中国作家协会主管

诗歌中的听觉场景分析与声音

在本篇简短的论文中，我想谈一下2011年以来我诗歌中的一个中心问题，一个既有主旨性又有技巧性的问题：诗歌中的声音。首先要说明的是，我指的“声音”并不是通常关于鲜明特征或风格的比喻含义。相反，我说的是声音的物质性是如何成为我们理解诗歌的一个因素的。当然，有一点要承认的是，在阅读纸上的文字时，只有纸张和笔墨的无言相伴，因而任何对此类声音的考量或讨论都必须是假想的或象征性的。但我希望能避免对声音作此类象征性处理，因为这通常是创意写作研习班的研究对象。

我将从另一位学者的声音开始说起，即比尔·梅德门特（Bill Maidment）的声音，他是悉尼大学的教师，我们大约相识于1991年，直至2005年他去世，他一直是我的良师益友。2006年初的一个早晨，在悉尼大学附近的格利伯区，我经历了一场入睡前幻觉。比尔的声音将我唤醒，还是生前老友的感觉，跟我说着什么。对此，我大吃一惊：就好像他与我同处拂晓前的房间，而我努力地分辨他在说什么，且事实是我认为自己是确确实实地听到了，而不是想象出来的。于是，就出现了几个问题并亟待解答：这些问题兼具哲理性、技巧性和伦理性。比尔的去世给我的创造力带来了危机，一直持续了好几年。期间对解答这些问题的关注引我走上了三条诗歌之路，且自那时我已意识到，这三条路即使不是相互纠缠，至少也是彼此叠盖。

首先是押韵的问题。在我迄今为止的诗歌实践中，对声音的考量一直都是颇具创意的推动因素，且之前我就已将押韵视为诸多诗歌创作手法中的一项，其他还有准押韵、头韵和对偶句的交错配列等，说到底，这些都是听觉重复原则上的各种变体。苏珊·斯图尔特（Susan Stewart）在她的文章《诗歌之声/声之诗歌》（The Sound, of Poetry/ The Poetry of Sound）（马乔里·佩洛夫（MarjoriePerloff）和克雷格·德沃金（Craig Dworkin）编辑，2009年）中，对押韵提出了类似的主张。2005年之前，我就创作了一些押韵诗，但强烈地感受到了几位澳大利亚当代诗人对押韵出于本能的蔑视，并将其视为（诗歌领域）政治保守主义的标志，为此我搜寻了更为创新的方法以探讨押韵问题。我首先研究了约瑟夫·布罗茨基（Joseph Brodsky）诗歌英译版本中的半押韵，随后很快转而探索保罗·马尔登（Paul Muldoon）在押韵上的创新，主要关注他们在诗节中的倾向性和非对称结构以及押韵诗行的长度差异，这些追根溯源都得自路易斯·麦克尼斯（Louis MacNeice）。

几次试验后，我得出了自认为原创性的解决方案。虽然我声称原创性，但我只是稍稍延迟地响应了他人的成果：由于在北京无法登录我的图书馆，我只能拿美国诗人莉兹·瓦尔德纳（Liz Waldner）作例子，因为她的方法和我尝试的方法一样有条理。我的想法是对押韵诗行的词尾音素进行重组，作为一种限定。实际上这要涉及到四个音节和偶尔的音素倒置。其结果就是梅健青（Kim Cheng Boey）所说的“流畅且铿锵有力的音调（mellifluous sonorousness）”，但这仅仅是系统性打乱音素群的一个副产物：它看起来、听起来都很美，但之前却是以断裂的概念开始的，即通过这种方式将语言拆分开，我实际上是在辨别所谓声音的组成要素，我们应称之为音素（phoneme），但我更倾向于称之为鼓膜素（tympaneme），容后详述。下面拿我的诗《海岸线》（Coastline）做个例子，大家或许能发现它并没有完全达到其灵感上的严苛，因为它将阳韵和阴韵与音素重组混合在一起（恐怕其结果是完全无法翻译出来的）：

I walked the cliff-top walk, totally alone

at the other end of love, on the way from one littoral

to another, balancing an act

in a world out of balance, piecing together words

to confront something, long ago put to the sword.

我行走于峭壁之巅，踽踽一人

在爱的彼端，从一个海岸

到另一海岸， 努力平衡着

在这失衡之世界，拼凑零言碎语

直面那，早已被剑斩杀之物。

Level with my eyes a seagull hovered, motionless

into the wind. I passed beer cans in modern middens,

dandelions on the path’s port side

while slowly from the north-east, thunderheads of mackerel-

mottled clouds began to coolly spit on the caramel-

视线同高处有海鸥翱翔，似静止般

隐入风中。途经垃圾箱中的啤酒罐，

道路左舷有蒲公英

东北侧缓缓来，云砧似鲭鱼

斑驳陆离 悠然轻吐向焦糖颜色的峭壁。

coloured cliffs. It’s funny how it worms

its way in, love, diasporated like a swarm

of angry bees bearding a heart.

The continents are the oldest divorcees, having drifted

apart for eons. Next to them we’ve barely tiffed.

多有趣 缓慢蠕动

似此番，爱，迁居如一群

挂在心上的愤怒的蜜蜂。

大陆乃是最老的离婚者，永远飘移分离。

相形之下 我辈拌嘴不足记。

比尔去世后，我追寻的第二条诗歌之路是对“声音”相当全面的考量：它是什么，意味着什么，有哪些不同的维度，如认识论的、现象学的、诗歌的、政治方面的、精神上的维度等。鉴于声音在表达上的中心地位——是最根本的必要性——，那么对哲学、心理分析甚至是语言学中声音的研究如此之少就很令人惊讶了。仿佛产生并维持人们的言语互动的工具不值得（或者是可能经不起）类似索绪尔对语言系统所做的持久分析。在尝试将声音恢复为诗歌本身的一项主体时，我的关注点又一次回到尝试从其基础层面来理解它，或者说剖析它。通过这种做法，我持续地返回到一个观点：声音，即音响的连续统一体，被细分为其组成要素，或者说音素。此处，我想到了拉康（Lacan）的观点，“指示结构的要素最终指的是音素，它使语言结构得以具体化。”音素是能指的主体，但它本身没意义：在指示动作中，意表主体或分裂的自我消失不见。对音素此方面的考量部分组成了《声音解剖学》的第二次分割，其中第二至第十首诗是由第一首诗中摘出的第一诗行组成，且在接下来的五个诗行中（或多或少地）在音素方面进行了重组。从主题性和互文性看，我还坚持这样一个观念：由其他声音组成声音，“里面有其他的东西掩盖着黑暗”，这是巴赫金杂语理论的一个版本。

现在我想要走出诗歌的技巧性考量，讨论下与我的声音调查有关的更为广泛的问题。第一个涉及从心理分析角度看声音是什么。婴儿首先会做的事情之一就是哭。从拉康主义的角度看，这显然是对其他人的识别和诉求。这对诗歌中任何声音的考量都十分重要：即它涉及了对他人的识别和诉求。第二点涉及声音如何呼吁他者：即通过听的方式。人们可以争论说，现代诗歌最为重要的发展在侏罗纪早期，大约在1.95亿年前，当时出现了吴氏巨颅兽，一种小型哺乳类动物，现已灭绝，约长3.2厘米，长着几乎完整的哺乳类动物的耳朵。人们无法想象，如果没有听力功能，哺乳动物（更不用说人类）会如何演变。但肯定的是，如果没有这个功能，诗歌就不会存在。无论是否有音乐伴奏，诗歌起初是用来吟唱的，这样一个假设受到了德里达（Derrida）的挑战。他的论据是书写先于言语存在类似于声音先于听觉这一假定。耳朵的发育是为了响应声音吗？还是说声音是在耳朵的存在下才产生的？又或者是，按照法国评论家罗杰·凯洛伊斯（Roger Caillois）的说法，我们不把声音视为具有通过自然选择赋予的功能，而是将其视为出现在世界上的一种模式，或是模式的一部分？

当我们认为进化论可能无法解释时，声音的第三个方面就出现了，并且我们需要神话来介入。在希腊神话中，听力、听觉和死亡有着密切的联系。听到赛壬（古希腊传说中半人半鸟的女海妖）的歌就等于走向死亡。赛壬远房姐妹斯芬克斯的谜语，听了但破解不了就要死去。当克劳狄斯将毒药倒入老哈姆雷特的耳朵里时，死亡就随着听觉一起来临。听觉只不过是内耳基底膜上的铭文，它是通过听小骨的机械运作产生的，这点又是由鼓膜的振动引起的：它是一种书写形式，其基本单位或可被称为“鼓膜素（tympaneme）”。正如德里达指出的，早在很久以前，书写就和死亡、腐朽放在一起理解，但是还有另一种方法来看待这种铭文式听觉和死亡之间的联系，那就是信息论。信息复制的热力学原理决定了精度和能量间必然有此消彼长的平衡。生物学层面信息复制的不精确一定会带来衰老和死亡，不幸运的话甚至可能会导致癌症，。从这个角度出发，在神话中，耳内信息复制的不精确应和死亡紧密联系在一起，这看起来就是十分恰当的类比了。

说到这点，我必须承认自己对信息论的认识受到对其数学运算的详细了解的限制（我怀疑我们中的大多数人都这样）。或许正是因为这个原因，信息论在文学研究领域的应用十分贫瘠，可为人尽皆知。同样，用信息论分析文学文本的“信息”，往往是基于对文本的精确理解，包括单词或字母所被赋予的二进制值。这点从计算型文学研究，正如我的同事，纽卡斯尔大学的休·克雷格（Hugh Craig）所实践的那样，延伸到尼尔·鲁本金（Neil Rubenking）的“Brekdown”项目中n-1马尔可夫阵列的使用。自20世纪80年代后期，澳大利亚诗人约翰·特兰特（John Tranter）好几次把该项目用在自己的作品中。在特兰特的例子中，表面上他力图创造“一个无作者的文学文本……一个没有作者意图，没有隐藏的文化、社会、经济和政治价值观，且没有隐藏的个性议题，而是只产出纯粹的‘文学’。尽管特兰特是在说反话，但我不确定这种纯粹究竟是否可行，因为鲁本金的项目和克雷格的研究也是无声的。但是，如果此类信息论技巧分析的信息单位是鼓膜素，结果会不一样吗？在某种意义上来说不会，概率分析将得以运用；但是从另外一种意义上看答案几乎是肯定的，因为不精确首先会成为一个更大的因素，声音本身的冗余和在其编码过程中捕捉到的周遭杂音也是如此。这就是我在试验Microsoft Word 2005曾经自带的语音识别软件时做过的事，它是我对研究声音的诗歌性反应中第三也是最不重要的一部分。关于这些实验，还有很多可以说的，但此处时间和篇幅有限。我就只列举自己阅读中的两个例子。鉴于特兰特已经使用“Brekdown”项目给每首诗取了个标题，且这些标题是那些听错的诗的变位词（anagram）。或许，正如雅克·阿塔利（Jacques Attali）所言，音乐的变化是文化变化的先兆，我们可能提及最多的是，20世纪90年代的低保真流行乐和摇滚乐唱片运动预示了这些试验的发生。

最后，我想根据前述内容再做些推断，并给出建议使研究更进一步。我们的听觉并不完美：完美是不可能的，因为听觉在转录过程中信息的能量会有所丢失。同样遗忘也需要能量，因此也不完整。然而，我们的文学制度在学术精度上有其基础，在西方，这种基础是建立在科学方法之上的。我们急于保存文本和某些含义，但是说到不确定性的对数测量（即一则消息所含的信息量）与含义的相对确定性（它或多或少指的是消除不确定性）之间的关系，我们可能会觉得别扭。思考这种关系的另一个方法，正如乔纳森·斯威夫特一首诗中所述“秩序出自混乱中/艳丽鲜花粪里来（order from confusion sprung/ such gaudy tulips raised from dung）”。此处说的是，正是由于我们容易误听、误解且选择性地接收信息的倾向，才使我们找到一些重要的可能途径来取得进展：请容许我强调，此处我并不是在说“虚假新闻”，或是对气候科学的顽固无知；而是把评论限制在诗歌领域。与我们对古典信息论的期待相反，比起在全是寂静的孤立状态下听取打断的言语，人类大脑在拥挤或吵闹的环境中，更容易正确听取并理解受阻信息。这就是阿尔伯特·S·布雷格曼（Albert S. Bregman）所说的“听觉场景分析”，我认为它对我们思考诗歌中的“声音”有些启示。如果杂音、误听等表面混乱不仅自然，还有助于产生新的含义，又会如何呢？如果不是单单在系统和连贯中找出抽象的难点和分离，如果这些的来源也是声音本身、音色、声调、重音和响亮度，还有氛围、共鸣和或许更为重要的多重性，又会如何？说话声音中的冗余标志着对可理解含义的充实的理解，正如它们在杰罗姆·罗滕伯格（Jerome Rothenberg）的翻译中可能做到的那样，但是它们也可能标志着对不同形式的押韵及它的近亲韵律重新燃起了兴趣。这样做将成为回归主体和分裂的自我的另一种方法，或许是通过这样一种方法，它会展开听觉场景中的指示意图，并带来通过诗歌的探索。

Auditory Scene Analysis and Voice in Poetry

David Musgrave

In this necessarily brief paper I want to talk about one of the central preoccupations in my poetry since 2011, which is both thematic and technical: the question of ‘voice’ in poetry. From the outset I want to be clear that I do not mean ‘voice’ in its customarily metaphorical sense of signature or style. Instead, I mean how the materiality of voice can be part of our understanding of poetry. Of course, I acknowledge that, in reading words on a page, any consideration or discussion of such voice must be imaginary or figurative, as there is only the silence of paper and ink to accompany it; but I hope to avoid the kind of figurative treatment of the subject which is often the subject of creative writing workshops.

I want to start with the voice of another, in this case that of Bill Maidment, academic at the University of Sydney and my friend and mentor from around 1991 until his death in 2005. One morning (in Glebe) in early 2006 I experienced a hypnagogic hallucination. It was Bill’s voice waking me, saying something to me in keeping with the nature of our friendship. I was startled: it was as if he was in the pre-dawn room with me, and I struggled to process what it was he said and the fact that I thought that I had actually heard and not imagined it. Several questions arose which demanded answers: these were philosophical, technical and ethical questions. In the following years of a creative crisis engendered by Bill’s death, my concern with answering these questions led me down three poetic paths which I have since realised were, if not implicated then at least imbricated with each other.

The first of these was a question of rhyme. In my poetic practice hitherto, considerations of sound had always been a driving, creative force, and I had always considered rhyme as one of a number of poetic devices, along with assonance, alliteration, chiasmus and so on, which are ultimately variations on the principle of aural repetition. Susan Stewart claims as much for rhyme in her essay in The Sound, of Poetry/ The Poetry of Sound (ed.Marjorie Perloff and Craig Dworkin, 2009) and prior to 2005 I had written several rhyming poems, but acutely aware of the almost visceral disdain for rhyme on the part of several contemporary Australian poets, as well as it being a marker for a (poetically) reactionary political conservatism, I searched for ways to explore rhyme in more innovative ways. My encounter with the half rhymes of the English translations of Joseph Brodsky’s poems was quickly superseded by the discovery of Paul Muldoon’s innovations with rhyme, in terms of their slantness and their asymmetrical arrangement across a stanza and the differing lengths of rhyming lines, which ultimately derive from Louis MacNeice.

After several experiments I arrived at what I felt was an original solution, although as with all claims to originality, I had merely belatedly echoed the achievements of others: without access to my library here in Beijing, I can only offer the example of the American poet Liz Waldner as someone whose approach is as systematic as I have tried to be. My idea was to adopt, as a constraint, the rearrangement of the terminal phonemes of rhyming lines, which in practice involved up to four syllables and which also involved the occasional inversion of phonemes. The result was what Kim Cheng Boey called a “mellifluous sonorousness”, but this was merely a by-product of a systematic derangement of the sense of a few phonemes grouped together: it looked and sounded nice, but had begun with what was a notion of fracture: that by taking apart the language in this fashion, I was in effect identifying what might be a constituent element of the voice, which we should call the phoneme, but which I would prefer to call the tympaneme – more of that later. Here is an example from my poem ‘Coastline’ which, as you may be able to see, does not completely live up to the rigour of its inspiration, as it mixes masculine and feminine rhymes with phonemic rearrangement (the effects I refer to are, I fear, completely untranslatable):

I walked the cliff-top walk, totally alone

at the other end of love, on the way from one littoral

to another, balancing an act

in a world out of balance, piecing together words

to confront something, long ago put to the sword.

Level with my eyes a seagull hovered, motionless 

into the wind. I passed beer cans in modern middens,

dandelions on the path’s port side 

while slowly from the north-east, thunderheads of mackerel-

mottled clouds began to coolly spit on the caramel-

coloured cliffs. It’s funny how it worms

its way in, love, diasporated like a swarm

of angry bees bearding a heart.

The continents are the oldest divorcees, having drifted

apart for eons. Next to them we’ve barely tiffed.

The second poetic path I went down after Bill’s death was a fairly extensive consideration of ‘voice’: what is it, what does it mean, what are its different dimensions: epistemological, phenomenological, poetic, political, psychic and so on. Given the centrality – indeed, fundamental necessity – of the voice to utterance, it is surprising how little has been devoted to the voice in philosophy, psychoanalysis and even linguistics. It is as if the vehicle which enables and sustains our verbal interactions is not worthy of (or perhaps not amenable to) the kind of sustained analysis that a de Saussure, for example, was able to undertake with regard to the system of language. In attempting to recuperate the voice as a subject for poetry itself, I again found myself concerned with trying to understand it in its fundamental aspects, anatomizing it if you like. In so doing, I kept coming back to the idea of a voice, which is a continuum of sound, being reduced to its constituent elements, or phonemes. Here I was reminded of Lacan’s suggestion that ‘the element that [the] signifying structure ultimately refers to is the phoneme, which materialises the structure of language.’ The phoneme is the subject of the signifier, but the phoneme itself lacks sense: the signifying body or divided self disappears in the act of signification. A consideration of this aspect of the phoneme partly constitutes the second partition of Anatomy of Voice, where the 2nd to 10th verses consist of a first line taken from the first verse and rearranged phonemically (more or less) in the succeeding five lines. Thematically and intertextually, I also adhered to the notion of voice being constituted by other voices: “inside it there are others/ tenting the dark”, a version of Bakhtin’s heteroglossia if you like.

I’d like now to move beyond poetically technical considerations to some broader issues to do with my investigation of voice. The first relates to what a voice is from a psychoanalytic point of view. One of the first things a baby does is to cry. From a Lacanian perspective, this is clearly a recognition of, and an appeal to an other. This is of profound importance to any consideration of voice in poetry: that it involves a recognition of and an appeal to the other. The second point relates to how the voice appeals to the other: through hearing. One could argue that the most significant development in modern poetry took place in the early Jurassic, around 195 million years ago, with the appearance of Hadrocodium, a small mammaliaform, now extinct, approximately 3.2cm long and possessing a nearly full mammalian ear. One cannot speculate how the evolution of mammals, let alone humans, might have occurred without the faculty of hearing, but it is certain that poetry would not exist without it. The assumption that poems were first sung, accompanied or not by music, has been challenged by Derrida. His argument, that writing precedes speech, has its analogue in the supposed precedence of voice to hearing. Did the ear develop in response to the call of the voice, or did voice arise at the invitation of the ear? Or, following the French critic Roger Caillois, do we see the voice as not having a utility bestowed upon it through natural selection, but rather as a mode, or part of a mode of appearing in the world?

A third aspect of voice emerges when we consider that the theory of evolution might have failed us, and we need myth to step in. In Greek mythology there is an intimate connection between listening, hearing and death. To hear the song of the Siren is to be lured to one’s death. To listen to the riddle of the Siren’s distant cousin, the Sphinx, and not to solve it is to die. Death arrives in the manner of the poison Claudius pours in Old Hamlet’s ear, with hearing. And hearing is nothing more than the inscription on the basilar membrane of the inner ear, by way of the mechanical operation of the auditory ossicles, which are set off by vibrations of the tympanum: a form of writing of which the fundamental unit may be termed the tympaneme. Writing has long been understood in terms of death and corruption, as Derrida has noted, but there is another way of looking at this connection between Inscriptive hearing and death, and that is through Information Theory. The thermodynamics of information copying dictates that there must be a trade-off between precision and energy. The imprecision of informational replication at the biological level may lead to cancer if one is unlucky, but certainly leads to senescence and death. The imprecision of informational replication in the ear should be closely associated in myth with death seems, from this point of view, to be more than an apt analogy.

Having arrived at this point, I must confess that my understanding of Information Theory is limited, like most of us I suspect, by any detailed understanding of its mathematics. Perhaps for this reason the application of Information Theory to literary studies has been notoriously sterile. Not the least reason for this is that the ‘information’ of a literary text subjected to Information Theoretical analysis tends to be based on the mathematical understanding of a text consisting of binary values assigned to words or letters. This extends from computational literary studies, as practiced by my colleague Hugh Craig at the University of Newcastle, to the use of n-1 Markov arrays in Neil Rubenking’s program ‘Brekdown’, which has been utilised by the Australian poet John Tranter in several of his compositions since the late 1980s. In the case of Tranter, he ostensibly seeks to create ‘a writer-free literary text… A text free of authorial intentions and without buried cultural, social, economic and political values and hidden personality agendas, giving forth only “literature” in its pure state.’ I am not sure such purity is at all possible, even if Tranter means it ironically, for both Rubenking’s program and Craig’s researches are also voice-free. But if the unit of information subjected to such Information theoretical techniques was the tympaneme, would the result differ? In one sense no, probabilistic analysis would style apply; but in another sense, the answer is most assuredly yes, for the imprecision would firstly be a much greater factor, as would the redundancy of the voice itself and any ambient noise captured in its encoding. This is precisely what I have done in my experiments with the speech recognition software that was once native to Microsoft Word 2005, which is the third, and least significant strand of my poetic response to researching voice. There is much that I could say about these experiments, but do not have the time or space here. I offer two examples in my reading, and in deference to Tranter’s use of ‘Brekdown’ have given each poem a title which is an anagram of the poem which has been misheard. Perhaps, as Jacques Attali has noted, changes in music function as a harbinger of cultural changes, and the most we might say of these experiments is that they were prefigured by the lo-fi pop and rock recording movement of the 1990s.

I’d like to conclude with some more speculation, drawing on what I have already put before you, and some suggestions for ways forward. We do not hear perfectly: to do so is impossible, for there is energy in the message lost through the act of aural transcription. Similarly, forgetting requires energy, and is also therefore imperfect, yet our literary institutions have their foundations in a scholarly precision which, in the west, is founded on the scientific method. We are anxious to preserve texts and certain meanings, yet we are uncomfortable, perhaps, with the connection between the logarithmic measure of uncertainty, which is the amount of information a message contains, and the relative certainty of meaning, which is more or less the elimination of uncertainty. Another way of thinking of this connection is as ‘order from confusion sprung/ such gaudy tulips raised from dung’, where it is through our propensity to mishear, to misapprehend and to selectively listen to the messages we receive that important possible paths forward are opened to us: and let me stress that I am not talking about ‘fake news’ here, or wilful ignorance of climate science; I’m restricting my comments to poetry. Contrary to what we would expect from classic Information Theory, the human brain’s ability to correctly hear and interpret an interrupted message is easier in a crowded or noisy environment than it is when an interrupted utterance is listened to in isolation where silence fills the gaps. This is what Albert S. Bregman calls ‘auditory scene analysis’ and it has, I think, implications for our thinking of the ‘voice’ in poetry. What if the apparent chaos of noise, mishearing and so on were not only natural, but conducive to the production of new meanings? And rather than exclusively finding in systems and coherences abstract aporias and disjunctions, what if a source of these was also voice itself, its timbres, intonations, stresses and sonorities, as well as its ambience, resonances and perhaps even more importantly, pluralities? The redundancies in the speaking voice signal an enriched understanding of intelligible meaning, as they perhaps do in Jerome Rothenberg’s translations, but they might also signal a revival of interest in rhyme, in all its varied forms, and its near relation, rhythm. To do so would be yet another way to return to the body and the divided self, perhaps in a way that unfolds the signifying intention in an auditory scene and which invites exploration through poetry.