Pirahã: a Linguistic Anomaly

Indo-European Language Tree

Imagine a rare, exotic language, that contains some of the most uncommon sounds pronounced by man; a language with no words for number or color, but that can be hummed or whistled instead of spoken.  You’ve imagined Pirahã, the language spoken by the 300 or so indigenous Pirahã peoples of Amazonas, Brazil. And, though the number of Pirahã speakers is small, use of the language is vigorous and most speakers are monolingual.  Pirahã isn’t going extinct any time soon.

Several incredible facts combine to make Pirahã, without a question, one of the strangest and most interesting languages out there.

Pirahã is a language isolate.  Language isolates—Basque and Korean are two famous examples—are incredibly rare and invariably interesting.  Assume that human language is a huge, old-growth tree, beginning with Indo-European and branching into Sanskrit, Celtic, Italic, Germanic, and so on: then, Pirahã is just a single sprout growing a ways away from the tree’s roots. *

There are two possible explanations for Pirahã’s isolation.

  1. A long, long time ago, back when the Indo-European tree was just a sapling, it cast a seed into the wind, which settled to plant the Pirahã sprout.
  2. The Indo-European tree and the Pirahã tree grew independently.  They may as well be of entirely different species!

Option 2 is the most interesting to me.  It suggests that the human race split off into the Amazonian jungle, to live in isolation, before language itself developed.  Further, it suggests that language evolved twice, once in the main tree and once in the Pirahã sprout.  This phenomenon might be compared to convergent evolution in evolutionary biology, where two animals that share similar traits acquired those traits through separate evolutionary pathways.  For example, wings evolved twice: once in birds, once in insects.

Option 1 is quite interesting as well, though, since it might tell us a lot about what Indo-European looked like when it was just a sapling.  This is especially true if we assume that Pirahã hasn’t changed much over the millennia (which, unfortunately, it probably has).  Still, language isolates may give us a glimpse into ancient language in general.

Pirahã doesn’t have words for number or color.  Counting is so unnecessary in tribal Amazonian life that the words used to count never evolved.  Color words haven’t evolved either, although the Pirahã do describe color using analogy: instead of the bird’s tail was red, they say, the bird’s tail was like blood.

One must wonder if these limits on language produce limits on thought.  Can the Pirahã truly see red, in the abstract?  Or do they simply see a color that resembles blood?  Perception of color is difficult to measure empirically, but numbers aren’t as difficult.  Research has shown that, if given a pile of ten pebbles to the left, and eleven to the right, the Pirahã can tell you which pile has more pebbles.  However, if the pebbles are removed, they are not able to recall which side had more pebbles before they were removed.  The words ten and eleven serve as semantic placeholders in our memory; it’s much easier to remember the word “ten” than to remember the appearance of ten pebbles.  Without these placeholders, the Pirahã struggle to generalize the concept of number.

This observation provides strong evidence for the concept of linguistic relativity, often known as the Sapir-Whorf hypothesis.  This hypothesis states that language shapes or even confines thought, and therefore, the lack of a word might result in the inability to grasp the concept denoted by that word.  The Pirahã certainly seem to have trouble grasping the concept of number.  They requested that they be taught to count, since they feared they were being cheated in trade with other tribes.  However, after eight months of lessons with linguist Daniel Everett, the Pirahã discontinued their instruction, since they felt that they were incapable of learning numbers.  Not a single student had learned to count to 10, or even to add 1+1 (Source 2).  This suggests that, since they didn’t learn words for numbers in childhood, the Pirahã face a permanent incapability to understand numbers later in life.  This theory could be tested by teaching the next generation how to count, but how would the parents do this, without words for numbers, or even an understanding of the concept?  Of course, Everett could try, but language barriers would make it difficult.

Pirahã lacks recursion.  Recursion is the ability to nest ideas within ideas, and those within more ideas, and so on.  It’s seen a lot in computer science.  For an example of recursion in language, consider the following sentence: Bob went to a party on the beach near Sally’s brother’s house, after running some errands downtown.  Many ideas are encapsulated in that one sentence, but the Pirahã simply can’t do that.  We say: “Sally’s brother’s house.”  They say: “Sally has a brother.  This brother has a house.”  Consider how long it would take to say my example sentence above, if each idea required its own sentence!  Yet, the Pirahã manage.  **

The Pirahã’s lack of recursion threatens to topple decades of established linguistic theory.  Noam Chomsky was a linguistic giant of the 20th century.  His principal theory was that of linguistic universality, which holds that certain elements are universal across all human language.  Chomsky’s theory might be seen as the inverse of the Sapir-Whorf hypothesis: while the latter holds that language shapes thought, the former holds that our inherent predispositions towards thought shape language.  Well, recursion is one of those elements of language that Chomsky had held to be universal.  Pirahã disproves recursion’s universality, and by extension, may disprove the fact that a necessarily universal component of language exists at all.  On the other hand, Chomsky claims that he’s still safe: he maintains that just because one language may lack one widespread aspect of language doesn’t mean that those aspects that tend to manifest themselves universally should be disregarded.

Pirahã can be whistled, sung, or hummed.  Pirahã language can be fully encoded into music, with vowels and consonants left out entirely.  In their place, meaning comes from rhythm, stress and pitch.  Mothers teach their children Pirahã through repetition of musical patterns, and use these tunes for comfort and lullaby (3).  Linguists have argued that those who study Pirahã miss a lot from the already nearly-indecipherable language by ignoring aspects of its musicality.  Perhaps there’s still much, much more that we don’t understand.

As it stands now, though, Pirahã stands in support for the Whorfian idea that we think through language, not alongside it.  Meanwhile, it stands against at least the strongest Chomskyan view, which is that, as humans, certain universal characteristics will inevitably appear in our language.  Maybe, though, Chomsky was completely right all along, and recursion is universal in language—our language, that is—and the seed that planted the lone Pirahã sprout came across wind and sea, from a strange tree entirely unknown.


  1. Pirahã language
  2. Cultural Constraints on Grammar and Cognition in Pirahã
  3. THE INTERPRETER: Has a remote Amazonian tribe upended our understanding of language?



* Two things here.  Firstly, calling Pirahã a lone sprout might be a little misleading. Pirahã used to have relatives, and may have once resembled more of a small shrub, altogether called the Mura family. But all of its parents and cousin languages have gone extinct, leaving Pirahã as an isolate. Options 1 and 2 below are still relevant, but they should be applied to Mura as a whole, not solely to Pirahã.  Incidentally, Basque is a true isolate: no language has ever been discovered that, to our knowledge anyway, even remotely resembles Basque.  Thus Basque is a true lone sprout.  For an even more divergent side-note, Martin Guerre was Basque.  Among the evidence that helped convict his impostor, Arnaud du Tilh, was the fact that Arnaud didn’t speak Basque!

Secondly, calling Indo-European its own tree might be misleading as well.  I do it for simplicity’s sake, but as one can see from the picture, Indo-European is portrayed as just a small offshoot of the massive language system beginning with the Mother Tongue.  I don’t completely agree with the picture, though, either.  While it is true that Indo-European is just a small part of the tree and many languages came before it, the assertion that all language came from a single mother tongue is highly debatable.  In option 2, below, I discuss the fact that Piraha, as well as other language isolates, could have evolved independently from the main language branch.  Thus two or several mother tongues could have existed.

** As an interesting side-note, some argue that the Pirahã’s inability to nest ideas causes their inability to count, since number theory requires recursion.  For example, 5 = 2 + 3 = (1+1) + (1+1+1).   If this were true, no amount of teaching could allow the Pirahã to count, that is, unless recursion were implemented into their language.


6 comments on “Pirahã: a Linguistic Anomaly

  1. Ben says:

    Some mathematicians, tasked with the presentation of a rigorous definition of the integer number system, define the set recursively. After axiomatically positing the existence of two operations, which we call addition and multiplication (and which are related by the distributive laws), and the existence of additive and multiplicative identities, which we call 0 and 1 (respectively), these mathematicians define 2 as (1 + 1), 3 as (2 + 1), 4 as (3 + 1), and so on.

    This is far from an “intuitive” conception of the integer number system. Indeed, it seems far from natural human understanding. Maybe it’s not as far off as we think.

  2. Richard says:

    “Pirahã disproves recursion’s universality, and by extension, may disprove the fact that a necessarily universal component of language exists at all.” I know Josh has noted that this is controversial, but let me reiterate: this is controversial. Most theoretical linguists do not believe that Piraha lacks recursion; there is an article, by now well known, by Andrew Nevins, David Pesetsky and Cilene Rodrigues, called ‘Piraha Exceptionality: A reassessment’ in Language, Volume 85, Number 2, June 2009, pp. 355-404 which aims to show that recursion and embedding is present in Piraha, albeit in a more rarefied form.

    Regarding the Sapir-Whorf hypothesis, as this is usually discussed academically, there are both strong and weak versions of it. On the weak version language – i.e. natural language, like Piraha or some Inuit language – simply exerts an influence on how we think. This seems fairly uncontroversial. There is something called, rather turgidly, ‘synesthetic cross-modal abstraction’ which shows that certain phonemes tend to be associated with corresponding shapes and visual stimuli. The word’s phonology will then influence how you expect the world to be. Suppose I show you two pictures of creatures. One is spiky and angular, the other is pudgy and curvy. Then if I say one is called ‘Kiki’ and one is called ‘Booba’, you will probably guess that the spike one is Kiki and the pudgy one is booba; at least, most people do. Also, the fact that we have language at all may influence our capacity for thought. If I’m speaking inside my head, I’m thinking. But without language, these linguistically structured thoughts might be impossible. However, some philosophers of language, notably one called Jerry Fodor, at Rutgers, believe that there is a ‘language of thought’ which has logico-syntactic structure but is composed of non-linguistic items, e.g. pure thoughts. Fodor thinks that a child learns its first language via a process of translation into this mental language, which he calls ‘mentalese’. So, perhaps if this child misses the crucial age window for language acquisition, like what happens with some Feral Children, there may nonetheless be a sophisticated, intelligently structured system of thought available to the child.

    The stronger version of Sapir Whorf is that our particular language – not just our general language faculty – determines our possible thoughts. This is, in the view of most, probably false.

    Obviously if you have no word for something you might think of it less. But maybe you have no word for it because you think of it less. ‘Eskimos have seven words for snow’ goes the urban legend. Maybe English speakers would too if we had more frequent exposure to snow. But from the fact that I don’t mark the conceptual distinctions Eskimos do with my language, it doesn’t follow that I cannot grasp those distinctions when they are put before me. The fact that semantically-faithful translation from one language to another is possible threatens the idea of linguistic relativity. If you took a piraha child and raised it in France with French parents, it would acquire French. Of this I have little doubt. In fact, I’ve just been told by a friend who studies linguistics full time that the latest generation of Piraha children have been provided with the opportunity to learn Portuguese and Maths and they’re doing fine. It may be that Piraha is not a language like others, and when the vital window for normal-language acquisition has passed, they cannot then master certain ideas. But if this is true, then it may be more grist to Chomsky’s mill. It may be that the universal faculty they possess as humans, and which underlies normal languages, simply hasn’t been developed. It may then be incorrect to say that what we mean by ‘language’ is not the same in the case of the Piraha ‘language’ as what we normally mean. But I’m skeptical of this.

    • Josh says:

      I think you make some good points.

      Sure, the Piraha can’t recall which pile had ten versus eleven stones. But this isn’t because their language lacks words for ten and eleven. Rather, they don’t use numbers much, and consequently, their language lacks words for ten and eleven. Their inability to tell the piles of stones apart is the primary phenomenon, not a secondary phenomenon.

      Still, I think the idea of language as a placeholder is somewhat compelling. After all, it’s NOT the case that they couldn’t tell ten stones from eleven. Rather, they couldn’t REMEMBER which pile had a different amount of stones. So, the argument is not that their (lack of) language shapes their thought, but rather that it impairs their memory. This might not line up with Sapir and Whorf’s ideas, but I think the point should be made nonetheless. I think a scenario in which the quick brown fox jumps over the lazy dog would be a bit harder to remember if we didn’t have words for the, quick, brown, fox, jumps, over, the, lazy, and dog.

      It is true that this might all be grist to Chomsky’s mill. The two theories aren’t mutually exclusive, though. It’s possible that language, which helps memory, is a universal phenomenon. But without language, we have decreased memory.

  3. eupf says:

    would you mind explaining why we should assume that indo-euro is the ‘root of human language’, to paraphrase?

    • Josh says:

      Dear Eupf,

      Thanks for your comment. I’m not sure if you got around to reading my footnotes, but I mentioned there that, as you imply, calling Indo-European the root of all language would be false. Quite a few languages emerged in isolation, and even Indo-European is probably just one branch of a much larger root language, which itself may or may not merit the description “mother tongue”.

      From my footnote:

      Secondly, calling Indo-European its own tree might be misleading as well. I do it for simplicity’s sake, but as one can see from the picture, Indo-European is portrayed as just a small offshoot of the massive language system beginning with the Mother Tongue. I don’t completely agree with the picture, though, either. While it is true that Indo-European is just a small part of the tree and many languages came before it, the assertion that all language came from a single mother tongue is highly debatable.

      • eupf horia says:

        Thanks for clarifying. As an anth student this kind of stuff is fascinating to me, especially the concept of a, or possibly multiple, mother tongue(s)

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s