文中讨论的波提切利的《维纳斯诞生》和《春》两件作品中主要人物组群的姿势都是舞蹈；并强调宁芙成为舞蹈性的女性情念范式中“非个人化的女主角”，就是因为“她身上结合了相当多可能角色的化身”，或者说她本身就是悖论的扭结，比如她既是女神又是女人，既是人间的维纳斯又是天上的维纳斯，既是仆人又是胜利女神，既是舞者又是狄安娜，既是阉割者犹滴，又是女性天使。

毫无疑问，在遵循理论理性的人看来，迪迪-于贝尔曼对瓦尔堡的思想有过度阐释的嫌疑，这一印象主要来自他对症候及其多元决定的特质的坚执。但从迪迪-于贝尔曼的角度说，他这么做恰恰是为了激活瓦尔堡的思想动能，因为这种动能在“二战”后被瓦尔堡的某些继承者和阐释者压抑了；并且也是出于理论理性的要求。迪迪-于贝尔曼的激进化就是为了激活潜藏在瓦尔堡思想中的多元、开放的图像思维，这一点在他对瓦尔堡“图集”的讨论中体现得尤为明显。

四、直面间隙：图集/ 蒙太奇

1924年瓦尔堡从克罗伊茨林根医院出院后，直到1929年去世前，都一直埋头于一个名为“记忆女神图集”（Mnemosyne Atlas，以下简称“图集”）的疯狂计划，即利用各类媒介的图像，包括艺术史、古代地图、手稿页面的黑白摄影图片及摘自当代报刊的插图，在用黑布蒙起来的木板（大约150厘米×200厘米）上反复进行编排组合，从而形成一个用图像唤醒生命能量和文化记忆的庞大工作装置。“图集”（最后一个版本现存63块图板）自诞生以来，尤其自20世纪末瓦尔堡热兴起以来，一直备受关注，甚至现今许多研究者认为，瓦尔堡研究就是“图集”研究，围绕“图集”形成了各种解释。作为瓦尔堡热的积极推动者，迪迪-于贝尔曼对“图集”同样投入了极大的精力，他不仅组织策划“图集”展览，还依托“图集”至少出版了两部专著。

瓦尔堡认为，“记忆女神”指的就是图像的记忆功能，或者说生命能量借着图像在历史时间中的存在和返回，所以情念范式的遗存依然是瓦尔堡关注的重点，换用迪迪-于贝尔曼的理解，“记忆女神图集”就是“症候图集”，其标示了图像与时间的辩证关系。

然而，“图集”对艺术史时间模式的刻意疏离，也标示了瓦尔堡对待艺术作品的一种特别姿态：它是对图像的“摆置”，是在摆置中完成的对图像时间和图像空间的撕裂，以及对图像本身的撕裂；它是在摆置中对图像关系的一种重组，由此让隐藏在新的可能关系中的图像价值得以涌现。因此，瓦尔堡对图像的摆置不是要在艺术史的理性原则指导下去寻找所谓风格或源头的相似性、相邻性或差异性，而是要在非连续的时间和非邻近的空间中激发图像的张力，找到生命能量在错时的形式中隐秘的回响。如当瓦尔堡在“图集”的同一块图板上将古代遗存中战败者的痛苦同文艺复兴时期征服者的荣耀并置在一起的时候，他就是在“重述同一姿势范式的使用价值，仅仅为了打破这一命运的时间统一性：那一范式的遗存只能以彻底的中断为代价，在范式意义的‘动力学倒置’中，可以看到这种中断”。实际上，瓦尔堡的摆置就是要创造这种“中断”，但这并不是目的，恰恰相反，中断只是“图集”作为一种图像阐释学的开始，瓦尔堡自己称之为“间隙的图像学”（Iconology of intervals）。

在2011年出版的《图集，或焦虑的快乐科学》（Atlas, or The Anxious Gay Science）中，迪迪-于贝尔曼一开始就提出一个看似不是问题的问题：何谓“图集”（atlas）？然后给出了一系列蒙太奇式的描述：图集从来没有固定的形式，它的开头常常是任意的，结尾通常会把人带到一个有待探索的新知识地带；图集不是由通常意义上的“页面”组成的，而是由图像列表或图板组成的；图集是“知识的一种视觉形式”和“观看的一种博学形式”，前者涉及视觉形式的美学范式，后者涉及知识的认知范式，但图集是对两种范式的倾覆，或者说以倾覆的形式对两种范式的扭结，因为它打乱了所有这些可理解性的框架，引入了“一种根本的不纯粹性”。最后他总结说：“图集直截了当地打破了框架，打破了对自身之真理确信无疑的科学和对自身之准则确信无疑的艺术自诩的确定性，在它们之间创造出探索的各种间隙地带和富于启发性的间隙。”正是基于对“图集”的这一理解，至少对于迪迪-于贝尔曼而言，直面“图集”就是直面间隙，而“间隙”的视角或者说他所阐发的“间隙”技术与他的症候阅读、他的撕裂图像、他的时间幽灵学恰好可以内嵌在一起。

迪迪-于贝尔曼认为，“间隙贯穿于瓦尔堡的每一个对象。它是佛罗伦萨节庆的短暂间奏，或文艺复兴绘画中分隔两个叙事场景的泥灰画区域。它是负载有记忆的微风，吹开雕塑的长袍的每一个衣褶。它是将纯图形作品转化为宇宙之象征的能量。它是一种扭曲，仿佛从内部激起人与动物之间残酷的拥抱，如拉奥孔群像的绝望姿态或美洲原住民祭司的神秘舞蹈。它是亲和力网络，使得将一个天体星座和一幅解剖图结合在一个整体中在理论上得以可能”。这一似乎带有泛化嫌疑的论述并非迪迪-于贝尔曼的过度阐释，瓦尔堡自己的确也有过类似的表述，制造“间隙”的“图集”其实是该表述的一种喻象化实践。

迪迪-于贝尔曼还引入本雅明的“辩证图像”概念来说明瓦尔堡的“间隙”。这一理论嫁接之所以成为可能，主要并不在于本雅明和瓦尔堡是同时代人，而在于他们都生活在一个充满危机的时代，他们都对在现代性的技术情境下距离的消失怀有深深的矛盾情感，他们都试图在艺术文化中直面历史时间的面孔。某种意义上说，本雅明的“拱廊街计划”也是一个蒙太奇式的工作形式：流行戏剧表演的摘录、文学和诗歌文本、幽默和讽刺诗句，以及商业橱窗装饰和商品广告，汇聚在一起，构成了一幅现代性的“辩证图像”或星座喻象。本雅明有一段名言：

图像与现象学的“本质”的区别在于它们的历史索引（海德格尔徒劳地试图抽象地通过“历史性”来为现象学拯救历史）。对这些图像的思考应完全与“人文科学”的范畴，如所谓惯习、风格等，区分开来。因为图像的历史索引不仅仅表明它们属于特定的时间，而是首先表明它们只有在特定的时间才能获得可读性……每一个现在都是由与之同步的图像决定的：每一个“当下”都是具有特定可识别性的当下。在其中，真理因为时间被带到爆发点（这个爆发点不是别的，就是意图的死亡，因而与真实的历史时间、真理的时间的诞生相吻合）。并不是过去照亮了现在，也不是现在照亮了过去；毋宁说，图像是过去与现在瞬间汇聚在一起形成的一个星座。换言之，图像是停顿的辩证法。因为虽然现在与过去的关系是纯时间的，但曾经与当下的关系是辩证的：本质上不是时间的，而是图像的——只有辩证图像才是真正历史的亦即非古风的图像。

本雅明的“星座”，如同瓦尔堡的“图集”，就是异质时间的辩证配置，是异质性的时间在当下时刻的对质。“辩证图像”之为星座，就是因为那里潜伏了异质时间的“结”，那里有着时间的旋涡，那里的“间隙”隐藏有图像的“风暴”或“暴动”。而这也是瓦尔堡的“间隙”想要做的。

在此，重要的不在于如何界定“间隙”，而在于如何把“间隙”转换为一种方法、一种阐释图像和“图集”的技术。对此，迪迪-于贝尔曼在对瓦尔堡的图板的阅读中有充分的实践。例如43号图板（图5），展示的是多米尼科·基朗达约（Domenico Ghirlandaio）在佛罗伦萨圣三一教堂为银行家萨塞蒂（Sassetti）装饰的私人礼拜堂。图板右上角是瓦尔堡的妻子为礼拜堂三面被装饰的内墙绘制的三幅素描稿，其他几组摄影图片则是一系列风格和图像史的比较：基朗达约与乔托之间（不同时代基于同一主题的图像史）；基朗达约与他的兄弟之间（同时代不同的北方风格）；基朗达约与波提切利之间（同时代不同的古典风格）。这一系列比较展示了礼拜堂装饰本身是记忆的艺术，是在墙壁上“展开的巨大相册”，是一个“图像志和时间蒙太奇的空间”。图板上还有从礼拜堂祭坛墙上的一幅壁画《教皇核准方济各会会规》截取的局部图像：一组是美第奇家族的几个孩子和人文主义者的头像；另一组是位于祭坛画后方、绘制在祭坛墙底部的萨塞蒂夫妇的肖像。从瓦尔堡的图像选择和他对萨塞蒂礼拜堂的专题研究来看，他在此要强调的是文艺复兴时期佛罗伦萨中产阶级的个人主义热情与宗教希冀之间、死亡恐惧与恐惧制服或复活渴望之间的情念张力。迪迪-于贝尔曼认为这种张力不是通过单幅图像的主题来表达的，而是通过图像之间的“间隙”——时间和空间的“间隙”、场景设置的“间隙”、形式遗存之间的“间隙”——制造出来的，比如委托人夫妇的肖像与他们的石棺之间，真实的石棺与祭坛画中的石棺之间，复活的孩童与从“地下”冒出来的儿童之间，神圣的仪式场景与错时插入的现实场景和人物之间，等等。

图5  “记忆女神图集”43号图板。

肖像的“现在”预期了赞助人必死的“未来”：这是一个墓地礼拜堂，萨塞蒂夫妇的肖像紧邻各自的石棺。方济各传说或基督故事的“过去”是复活的“未来”的范本：圣子在祭坛桌上诞生，一个古老的石棺支撑着他的头；相邻画面描绘的场景中，一个死去的孩子复活了，这是萨塞蒂家族戏剧的暗示，而在最高一层的礼拜堂装饰中，一群孩子从地下冒出来……就这样，基朗达约在壁画中汇集了神圣与世俗、私人与公共、遥远空间（伯利恒）与当地空间（佛罗伦萨）、基督的生平与方济各的生平（模仿前者）、北方写实风格与南方古典化风格、中世纪价值与文艺复兴价值、人文主义的智性与资本家的“唯物主义”、各种类型的诞生与死亡等，他将它们全都整合在一个宏大的、被古代异教遗存所萦绕的基督教形象体系中。

“图集”作为对图像的摆置，就是制造具有张力的“间隙”，其最基本的操作就是对图像或图像体系的拆解和重新组装，迪迪-于贝尔曼借由从电影技术获得的启示，称之为“蒙太奇”。这一观点并不为迪迪-于贝尔曼所独有，法国另一位著名的瓦尔堡研究者菲利普-阿兰·米肖（Philippe-Alain Michaud）曾经说：“‘记忆女神图集’尽管与电影技术本身没有任何关系，但仍是一种电影式的编排。”迪迪-于贝尔曼描述说，瓦尔堡喜欢把图片挂在黑布为背景的展示板上，形成图板进行拍照，拍照后拆解掉既有图板，开始下一板，完毕后再破坏掉重新开始。这一近乎疯狂的、西西弗式的劳作就像是瓦尔堡对疯癫最后的留存和救赎。瓦尔堡对图像或图片看似混乱实则高度谨慎的调度，就是为了赋予组装或蒙太奇一种辩证的知识生产能力。

为什么蒙太奇式的拆解和组合可以生产出充满能量的辩证知识呢？因为蒙太奇式的操作首先就是在摆脱或扰乱图像原有的空间语境和叙事语境，“改变显现的空间、展示的空间，或编排被看之物的空间”，这样其原先的可见性状态，以及原先的类比关系或象征关系也会随之改变；再为其准备一张“桌子”或工作台，将其置于不同的表征系统或关系网络中，使其面对一个断裂的“间隙”，“来容纳这目光和意义的转变，来收集有待被看到的喻象多重性的包裹”。就像43号图板，被截取和摆置在那里的图片/图像仅仅是蒙太奇式的素材，所形成的只是一个互文的语境。要对它们进行蒙太奇式的组合或重组，则仰赖“间隙”的加入和“间隙”对图像的撕裂和重新照亮，比如委托人夫妇的石棺与委托人夫妇的肖像似乎构成了某种指涉关系，但如果将它们与祭坛画上圣婴身后的古典石棺相关联，如果将它们的异教浮雕装饰同图绘中石棺上的异教铭文相关联，古代遗存及其情念范式就在这一错时的返回中，在这一时间和事件的“间隙”中构成了图像的症候。所以，迪迪-于贝尔曼说：“从异质的时间持续在里面一起工作的意义上说，图集是一种错时的物件。”或者说，图集作为一种工作方式，就是要以蒙太奇式的拆解和重组来激活图像内部及图像之间异质的时间性。

结语

迪迪-于贝尔曼的症候图像学并非学院意义上的图像学体系，他所寻求的根本不是逻各斯中心主义的“学/知识”，恰恰相反，他的求真意志总是受到一种“非知”的开放性驱使，总是在不断的破坏和重建中重申图像及其意义的不稳定性。因此，我们不能企望有一个普遍有效的症候图像学来帮助处理所有的图像问题。说得更明确一点，迪迪-于贝尔曼的症候图像学关注的是图像中的“非知”，诸如与主题无关的某个失序的细节、图像中的断裂和“错误”、扰乱图像秩序的“非法闯入者”，以及如同“物自体”一般中断图像可读性的物质性等。它们构成图像的症候，不是因为艺术家的病态或失败，而是因为艺术本身的原始冲动：与生活对质、与知识对质、与艺术本身对质。因此，不论是直面图像、直面时间还是直面间隙，对于迪迪-于贝尔曼而言，都是在直面症候：直面图像的症候、直面时间的症候、直面我们自身生命的症候。 

 

The Boundless Deep by Richard Holmes review – wild times with young Tennyson

A masterful account of the poet’s early life during the tumultuous 19th century crisis of faith

Lucy Hughes-Hallett

Mon 6 Oct 2025 09.38 CEST

Alfred Tennyson was a divided soul. He even wrote a poem called The Two Voices in which dual versions of himself argued out the pros and cons of suicide. In this illuminating book, Richard Holmes has chosen to focus on the lesser known of the poet’s personae.

The year 1850 was pivotal for Tennyson. He published the great poem sequence In Memoriam, over which he had laboured for nearly two decades. He became, as a result, both famous and rich. He got married, after a 14‑year courtship. He had been living in rented homes with his mother and siblings, or dossing down with bachelor friends in London, or lurking alone in a ramshackle cottage on one of his native Lincolnshire’s bleak beaches. Now he took a house where he could receive distinguished visitors. (When Prince Albert came calling, Tennyson was so far from obsequious that he forgot to invite the queen’s consort to sit down, though he did at least offer the poor man a drink.) He was appointed poet laureate. His life as a Great Man began.

For Holmes, this settling down – emotional, financial, social and poetical – is the end of the exciting part of Tennyson’s life. The Boundless Deep is not about the bestselling author of the nation’s favourite poem – The Charge of the Light Brigade – or the laureate lord, pacing along the clifftops of the Isle of Wight or posing for Julia Margaret Cameron’s famous photograph (known derisorily in his family as “the dirty monk”). Instead, Holmes gives us “young Tennyson”, the wildly talented youth whose story is far more rich and strange than that of the bearded celebrity.

From his teens he was imposing, even glamorous. He was very tall, unkempt but handsome

The Tennysons, wrote Alfred, were a “black-blooded race”, meaning prone to moods and melancholy. His father, a reluctant clergyman, was angry and very often drunk. There was an incident, the details of which are obscure, that resulted in the family cook being burned to death in the rectory kitchen. One of Alfred’s brothers was confined to a lunatic asylum as a boy and stayed there for life. Another suffered from profound depression and followed his father into alcoholism. A third became addicted to opium. Alfred himself suffered from episodes of paralysing gloom and what he called “weird seizures”. His Maud is narrated by a madman: he must often have wondered whether he was or would become one himself.

From his teens he was imposing, even glamorous. He was very tall, unkempt but handsome. Even before he began to wear a black Spanish cloak and sombrero, he could dominate a room. But, having grown up hugger-mugger with his siblings – three brothers to an attic room – as an adult he sought out solitude, retreating into silence when in company, vanishing for solitary walking tours.

Holmes, master biographer that he is, vividly conjures up this awkward, compelling figure. What gives his book its exceptional energy, though, is not what is happening on the surface of Tennyson’s life and Holmes’s narrative. It is the powerful undertow of threatened belief and existential anxiety tugging the reader down into the “boundless deep” of the title, where 19th-century thinkers wrestled with terrible thoughts. In 2008, Holmes published The Age of Wonder, a group biography charting the way that Romantic poets responded to the discoveries of their scientific contemporaries such as Humphry Davy, William Herschel and Joseph Banks. That book was full of exuberant hopes. This one, set a generation later, is its darker, sadder sequel.

In Tennyson’s lifetime, geologists, astronomers and those “natural philosophers” who were beginning to think with Darwin about the origin of species, were raising appalling questions. If the history of life on Earth had begun millions of years before the appearance of the human race, then how to believe that the world had been made for humanity’s benefit and enjoyment? “It is inconceivable,” wrote Tennyson, “that the whole Universe was merely created for us, who live on a third-rate planet of a third-rate sun.” The new telescopes and microscopes revealed spaces infinitely large and beings infinitesimally small: how to hold to one’s faith, given such evidence, in a God who had made man in his own image? If dinosaurs had become extinct, then might the human race do so too?

These conundrums were repressed. It was hard to confront them in public without being accused of blasphemy, or in private without seeming to give way to despair. It is easy, in retrospect, to make fun of those 19th-century creationists who strove to believe that God had hidden the fossils in the rocks for some good, if inscrutable, reason. Holmes doesn’t indulge in such patronising levity. He takes seriously the pain of those living through the crisis of faith, and writes about it with such sympathy that even his most secular-minded modern readers feel the shock of finding oneself alone and unloved in a God-forsaken universe.

Holmes binds his narrative together with two recurrent motifs. The first he introduces on his second page – it is the image of the Kraken, the legendary sea monster. Tennyson was a 20-year-old undergraduate when he wrote his poem about it. In Holmes’s view, with its mix of “Norse mythology, 18th-century zoology, 19th-century science fiction and the Book of Revelations”, the 15-line sonnet introduces themes to which Tennyson would keep returning. Its sense of something immense, unspeakable and tragic, submerged beyond reach of human inquiry, prefigures the mood of In Memoriam. It marks Tennyson’s debut as a master of metre and as the creator of images in which awful mystery is compressed into a few dazzlingly suggestive words.

The second theme is the Kraken’s opposite. Where the imaginary sea monster epitomises all that is lugubrious about Tennyson, his relationship with a real-life person, Edward FitzGerald, of whom he would write “I had no truer friend”, summons up all that is affectionate and playful in the poet. With him, Holmes introduces us to a side of Tennyson seldom before encountered. A Tennyson who, after intoning some of his grandest lines with “grotesque grimness”, would suddenly roar with laughter at his own solemnity. A Tennyson who, after visiting “dear old Fitz” at home, wrote a thank-you letter in verse describing him in his rose garden with his tame doves sitting all over him, planting their “rosy feet … on shoulder, hand and knee”, and even on his head. It’s an image of pleasure nicely adapted to FitzGerald’s great celebration of hedonism – his version of The Rubáiyát of Omar Khayyám. It also summons up the brilliant nonsense of the two poets’ mutual friend Edward Lear. It’s pleasing to be told that Tennyson, the mournful Great Man, was also the inspiration for Lear’s rhyme about the old man with a beard in which “two owls and a hen, four larks and a wren” built their nests.

Holmes is not an uncritical fan – he is tart about the pseudo-medieval flim-flam of Idylls of the King – but when he loves a poem, he writes about it with a wonderful capacity for noticing every pulse of metre, every flicker of nuance. This biography is a compelling story of an odd, brilliant, charismatic character, and a reappraisal of a man who had become so very established we could no longer see him. In giving us young Tennyson – gauche, beautiful, veiled in stinking tobacco smoke – Holmes has given the great poet back to us.

 The Boundless Deep: Young Tennyson, Science and the Crisis of Belief by Richard Holmes is published by HarperCollins (£25). To support the Guardian, order your copy at guardianbookshop.com. Delivery charges may apply. 

The Kafka Challenge: Translating the Inimitable

 

The Kafka Challenge

Translating the Inimitable

Paul Reitter

Paul Reitter is professor of German Languages and Literatures and former director of the Humanities Institute at Ohio State University and the author of The Anti-Journalist: Karl Kraus and Jewish Self-Fashioning in Fin-de-Siècle Europe and many other books. The most recent of his translations is Karl Marx’s Capital.

When I taught German in graduate school back in the late 1990s, my fellow instructors and I often used a line from Franz Kafka’s novel The Trial to illustrate a point about grammar that was also a point about untranslatability.¹ In German, as in English, the regular subjunctive form goes mainly with wishes, counterfactual conditions, statements, and questions, as well as with polite requests. But the German form has an additional function: It can mark speculation—or, really, ambiguity—in a way that’s hard to match in English. Kafka’s line evokes a vivid sense of this gap, which, in the first place, is why we turned to it here. However, we had further reasons for doing that, starting with the fact that untranslatability is one of Kafka’s great themes. 

Untranslatability is also one of George Steiner’s great themes—and one of his central concerns in his commentary on Kafka. It would be hard to think of a literary scholar or critic who has done more to draw attention to this aspect of Kafka’s work, to reveal it as a guiding principle. In his essay “K,” for instance, Steiner cites, at length, a previously underexamined diary entry in which Kafka discusses how for him the German words Mutter and Vater fail “to approximate to” Jewish mothers and fathers. Kafka suggests that his psychic life was shaped by this linguistic misalignment; as a result of it, he “did not always love” his mother as “she deserved” to be loved and as he was capable of loving her. Steiner goes on to read “The Burrow,” one of Kafka’s last stories, as “a parable” of “the artist unhoused in his language,” a point he makes to explain nothing less than “the fantastic nakedness and economy” of Kafka’s prose.²

In After Babel (1975), which appeared more than a decade after “K,” Steiner went further still, deepening his engagement with how “the theme of Babel haunted” Kafka, who felt himself to be caught between the incompatible impossibilities of “writing in German” and “writing differently.” He traces the theme through a number of Kafka’s works, including “The Great Wall of China” and “The Burrow,” that enigmatic late work on the way in which the apt metaphor for verbal communication is less a building of bridges to the world than the creation of structures that seal us off from it.³ “In Kafka,” as Steiner puts it, “speech is the paradoxical circumstance of man’s incomprehension.”⁴ Borrowing a phrase from the historian Gershom Scholem, he elsewhere calls Kafka a “borderline case of wisdom, representing, as no other writer has, ‘the crisis of the sheer transmissibility of truth.’”⁵

Josef K. in the Dock

Against this background, in the late 1990s the prospect of new English translations of Kafka’s novels The Trial and The Castle, the first done by scholars—Breon Mitchell and Mark Harman, respectively, and the first to be based on unexpurgated manuscripts—elicited excitement but also skepticism. There was no shortage of the latter attitude among graduate students of German, and I remember feeling both disappointed and vindicated when Mitchell’s rendering of The Trial appeared in 1998 and I saw that he hadn’t fared much better than Willa and Edwin Muir with the line we enlisted to stress the difficulty of the subjunctive. Our suggestions about its untranslatability had held up. 

If anything, Mitchell’s translation appeared to be less effective than the Muirs’ original English version (1937), which made the line seem all the more well-suited for the role we assigned it. The first sentence of The Trial reads, in German, “Jemand mußte Josef K. verleumdet haben, denn ohne daß er etwas Böses getan hätte, wurde er eines Morgens verhaftet.” (“Someone must have slandered Josef K., for without having done anything wrong, he was arrested one morning.”⁶) The subjunctive form is “hätte,” which, when paired with a past participle (e.g., “getan”), would most often be translated as “would have” or “had,” as in “if he had done something wrong, we would have found out,” or “if only he had done something wrong.” Neither a condition nor a wish nor a request in the case at hand, it signals, unobtrusively but importantly, that we don’t have here an unambiguous statement of fact. Josef K. may have “done something wrong” (“etwas Böses getan”), or maybe not. Maybe the narrator knows but is not saying whether it’s one way or the other, and the more decisive first and last parts of the sentence—“Someone must have slandered Josef K.,…he was arrested one morning” (“Jemand mußte Josef K. verleumdet haben…wurde er eines Morgens verhaftet”)—don’t ultimately make it all clear. 

This element of uncertainty is not preserved in the Muirs’ rendering. Since it turns on a use of the subjunctive that, you could say, lies outside the cognitive field of the English counterpart feature, you can’t keep it without resorting to paraphrase—“even though he may not or would not have done anything wrong”—which still points too much in the direction of guiltlessness. In the Muirs’ translation, Josef K. is thus introduced as someone who was arrested one morning “without having done anything wrong.” He’s presented as a victim, in other words, and Anglophone readers have tended to see the book as being about an innocent man who is arrested and punished by a bizarrely opaque and capricious legal system. This, for example, is how the novel has been portrayed in US legal reasoning, where it has figured more prominently than you might think. According to one estimate, “Since the mid-1970s…Kafka’s name has appeared in more than 400 opinions written by American state and federal judges.”⁷

Mitchell tried to finesse the situation, following the Muirs into the indicative mood but inserting the word “truly” into the sentence, so that Josef K. is arrested “without having done anything truly wrong.”⁸ He’s no longer simply innocent: Mitchell’s translation tells us that Josef K. may have transgressed, just not in a substantial or serious way. In essence, then, Josepf K. remains a victim, only now we have to contend with a distinction that plays no role in the German text: something wrong versus something truly wrong.

So if there is a politics to this particular instance of untranslatability—one that has to do with things postwar critics haunted by totalitarianism prized in Kafka, among them Steiner and Theodor Adorno, things such as developing in readers a capacity to live with doubt, or, perhaps, to understand that interpretation involves overreach and thus guilt—then it is a fragile politics. Necessarily, the political content changes in English, and the consequences for Kafka’s Anglophone reception have been significant. 

The Kafkaesque Untranslatables

In the years after Mitchell’s translation was published, untranslatability won greater prominence as a topic in literary studies, thanks in large part to the work of the scholar Barbara Cassin, whose 1,344-page Dictionary of Untranslatablesappeared in French in 2004 and as an expanded English edition ten years later.⁹ Like Steiner, and Kafka, too, Cassin stopped short of making untranslatability “a criterion of truth.” But her project, carried out with scores of collaborators, certainly thickened the association between untranslatability and semantic depth and heft, since the list of terms included—they are all terms with a place in the history of philosophy—reads like a who’s who of philosophical key terms: e.g., “Geist” and “Dasein.” During this time, as the English retranslations of Kafka kept coming, I continued to think about the translation challenges his works pose, or, as Cassin says with reference to philosophical untranslatables, about how the translation of his works can “create a problem.” It seemed, and still seems, like a productive way to reflect on his writing and the sources of its special appeal. 

Consider what happens when we compare two recent translations of The Metamorphosis—Susan Bernofsky’s, from 2014, and Mark Harman’s, from 2024. The novella begins with Gregor Samsa waking up to find that he has been transformed into a “monstrous insect.” This turn of events clearly comes as a surprise, one that perplexes Gregor, prompting him to ask, “What has happened to me?” But he never follows up on his question, and it’s another unexpected occurrence, which he notices a little later, that leaves Gregor truly baffled and looking for answers: He has overslept. He wants to know how he could have done that, as well as what the best way to respond might be. Here, in Harman’s translation, are a few sentences from this part of the story:

Shouldn’t the alarm clock have rung? He could see from his bed that it was correctly set for four o’clock; surely it must have rung. Yes, but was it possible to sleep through this furniture-rattling alarm? Well, he had certainly not slept quietly, but probably all the more deeply for that. But now what should he do? The next train left at seven o’clock; to catch up with it he would have to hurry like mad, the cloth samples were not yet packed, and he certainly didn’t feel especially fresh and active. And even if he caught up with the train, it would be impossible to avoid a tongue-lashing from the boss, for the office assistant would have been waiting beside the five o’clock train and would have reported his negligence long ago.¹⁰

And here are the same sentences in Bernofsky’s translation:

Could the alarm have failed to ring? Even from the bed one could see it was properly set for four o’clock; it must have rung. Yes, but was it possible to sleep tranquilly through this furniture-shaking racket? Well, his sleep hadn’t been exactly tranquil, but no doubt that’s why it had been so sound. But what should he do now? The next train was at seven o’clock; to catch it, he would have to rush like a madman, and his sample case wasn’t even packed yet, and he himself felt far from agile or alert. And even if he managed to catch this train, his boss was certain to unleash a thunderstorm of invective upon his head, for the clerk who met the five o’clock train had no doubt long since reported Gregor’s absence.¹¹

Right from the start—“Could the alarm have failed to ring?” versus “Shouldn’t the alarm clock have rung?”—we see that Bernofsky opts for a more formal register. Since we are sliding into Gregor’s perspective, and thus getting the phrasing in his head, this makes for a little more wackiness—the monstrous insect with the genteel formulations: “Even from the bed one could see,” “tranquilly,” “rush like a madman,” “thunderstorm of invective upon his head.” Not only is the diction in Harman’s rendering less elevated, but the prose tends toward greater compression—“he could see,” “hurry like mad,” “tongue-lashing,” “would have reported his negligence long ago” versus “had no doubt long since reported Gregor’s absence.” Similarly, Harman’s syntax is a little less complex, and as a result of these differences, the sentences in his translation move forward faster. 

In the German, the repetition of words is more conspicuous than it is in both translations. Kafka uses the same word for “ring/noise” (“läuten”) in three consecutive sentences. Both translators reduce that to two occurrences and the two occurrences of “ja” (“yes”) to one. The word “selbst” is employed two different ways (“himself” and “even”) in consecutive sentences, which makes the repetition difficult to preserve. But, unlike Harman, Bernofsky compensates for that loss by repeating a phrase (“no doubt”) where there isn’t repetition in the source text. Furthermore, she keeps the doubling of the term “ruhig” (“tranquil” in her translation), whereas Harman doesn’t, and in contrast to him she retains all three instances of “and” (“und”) in the last two sentences, thereby reproducing more of the dynamic of prose cycling around. 

The differences with respect to forward thrust and repetition bring us to the key translation challenge: Kafka’s sentences manage to be both recursive and propulsive. Gregor’s thoughts wind around and around, and yet even many of the longer sentences drive forward, since as he repeats words, Kafka takes advantage of resources for compression that we lack in English (e.g., gendered nouns and heavy case inflection make it possible to lean on pronouns without risking confusion). Because the recursive element plays the greater role in producing the atmosphere readers have come to think of as “Kafkaesque,” which is characterized in part by recursive reflection going on where you don’t expect it to (the former traveling salesman wondering in detail about how he could have overslept and ignoring the matter of how he turned into an insect), translators have made conveying it a priority. 

Harman’s bold move in his translation of The Metamorphosis, which he calls The Transformation, is to back away from this priority—not a lot, but perceptibly—and to allow the propulsive character of Kafka’s prose to be brought into English more fully. As a longtime reader of that prose, I had certainly experienced the tension in its sentences between two contrasting, even opposing, kinds of movement. But the tension didn’t come into relief for me as something I was consciously aware of until I thought about how Harman’s translation differs from earlier ones and about the reprioritization—whether intentional or not—his project entails. 

When we compare the versions of The Metamorphosis by Harman and Bernofsky, we encounter another pair of dichotomous tendencies, which is perhaps even more striking. Neither version manages to preserve the full effect of Kafka’s sentences—the translation of his work does indeed “create a problem”—and yet both renderings are excellent: careful, thoughtful, and polished. They give us a Kafka who is recognizably himself. In fact, for an author so much associated with radical singularity, linguistic self-alienation, and Talmudic indeterminacy, or, in a word, untranslatability, Kafka has proven to be remarkably translatable.¹²

Parables Without a Key

The Muirs’ translations are still used, appreciated for what Steiner himself described, in speaking of their version of The Trial, as a certain “freshness of encounter,” by which he may have meant they preserved the antic mood one finds so often in the social world of Kafka’s fiction. But there are now many other successful Anglophone efforts: Harman’s translations of The Castle (1998) and Amerika (2008), Idris Parry’s translation of The Trial (2000), J.A. Underwood’s translation of The Castle (1997), Michael Hofmann’s translations of Amerika (2002) and The Metamorphosis (2007), Anthea Bell’s translation of The Castle (2009), Shelley Frisch’s translation of Kafka’s aphorisms (2022), and Ross Benjamin’s translation of his diaries (2023), to name a few beyond the ones already mentioned. 

Part of what has brought about this situation, I think, is simply that Kafka, an uncompromising innovator with high-modernist aura and a broad readership, has attracted very talented, ambitious translators. Another part is that he maintained diverse allegiances. In this he bears an affinity with Steiner, who was devoted to modernist “Gnosticism,” to use his (Steiner’s) term, and who celebrated Walter Benjamin as a “theologian of language” with an unrivaled eye for the paradoxes of Kafka’s mysticism, yet whose own prose differs so starkly from Benjamin’s in terms of directness and limpidity. There is a side of Kafka that not only draws on the realist tradition of Dickens, Flaubert, Balzac, and Tolstoy, all writers whom he studied attentively, but also tries to streamline it. 

Compare the first lines of the 1912 short story “The Judgment,” which Kafka regarded as his best work, with the opening of Thomas Mann’s novella Death in Venice, which appeared the very same year and lifted Mann out of an extended writing slump. (The two stories have quite a bit in common, by the way, including the theme of deindividuation and the respective watery demises of their protagonists.) With his self-narrative breaking down more and more—he can hardly remember anything by the end of the story—Kafka’s Georg Bendemann hastens to carry out his father’s dramatic judgment: He hurls himself into a river, vanishing without a trace. (Mann’s Gustav von Aschenbach, exhausted by his lifelong struggle to balance the Dionysian and Apollonian tendencies within him, finds himself drawn irresistibly to the formless “liquid element.”¹³) Kafka gives us a series of lean, descriptive sentences with straightforward syntax: They begin with a subject followed immediately or nearly immediately by a verb and contain only a couple of subordinate clauses. Mann’s opening sentences are so full of extended modifiers and internal clauses that an acclaimed recent Anglophone translation simply drops one of those clauses for the sake of getting the sentences into literary English. In contrast to Mann’s fiction, moreover, Kafka’s largely avoids local references and also dialects, two things that can bedevil translators. Whereas Mann cultivated a musical style, at times echoing the rhythms of Wagner’s compositions, Kafka strove, as Mark Anderson has put it, to make his prose “non-musical,” even boasting of his “unmusical” nature in letters to his Czech translator Milena Jesenská.

But in at least one respect, the Kafkaesque itself, or the atmosphere that counts as that, lends itself to translation. It does so because this atmosphere depends, to a large extent, on structural conceits. That this is so should already be clear, in fact. Recall the passage where Gregor realizes he has overslept. Whatever an individual translator’s priorities may be, the basic incongruousness of Gregor’s response to his situation will likely come across in the translation because it is part of the structure of the scene: Again, Gregor puzzles over how he managed to oversleep rather than over the forces that metamorphosed him into a giant insect. Furthermore, the same goes for the strangeness of such an insect engaging not only in thought but in intricate, lawyerly, recursive thought, in which the doubts pile up—and even if x problem were solvable, what about y problem?—yet the most obvious problem is omitted. How can Gregor hope to pack up his samples, catch his train, and carry out his job, given that he no longer has a human body? 

Let’s consider the Muirs’ translation of the passage:

Had the alarm clock not gone off? From the bed one could see that it had been properly set for four o’clock; of course it must have gone off. Yes, but was it possible to sleep quietly through that ear-splitting noise? Well, he had not slept quietly, yet apparently all the more soundly for that. But what was he to do now? The next train went at seven o’clock; to catch that he would need to hurry like mad and his samples weren’t even packed up, and he himself wasn’t feeling particularly fresh and active. And even if he did catch the train he wouldn’t avoid a row with the chief, since the firm’s porter would have been waiting for the five o’clock train and would have long since reported his failure to turn up.¹⁴

As with the first sentence of The Trial, the Muirs are prepared to let subjunctive forms go here—for example, note their “Had the alarm clock not gone off?” as compared with Harman’s “Shouldn’t the alarm clock have rung?” and Bernofsky’s “Could the alarm have failed to ring?” The loss of the subjunctive doesn’t substantially alter the meaning in this case by eliding a crucial ambiguity, but neither is it an unimportant change. Holding a little closer to the German, as Harman and Bernofsky do with the slightly non-colloquial “Shouldn’t the alarm clock have…” and “Could the alarm have failed…,” and as Bernofsky does with “thunderstorm of invective” for Kafka’s “Donnerwetter,” whose non-metaphorical meaning is “stormy weather,” creates an air of linguistic scruple that matches Kafka’s. The Muirs’ “row” doesn’t begin to reproduce the vividness of Kafka’s expression and suggests “quarrel” when Gregor is unlikely to stick up for himself—hence Harman’s “tongue-lashing.” Such translation differences have cumulative effects, making for profoundly different reading experiences, and yet the point still stands: The Kafkaesque is palpable in all three translations. The structural conceits are of course present in all three, and if, as is often the case with first translations of works that become classics, the Muirs’ translation of the passage feels almost a little slapdash compared with the carefully weighed out later versions, it is at least as attuned to other basic elements of the Kafkaesque, such as repetition. 

Another of Kafka’s great themes is paradox—“plenty of hope, just not for us,” “what do I have in common with the Jews? I hardly have anything in common with myself.” And then there are the paradoxes of his work—parables without “a key,” to speak with Adorno, fiction so tied to a Czech-Jewish-German milieu and yet able to resonate with the most diverse audiences, to the point that an early biographer saw fit to describe Kafka as “representative man.” We can add this paradox to the list: eminently translatable, untranslatable.  

 

The Software Essays that Shaped Me

by Michael Lynch, published September 30, 2025

I started reading software blogs before I got my first programming job 20 years ago. At this point, I’ve read thousands of blog posts and essays about software, but only a small handful stuck in my mind and changed the way I think.

1. “The Joel Test: 12 Steps to Better Code” by Joel Spolsky (2000)
2. “Parse, don’t validate” by Alexis King (2019)
3. “No Silver Bullet - Essence and Accident in Software Engineering” by Fred Brooks (1986)
4. “Choices” by Joel Spolsky (2000)
5. “Application compatibility layers are there for the customer, not for the program” by Raymond Chen (2010)
6. “Don’t Put Logic in Tests” by Erik Kuefler (2014)
7. “A little bit of plain Javascript can do a lot” by Julia Evans (2020)
8. “Choose Boring Technology” by Dan McKinley (2015)
9. “I’ve locked myself out of my digital life” by Terence Eden (2022)
10. Bonus: Brad Fitzpatrick on parsing user input (2009)

“The Joel Test: 12 Steps to Better Code” by Joel Spolsky (2000)🔗

Joel Spolsky is the greatest software blogger of all time. His essays have informed so much of my approach to software that it was hard to pick out just one, but “The Joel Test” is my favorite.

The Joel Test is a set of 12 questions that employers can ask themselves to see how well they’re investing in their software team:

1. Do you use source control?
2. Can you make a build in one step?
3. Do you make daily builds?
4. Do you have a bug database?
5. Do you fix bugs before writing new code?
6. Do you have an up-to-date schedule?
7. Do you have a spec?
8. Do programmers have quiet working conditions?
9. Do you use the best tools money can buy?
10. Do you have testers?
11. Do new candidates write code during their interview?
12. Do you do hallway usability testing?

Some of the questions are dated, but the point was never the questions themselves but rather the meta-point of the questions.

Joel was really asking employers: do you respect developers?

The questions all assess whether an employer prioritizes their developers’ time and focus over things like cheap office space and short-term deadlines.

Joel published this article at the height of the dot-com boom, when skilled developers were a precious resource, but not everyone realized it, including developers themselves.

Joel’s blog always presented programmers as rare, delicate geniuses that employers needed to pursue and pamper. I liked that.

Throughout my career, I sought out employers that scored well on the Joel test, and I’m grateful to Joel for giving me the map to find them.

“Parse, don’t validate” by Alexis King (2019)🔗

This essay is about leveraging the type system in Haskell to — wait, wait! Don’t go to sleep.

If you don’t care about type systems or Haskell, I get it. I don’t either. But this essay radically changed the way I think about software. You can use Alexis’ technique outside of Haskell in any language that supports static types (e.g., Go, C++, Rust).

The highly abridged version of the essay is that whenever you validate any data, you should convert it to a new type.

Suppose that your app has a rule limiting usernames to a maximum of 20 alphanumeric characters. The naïve solution would be to define a function that looks like this:

func validateUsername(username string) error { ... }

With the above function, you run validateUsername anytime you receive a username from a user.

The problem with this approach is that your code is unsafe by default. You have to remember to validate every username you receive, so it’s easy to create a code path that accidentally processes a username without validating it. If a nefarious user notices the mistake, they can do tricky things like embed malicious code in the username field or stuff it with a billion characters to exhaust server resources.

Alexis’ solution is to instead use a function like this:

func parseUsername(raw string) (Username, error) { ... }

In the rest of your codebase, instead of passing around a string called “username,” you use a custom type: Username. The only function that can create a Username is parseUsername, and it applies validation rules before returning a Username instance.

Therefore, if you have a Username instance, it must contain a valid username. Otherwise, it couldn’t exist.

You can’t forget to validate a username because untrusted input will always be a string, and you can’t pass a string to a function that expects a Username.

Before Alexis’ essay, I thought type systems were just a fun way to distract language nerds. “Parse, don’t validate” opened my eyes to how valuable compiler features can be in improving an application’s security and reliability.

“No Silver Bullet - Essence and Accident in Software Engineering” by Fred Brooks (1986)🔗

In college, I read The Mythical Man-Month, a collection of essays about software engineering by Fred Brooks, drawing on his experience directing IBM’s OS/360 project.

The essays were hit or miss. Some felt too old to be relevant, even in 2002, but the one that stuck with me was, “No Silver Bullet.”

The essay argues that you can divide software work into two categories: essential complexity and accidental complexity.

Essential complexity is the work that you absolutely have to do, regardless of your tooling and hardware. For example, if you write software that calculates bonuses for salespeople, you have to define formulas for those bonuses and cover all possible edge cases. This work is the same if you have a $5B supercomputer or a $1 microcontroller.

Accidental complexity is everything else: dealing with memory leaks, waiting for your code to compile, figuring out how to use a third-party library. The better your tooling and hardware resources, the less time you spend on accidental complexity.

Given this model, Brooks concluded that it was impossible for any advancement in tooling or hardware to create a 10x improvement in developer productivity:

How much of what software engineers now do is still devoted to the accidental, as opposed to the essential? Unless it is more than 9/10 of all effort, shrinking all the accidental activities to zero time will not give an order of magnitude improvement.

Throughout my career, people have been trying to find ways to eliminate programmers from software. For a few years, no-code platforms generated buzz by promising non-programmers all the powers of a seasoned web developer.

Brooks’ essay always reassured me that the latest buzzword platforms could never replace developers, as the platforms focused on the accidental, not the essential. Even if the platforms could magically create working code from a functional specification, you still need someone to write the spec:

I believe the hard part of building software to be the specification, design, and testing of this conceptual construct, not the labor of representing it and testing the fidelity of the representation.

Modern AI has thrown a wrench into Brooks’ theory, as it actually does reduce essential complexity. You can hand AI an incomplete or contradictory specification, and the AI will fill in the gaps by cribbing from similar specifications.

Even if AI eliminates programming as we know it, Brooks’ essay gives me hope that we’ll still need people to manage essential complexity at whatever level of abstraction that ends up being.

“Choices” by Joel Spolsky (2000)🔗

I said above that it was hard to pick a single favorite Joel Spolsky essay, which is why I’ve chosen two.

“Choices” is about creating user interfaces and the subtle costs of giving a user power:

Every time you provide an option, you’re asking the user to make a decision. That means they will have to think about something and decide about it. It’s not necessarily a bad thing, but, in general, you should always try to minimize the number of decisions that people have to make.

As an example, Joel shares a ridiculous dialog that appears in Windows 98 when you try to search the help documentation:

The dialog infuriates Joel because it interrupts the user while they’re trying to get help, and it asks them to make an uninformed decision about database optimization. Windows was shirking a decision and pushing it onto the user.

Joel’s essay focuses on graphical user interfaces, but I think about it wherever people might encounter my code, including on the command-line or other developers calling functions I wrote. Can I make a useful decision on my user’s behalf while still giving them power over things they care about? There are countless times where Joel’s essay has saved me from pushing a decision onto the user that I could make myself.

“Application compatibility layers are there for the customer, not for the program” by Raymond Chen (2010)🔗

Raymond Chen is one of the longest-serving developers on the Microsoft Windows team. His blog has thousands of informative, entertaining stories about the history of Windows programming, but the one I think back to most is one about compatibility mode in Windows Vista.

A customer had contacted Raymond’s team with this request:

Hi, we have a program that was originally designed for Windows XP and Windows Server 2003, but we found that it runs into difficulties on Windows Vista. We’ve found that if we set the program into Windows XP compatibility mode, then the program runs fine on Windows Vista. What changes do we need to make to our installer so that when the user runs it on Windows Vista, it automatically runs in Windows XP compatibility mode?

Raymond proceeds to characterize the customer’s request as follows:

I normally toss my garbage on the sidewalk in front of the pet store, and every morning, when they open up, somebody sweeps up the garbage and tosses it into the trash. But the pet store isn’t open on Sundays, so on Sundays, the garbage just sits there. How can I get the pet store to open on Sundays, too?

I loved this analogy. The metaphor was so funny that I didn’t notice until just now that Raymond is in the wrong. He’s making fun of a developer whose sin is expecting Windows not to break their app after a single release.

But as is the case with a lot of Raymond Chen’s writing, it’s so funny and sharp that I can look past the flaws.

Even though I disagree with the specifics, Raymond’s post is an excellent lesson in influencing user behavior.

If you want to nudge the user to do something that helps you, think carefully about the path of least resistance from the user’s perspective, because that’s the path they’ll take.

If you show the user that dumping garbage on the sidewalk completely solves their problem, they’re going to keep dumping their garbage on the sidewalk.

“Don’t Put Logic in Tests” by Erik Kuefler (2014)🔗

I’ve always loved unit testing and took great pride in my test code. That’s why I was so horrified when this essay appeared in my bathroom and revealed that I’d been writing awful tests my whole career.

Erik’s essay shows the following unit test, which has a subtle bug:

@Test public void shouldNavigateToPhotosPage() {
  String baseUrl = "http://plus.google.com/";
  Navigator nav = new Navigator(baseUrl);
  nav.goToPhotosPage();
  assertEquals(baseUrl + "/u/0/photos", nav.getCurrentUrl());
}

When I first read the essay, I thought, “That’s exactly how I write unit tests!”

Why duplicate the http://plus.google.com/ string in two places? Create a single source of truth, just like in production code. I did this all the time, adding helper functions, variables, and loops to eliminate redundancy from my tests.

The problem with the approach above is that it masks a subtle bug. It’s actually asserting that the URL looks like this:

http://plus.google.com//u/0/photos
                      ^^
                    whoops

Erik’s essay showed me that I shouldn’t treat test code like production code at all. The two have completely different goals and constraints.

Good test code should be, above all, clear. Test code doesn’t have its own test code, so the only way to verify correctness is by inspection. A test should make it blindingly obvious to the reader what behavior it asserts. In service of that goal, you can accept redundancy to reduce complexity.

“A little bit of plain Javascript can do a lot” by Julia Evans (2020)🔗

As a software engineer, I was embarrassingly late to the web. For the first 10 years of my career, I only wrote code for desktop apps and backend servers. I never bothered with HTML or JavaScript until 2017.

By the time I got serious about learning frontend development, my impression was that JavaScript was a mess of a language, hacked together in 10 days, and it had drastically different behavior in different browsers. If I was going to write web apps, I wanted something modern and sleek to protect me from all of JavaScript’s bile and warts.

So, I tried the popular web frameworks of the day: Angular, React, and Vue. I learned enough Vue to make my way around, but I was still spending an enormous amount of my time on dependency issues and framework gotchas. After all the work these frontend frameworks did to fix JavaScript, web programming still sucked.

Then, I read Julia’s essay and realized I’d been so confident that JavaScript needed fixing that I never gave it a chance.

At the time, I was working on the prototype of TinyPilot, which would become my first commercially successful software product. TinyPilot had a web interface that I was planning to implement with Vue, but Julia’s essay inspired me to see how far I could go with plain JavaScript. No framework, no wrapper libraries, no build step, no Node.js, just regular old JavaScript. Okay, not “old” — more like ES2018, but you know.

I kept expecting to hit some problem where I’d need to switch to some kind of framework or builder, but it never happened. There were still some gotchas, especially around WebComponents, but it was nothing compared to the suffering I endured with Vue and Angular.

I loved being free of the frameworks. When I had a runtime error, the stack trace wasn’t some minified, transmogrified, tree-shakified fever dream of my code. I was debugging my code, exactly as I wrote it. Why hadn’t I tried this sooner?

My biases about JavaScript were wrong. Modern JavaScript is pretty nice. It absorbed a lot of ideas from wrapper libraries, so now you don’t need the wrappers. And browsers got their act together to ensure consistent behavior across platforms and devices.

I haven’t integrated a JavaScript framework or build step into any new project since 2020, and I’ve never looked back. Plain JavaScript gets me 90% of the benefit of frameworks with 5% of the headache.

“Choose Boring Technology” by Dan McKinley (2015)🔗

This is an odd essay to include in this list because I’ve never actually read it.

People have quoted this essay to me, and once I understood the idea, it felt so intuitive that I didn’t need to read it. In my interview with CoRecursive podcast host Adam Gordon Bell, he talked about how there are certain non-fiction books where, once you understand the idea, all you need is the title. “Choose Boring Technology” is that for me.

Dan’s argument is that when you start a new project, you’re tempted to use cutting-edge technology that has lots of buzz. Google just announced a new database that scales to exabytes, and it’s 40% faster than Postgres at 20% the cost. You’d be an idiot to use Postgres when this sexy new alternative is right there!

In practice, the new technology has bugs and weaknesses, but they’re not obvious to you yet; they’re not obvious to anyone yet. So, when you run into them, you’re stuck. Postgres has its issues, but after 30 years in the field, it has battle-tested solutions for any problem you’re likely to encounter.

Dan concedes that you should use new technologies sometimes but only strategically and in limited quantities. He suggests that every business gets three “innovation tokens” to spend. If you want a flashy new database, you’ll have to spend one of your tokens.

Dan’s essay dovetails naturally with Julia’s essay. I wish I’d read either of them before I wasted all that time with frontend frameworks.

“I’ve locked myself out of my digital life” by Terence Eden (2022)🔗

Terence Eden is a delightful and eclectic technology blogger. He writes several new posts each week, but the one that impacted me the most was “I’ve locked myself out of my digital life.”

The article plays out what would happen if lightning struck Terence’s house and destroyed all of his possessions. He keeps his passwords to everything in a password manager, but if all his devices get destroyed, he can’t access his password manager. And he can’t fall back to hardware passkeys because those were in his house, too.

I always felt like I was pretty safe about my data because I store everything on redundant drives, and I have offsite backups on three continents with two vendors.

Terence’s post got me thinking about the many credible threats that could wipe out all of my devices simultaneously: fire, flood, electrical surge, criminal investigation. All of my data is encrypted with passwords that live in my head, so add to that list memory loss, incapacitation, or death.

Online services are bad at helping users recover from disaster. I use several services that assume it’s impossible for me to ever lose my phone, let alone my email account and every digital device in my possession.

Ever since I read Terence’s essay, I’ve been thinking more about which services and devices are critical to me, and how I could recover from a scenario like the one Terence described. The next time I bought a laptop, I set it up at the library to test whether I could recover access to my password manager and critical accounts without any of the devices in my house.

I still could do a better job at digital disaster preparedness, but Terence’s post always echoes in my head whenever I think about how to secure my devices and data. What if everything was suddenly destroyed?

Bonus: Brad Fitzpatrick on parsing user input (2009)🔗

It’s technically not an essay, but there’s a quote from a software interview I constantly think about.

In 2009, as a result of Joel Spolsky’s gushing review, (yes, again with the Joel), I read Coders at Work, a collection of interviews with accomplished programmers.

Brad Fitzpatrick, creator of LiveJournal and Memcached, appears in the book as one of the interviewees. He was only 28 years old at the time, the youngest programmer in the book and also the sweariest and most entertaining.

In response to a question about ethics in software engineering, Brad goes on an impassioned rant about input validation:

I would like to ask that everyone is consistent on their credit-card forms to like let me put in fucking spaces or hypens. Computers are good at removing that shit. Like don’t tell me how to format my numbers.

-Brad Fitzpatrick, in Coders at Work

I think back to this quote whenever I try to paste a phone number into a web form, and it whines that parentheses or spaces aren’t allowed. Or worse, it truncates my phone number because of the parentheses, and also complains that parentheses aren’t allowed.

Whenever I create input fields in my software and think about unexpected characters, I hear Brad Fitzpatrick say, “Computers are good at removing that shit.”