动物有语言吗

一只蜜蜂能把几百米外的方位抖出来，但它一辈子只会说这一类话

先看清这支舞在编码什么。巢里是黑的，巢壁是竖直的。蜜蜂跑八字，中间那一段一边直走一边左右摇摆，叫摇摆段。摇摆段偏离「重力的正上方」多少度，花就偏离「太阳的方位」多少度——正上方替太阳站着。摇摆段持续得越久，花越远，每多一秒大致多出将近一公里（具体换算随蜂群和地区有差别）。

这事被怀疑过几十年：会不会蜜蜂只是闻着舞者身上的花香飞？2005 年 Riley 和 Menzel 等人给蜜蜂背上贴了微型雷达应答器，把看过舞之后飞出去的蜂的航线一条条画出来。它们按舞里指的方向和距离直飞，中途还把侧风吹偏的量修了回来——而那个地方它们从没去过。飞到之后往往还要盘旋搜一阵才找到花：这支舞给的是指路，不是坐标。

再看它编码不了什么。这套编码里没有「词」。角度偏五度，意思就跟着偏五度，不会因此变成另一个意思。人的语音正相反：把「妈」的声母换成「趴」的声母，得到的不是「偏了一点的妈」，是另一个词——中间没有过渡地带，这就是「大脑给声音画的格子」那期讲的范畴知觉。

所以蜜蜂的位移性（displacement，能谈不在眼前的事）是真的，这一条上它比很多哺乳动物都强。但它一辈子只会说这一类话：方向、距离，加上「值不值得去」（舞跳得越久越卖力，招来的蜂越多）。分家找新巢址时用的还是同一支舞。它说不出「昨天那片花被割掉了」，也没法把「那个方向」和「那边有危险」拼起来——没有零件可拼。

EN A returning forager runs a figure-of-eight on the vertical comb inside the dark hive. In the straight middle stretch — the waggle run — the angle away from straight-up-against-gravity equals the angle of the flowers away from the sun's azimuth: vertical stands in for the sun. The longer the waggle run lasts, the further the food, at very roughly a kilometre per extra second, with the conversion varying by population and region. For decades people suspected the recruits were just following flower scent on the dancer. In 2005 Riley, Menzel and colleagues glued tiny radar transponders onto bees and plotted the actual flight tracks of recruits: they flew straight out on the indicated bearing for the indicated distance, correcting for crosswind drift, to a place they had never visited — then usually had to circle and search before finding the flowers. The dance gives directions, not coordinates. Now the other half. This code has no words in it. Five degrees off in the angle means five degrees off in the meaning; it never becomes a different meaning. Human speech is the opposite: swap the initial consonant of 妈 (mā, "mother") for that of 趴 (pā, "lie face down") and you do not get a slightly-off 妈, you get another word, with no transitional zone between them — the categorical perception from the topic on the grid the brain draws over sound. So the bee's displacement — talking about what is not present — is real, and on that one criterion it outperforms many mammals. But it only ever says this one kind of thing: direction, distance, and how much the trip is worth, since a longer and more vigorous dance recruits more bees. A swarm looking for a new nest site uses the same dance. It cannot say that yesterday's flowers were mown down, and it cannot combine "that direction" with "there is danger there" — there are no parts to combine.

教猩猩打手势之后，争论的焦点变成了谁在读那盘录像带

1966 年起，Gardner 夫妇把一只叫 Washoe 的黑猩猩当孩子养，教它打手势；后来有大猩猩 Koko、黑猩猩 Nim、倭黑猩猩 Kanzi——Kanzi 用的不是手势，是一块印满图符的键盘，一个图符代表一个词，这种符号叫 lexigram。这些项目当年都上了封面：猩猩能打出几百个手势，还会组合。

1979 年 Terrace 和同事把 Nim 的录像倒回去逐段看，整件事转了向。Nim 的大多数「句子」是在模仿老师刚打完的手势，或者插在老师还没打完的时候；它极少自己开启一个话题。它最长的那串有十六个词：「give orange me give eat orange me eat orange give me eat orange give me you」（给 橘子 我 给 吃 橘子 我 吃 橘子 给 我 吃 橘子 给 我 你）。长度是堆出来的，不是嵌出来的——人的句子变长靠往里套，这串只是往后接。

这不等于说猩猩笨。Kanzi 的理解测试是另一个样子：660 条口头英语指令，很多要求拿东西做没做过的事（「把牙刷放到苏打水里」这种），Kanzi 答对 74%，同一套题给一个两岁幼儿 Alia 做，对 65%。但这是一个个体对一个个体，没法复制，证据强度就这么弱。

真正的裂口在「谁在读那盘带子」。一百多年前德国有匹叫汉斯的马会做算术，后来发现它读的是提问者无意中泄露的身体姿态，这件事后来成了专有名词，聪明的汉斯效应（Clever Hans effect）；所以这类实验必须双盲。Pinker 书里引过 Washoe 项目组里唯一一位以手语为母语的聋人成员的回忆：每见猩猩打一个手势就要记一笔，「他们老抱怨我记下的手势太少」。一个人的回忆不是测量，但它点出的问题是实打实的：同一盘带子，判读标准松一点紧一点，结论就不一样。

还有一件要说清：手语是完整的语言，有自己的语法、自己的最小单位层级（「手语是完整的语言」那期讲的就是它）。这些项目里猩猩学到的是从手语里借来的一些手势，不是手语的语法。把两件事混着说，既抬高了项目的成绩，也亏待了手语。

EN From 1966 the Gardners raised a chimpanzee named Washoe like a child and taught her manual signs; later came the gorilla Koko, the chimpanzee Nim, and the bonobo Kanzi — Kanzi using not gestures but a board of printed symbols, one symbol per word, each called a lexigram. The projects made covers: hundreds of signs, and combinations. In 1979 Terrace and colleagues went back through Nim's videotapes, and the picture turned. Most of Nim's "sentences" imitated a sign the teacher had just made, or cut in before the teacher had finished; he rarely started a topic himself. His longest string ran sixteen words: "give orange me give eat orange me eat orange give me eat orange give me you". The length is stacked, not nested — human sentences grow by embedding, this one grows by appending. None of which makes apes stupid. Kanzi's comprehension test looks different: 660 spoken English instructions, many asking for novel uses of objects, such as putting the toothbrush in the soda. Kanzi got 74% right; a two-year-old child, Alia, given the same items, got 65%. But that is one individual against one individual, unrepeatable, and the evidence is that weak. The real fault line is who reads the tape. A century ago a German horse named Hans appeared to do arithmetic; he was in fact reading posture cues his questioner leaked without knowing it, and the name stuck as the Clever Hans effect. Studies of this kind have to be run blind. Pinker quotes the recollection of the only native signer on the Washoe team, a deaf member asked to log every sign the chimp made: "they were always complaining because my log didn't show enough signs." One person's memory is not a measurement, but the problem it names is real: the same tape, scored a little more loosely or a little more strictly, yields a different conclusion. One more thing has to be said plainly. Sign languages are full languages with their own grammar and their own level of minimal units — the subject of the topic on sign language being a complete language. What these apes acquired was a set of gestures borrowed from a sign language, not its grammar. Blurring the two both inflates the projects and shortchanges the languages.

把判据拆开，每一条上都有动物过关，可没有一个物种过齐

先看一条过得干脆的。1980 年 Seyfarth、Cheney 和 Marler 在肯尼亚把喇叭藏进草里，在没有任何捕食者的场合播放长尾黑颚猴（vervet）自己的警报声。放「豹」那一声，地上的猴往树上跑；放「鹰」那一声，树上的猴往下钻、往灌木里躲；放「蛇」那一声，猴站起来，低头在地上四处看。声音和捕食者种类对得上，而且不是靠看同伴在看哪儿——喇叭是藏着的。这是指称性的硬证据。

但它不像教科书画得那么干净。2015 年有人把录音拿去做定量聚类：三种针对捕食者的叫声确实分得开；可一旦把争斗场合发出的叫声也放进来就开始重叠——雌猴对蛇、对鹰和在争斗中发的声音彼此交叠，雄猴的吠声在「有豹」和「在打架」两种场合也混在一块。所以这些叫声里同时装着「判断了什么」和「当时多紧张」。

再看一条更往前的。2016 年 Suzuki 等人在野外播日本山雀（Japanese great tit）的叫声：ABC 这串音符让鸟警戒地四下张望，D 这个音符让鸟朝发声处靠近；把 ABC-D 连起来播，鸟两件事都做。关键是倒过来播成 D-ABC，鸟基本没反应。顺序会改变效果这件事是硬的，后来在别的鸟身上也做过。

不过「这算不算组合出来的意思」有争论。批评的一方说，鸟也可能只是把 ABC-D 整体当成第三个固定信号记住了，或者只是把两句话连着听而已；人的组合性是一个能自由套用的办法，拿任何两个合适的成分都能算出一个新意思，这一点在山雀身上没被证明。这条争论今天还在。

于是判据的样子清楚了：它不是一道门，是一张清单，每一条上都有动物过关。位移性——蜜蜂过了。指称性——长尾黑颚猴过了。顺序敏感——日本山雀过了。只有「开放性」这一条，到今天还只有人过：人能造出的新意思没有上限，因为声音先被拆成几十个可反复使用的零件，再拼成词、拼成句，这就是「语言不是一本词典」那期讲的双重分节。没有哪个非人物种被发现有这一层。

所以「动物有没有语言」这个问法容易把事情变成给物种排名。换成「这套系统能表达什么、被什么卡住」，才问得出上面这些实验。

EN Start with a criterion something clearly passes. In 1980 Seyfarth, Cheney and Marler hid loudspeakers in the Kenyan grass and played vervet monkeys their own alarm calls with no predator present. On the leopard call, monkeys on the ground ran for trees; on the eagle call, monkeys in the canopy dropped down and into bushes; on the snake call, they stood up and scanned the ground around them. Call matched predator class, and not by watching where a caller was looking, since the speaker was hidden. That is hard evidence of reference. It is less tidy than the textbooks draw it. In 2015 the recordings were reanalysed by quantitative clustering: the three predator calls do separate, but once calls given during aggression are included the categories start to overlap — females' snake, eagle and aggression calls run into each other, and males' barks mix across leopard and fight contexts. These calls carry both what was judged and how aroused the caller was. Next, a criterion that goes further. In 2016 Suzuki and colleagues played Japanese great tits their own calls in the field: the ABC notes make a bird scan warily around, the D note makes it approach the source, and played as ABC-D the bird does both. Crucially, reversed to D-ABC it mostly does neither. The order-sensitivity is solid and has since been run on other birds. Whether it amounts to composed meaning is argued over. Critics point out the bird may simply have memorised ABC-D whole as a third fixed signal, or be hearing two utterances back to back; human compositionality is a procedure applied freely, computing a new meaning from any two suitable parts, and that has not been shown in tits. The dispute is live. So the criterion turns out not to be a gate but a checklist, with something passing each line. Displacement: bees pass. Reference: vervets pass. Order-sensitivity: Japanese great tits pass. Only openness is still passed by humans alone — the new meanings a person can build have no ceiling, because sound is first broken into a few dozen reusable parts and only then assembled into words and sentences, the duality of patterning from the topic on language not being a dictionary. No non-human species has been found to have that layer. Which is why asking whether animals "have language" slides so easily into ranking species. Asking instead what a given system can express, and what stops it, is the question that produced every experiment above.