Wenlei Qu

鱼类对话

当你徜徉在蔚蓝的海面上,漫不经心地享受阳光时,一个声音突然从海底传来:“朋友,想加入海草游戏吗”?这似乎过于天方夜谭,但巴哈马群岛上通过专门工具与海豚进行交流的某研究计划宣称他们确曾听到过类似声音。


在过去的 29 年中,“野生海豚计划”(Wild Dolphin Project)一直在研究海豚的社会结构与语言。现在,该计划正践行“处彼之世界,守彼之规则(In their world, on their terms)”的原则。他们将海豚的啸声与滴答声组合起来,形成海豚语言。他们还运用谷歌眼镜技术负责人 Thad Starner 开发的鲸类声音遥测(C.H.A.T.)系统,将海豚语言播送到水中,并倾听之后的回应。最终,他们收到了期盼已久的回应;该项目中的一只小海豚使用了研究员创建的某个指代“马尾藻类海草”的词语,即某种海藻、玩具,据推测这表示小海豚希望玩某个游戏。


“野生海豚计划”的研究是独一无二的,他们在广阔的海面上已对同一海豚群体进行了超过 25 年的研究。其调查结果表明,海豚拥有复杂的社会结构——它们采用“母系社会”的组织形式,且往往会有 2-40 头海豚组成“豆荚”群体出行,并显示出某些我们可以称之为“人类”的特征。该项目侧重于研究海豚语言,如今,得益于信息论——这一发现数据结构的数学分支学科的进步,与海豚的第一次沟通可能即将成为现实。


运用信息论分析工具,访问“野生海豚计划”数据的科学家正在破译海豚语言。通过计算机输入海豚的录音,他们已经能够随机分离二进制代码,并发现海豚发出的啸声与滴答声实际上正是在传递信息。


这并不等同于真正意义上的“翻译”。使用鲸类声音遥测系统的研究人员一直在尝试使用他们针对海豚生活环境常见现象所创建的语言——例如,鲸类学家总是使用代表玩具球的悠长、震颤的啸声。这些人工创建的语言仍与海豚自身的语言有着很大差别。海豚确在讲一门语言,但其词汇与语法结构仍是个谜。


当谈及海豚语言到英语转换这一界面实现的可能性时,CSOFT 机器翻译专家及产品开发高级经理 Robert Derbyshire 表示:“我认为前景并不十分乐观。(机器翻译的)主要先决条件在于我们可以告诉计算机‘这一语言里的 A-B-C 意味着另一语言里的 X-Y-Z’,你需要一系列这样的内容,约有 10,000 组。


“野生海豚计划”负责人 Denise Herzing 希望通过其团队有限的跨物种沟通,最终能够揭示海豚语言的基础。Denise 指出:“如果你确已掌握这类数据,便有可能成功开发机器翻译引擎,进行从海豚语言到英语的音频文件翻译。”


在实现真正意义上的翻译设备前,我们仍有很长的一段路要走。但是,我们的努力对于人类在地球上的“地方感”有着深远意义,能帮助人们构建与环境的情感联系。或许,我们可以通过建立与地球上另一种智慧生命体的双向沟通,促进人类与自然和谐相处。


跨物种沟通或许能将我们变为严格的素食主义者——食用能够用言语提出抗议的动物将令我们深感不安。当麻雀能够抱怨发动机尾气为他们带来的哮喘病痛时,我们可能都会再次使用自行车。当栖居于世界不同角落的非人类物种能够对人类提出反驳时,我们该说什么呢?我想,我说出的第一句话应该是“很抱歉,我们做得一塌糊涂”。


Fishy Conversations

Floating on a cerulean sea, you’re mindlessly soaking up the sunshine when a voice calls from below, “Care for a game of seaweed?” That sounds silly, but a research program in the Bahamas using a gadget designed to communicate with dolphins claims they’ve heard something similar.


The Wild Dolphin Project has been studying dolphins’ social structure and language for the past 29 years. Now they’re living up to their “In their world, on their terms” principle. They’ve taken the dolphins’ whistles and clicks and strung them together to form words. Using a system called C.H.A.T.—Cetacean Hearing and Telemetry—developed by Thad Starner, the technical lead on Google Glass, they broadcast these words into the water and listen for responses. They finally got a long awaited reply; a younger dolphin the project works with used the researchers’ word for “sargassum,” a type of seaweed and plaything, presumably in an attempt to get a game going.


The Wild Dolphin Project’s research is unique in that it has worked on the open ocean with the same group of dolphins for over a quarter of a century. Their observations have revealed dolphins’ complex social structures—matriarchal societies, usually travelling in groups called “pods” of 2-40 animals and displaying characteristics we might call “human.” The project focuses on dolphin language and now, thanks to advances in information theory—a branch of mathematics that finds the structure within data—the first conversation with a dolphin may soon be a reality.


Armed with the analytical tools of information theory, scientists with access to the Wild Dolphin Project’s data are decoding dolphins’ speech. By feeding the recordings of the dolphins’ noises through their computers, they have separated binary code from randomness and found that the whistles and clicks are actually transmitting information.


This isn’t exactly “translation.” Researchers using the C.H.A.T. system have been attempting to use their invented language for common objects in the dolphins’ environment—e.g. a long, trilling whistle for the toy ball scientists always bring with them. These human-created noises are still very different from the dolphins’ own words. The dolphins are speaking a language, but its vocabulary and grammatical structure are a mystery.


“I’d say the prospect isn’t particularly hopeful,” said Robert Derbyshire, a machine translation specialist and Senior Manager of Product Development at CSOFT International, when asked about the likelihood of a dolphin-to-English interface. “The main prerequisite [of machine translation] is something where you tell the computer, ‘A-B-C in this language means X-Y-Z in that language,’ and you need a bunch of those, around 10,000.”


Denise Herzing, the head of the Wild Dolphin Project, hopes that through her team’s very limited interspecies interaction, they will eventually be able to uncover the foundation of the dolphins’ speech. “If you did have that sort of data,” said Derbyshire, “then feasibly you could develop an M.T. engine using audio files translating from dolphin into English.”


A true translation device is a long way off but the implications for our sense of place on the planet are profound. Perhaps establishing two-way communication with another Earthly intelligence will encourage environmental stewardship.


Maybe interspecies conversation will turn us all vegan—too uncomfortable eating animals capable of verbal objection. Maybe we’ll all dust off our bicycles when sparrows can complain about their engine exhaust-induced asthma. What will people say when the world’s innumerable inhuman inhabitants are able to talk back? My vote for first note: “Sorry about the mess.”


 


评论