Corpus of the Pirahã language
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.

MIT Pirahã Corpus


This corpus provides a transcription and word-by-word translation of 1149 sentences of Pirahã that were collected during fieldwork by Steve Sheldon and Dan Everett. Our goal is to provide primary linguistic data from Pirahã in a human- and machine-readable format that includes annotation of the lexicon, morphology, and syntax of the language.

See the publication "A Corpus Investigation of Syntactic Embedding in Pirahã" by Richard Futrell, Laura Stearns, Daniel L. Everett, Steven T. Piantadosi, and Edward Gibson in PLOS ONE.

This corpus is distributed under a Attribution-ShareAlike (CC BY-SA) license, allowing free modification and re-distribution of the corpus so long as derivative work is released under the same terms. You may disregard any comments in the source files to the effect that citation is limited.


We welcome any contributions, including suggested changes to the core parsed data. To contribute, fork this repository, make your changes in your fork, and submit a pull request.

Data Sources

The majority of the texts are transcriptions of a single speaker describing an event. Some dialogue appears at the beginning of the Sept29amfx3 text. XigAbaI_almostbittenbysnake, Sept29amfx3, and panther were recorded, transcribed, and translated by Daniel L. Everett. All other texts were initially recorded, transcribed and translated by Steve Sheldon in the 1970s. Some of the translations, word boundaries, and sentence boundaries have been since been edited to reflect the current best translation. The original sentence translations have been preserved in the first of the two glosses given above each sentence, except for in Katosbabyandthefire, for which these first glosses reflect a 2009 translation by Daniel L. Everett.

Dan Everett texts

These stories were originally translated by Dan Everett.

Number Speaker Date Transcriber Source file
1 Xaogioso (f) 1970s Steve Sheldon 01 KATO'S BABY FALLS NEAR THE FIRE.pdf
2 Xahoápati (m) c.a. 2000 Dan Everett 02 XigAbaI almost bitten by snake.pdf
3 Kaioá (m) Sept. 29, 2009 Dan Everett 03 Sept23AMFX3.pdf
4 Xahoápati (m) July 28, 1980 Dan Everett 04 panther.pdf

Steve Sheldon texts

These texts were all collected in the 1970s and transcribed and originally translated by Steve Sheldon.

Number Speaker Source file
6 Tisahai (f) 06 APIBAI GOES, I STAY.pdf
7 Maigiphoasi (f) 07 BIRTH OF PE'S BABY #2.pdf
8 Kaboibagi (m) 08 CASIMIRO DREAMS.pdf
9 Itaibigai (f) 09 DEATH OF AOGIOSO #2.pdf
10 Baigibohoasi (f) 10 IOHABI, THE DOG, AND THE SNAKE.pdf
11 Kaboibagi (m) 11 ISAGUE AND THE ONCA.pdf
12 Tisahai (f) 12 ITAISOI HUNTS FOR TAPIR.pdf
13 Itaibigai (f) 13 MARTINS VISIT.pdf
14 Itaibigai (f) 14 MIGIXISTI DIES.pdf
15 Tisahai (f) 15 OPISI GETS A DOG.pdf
16 Tisahai (f) 16 TRIP TO SEE A PLANE.pdf
17 Tisahai (f) 17 PORTO VELHO IS BIG.pdf

Phonemic Transcription

Pirahã phonemes are /i/, /a/, /u/, /p/, /t/, /k/, /h/, /s/, /b/, /g/, and /ʔ/. In the corpus, /ʔ/ is written x and /u/ is written o.

Pirahã vowels have phonemic tone, high or low. High tones are marked with a capitalized vowel. These are based either on a diacritic marking in texts transcribed by Daniel L. Everett or tone 1 or tone 2 marker in texts transcribed by Steve Sheldon.

Low tones are marked with a lower case vowel. These are based either on a lack of a diacritic marking in texts transcribed by Daniel Everett or a lack of a tone marker or a tone 3 marker in texts transcribed by Steve Sheldon.

Otherwise, the orthography is straightforward.

Word and Sentence Boundaries

Pirahã words are given broken into morphological units, with exceptions for proper nouns, foreign words and phrases, or words whose morphological details are unknown. Word boundaries are indicated by units which are not preceded by a dash; dashes indicate a unit is an affix.

It is possible that many of the pronouns or short nouns whose definitions are marked with _clitic may not be independent words, yet are labeled as such in this corpus.

Basic Organization of the Corpus

Each text in the corpus is preceded by three lines of hashes and information about the source of the text.

The corpus is divided into stories; stories are divided into "utterances"; and utterances are divided into sentences. Utterances correspond to the sentence breaks in Steve Sheldon's original glosses.

Each utterance is preceded by two English glosses (free translations). The first is labeled with a hash and a code of two numbers, and reflects the English translation given by Steve Sheldon or Daniel L. Everett in their original translation. The second is labeled with a hash and a code of three numbers, and reflects the current best translation, as judged by Daniel L. Everett, Steve Sheldon, and the other authors. Clarifications are provided in square brackets in these glosses. The glosses allow simple text searching to find rough equivalents of English words and phrases (e.g. what does Pirahã use to convey meanings glossed as "and"?).

Numbering Scheme

Each Pirahã sentence is labeled with a unique code of three numbers, appearing on the preceding line along with its current best English translation. The first number indicates the text in which the utterance appears. The second number indicates the utterance's sequential placement within the text. This second number reflects the utterance boundaries present in the original transcriptions (e.g. by Dan Everett and Steve Sheldon). In many cases, text which was originally translated into a single utterance actually includes a group of Pirahã sentences according to our current best translations. When this occurs, the third number indicates the order of the Pirahã sentences within this grouping. Otherwise, the third number is simply 1.

Part-of-Speech Tags

Each morphological unit is in an A/B/C triple, where A is a Pirahã word, B is an English translation, and C is a part of speech tag. For instance kagi/basket/NN means that the Pirahã word kagi is best translated into "basket", a noun (NN). In the English translations, the numerals 1, 2, 3 indicate first, second, and third person pronouns respectively, and underscores are used where single English words do not suffice. In cases where morphemes are judged to be affixes, only the root is labeled with a tag.

Part-of-speech tags are assigned based on the judgment of the authors and are based on those used in the Penn Treebank. Listed below are the tags used in the Pirahã corpus, along with examples:

Tag Description Examples (Pirahã/English definition)
NN Noun kagi/expected_associate ao/foreigner s/animal_clitic xahoa/day hoaI/fire aho/leg baOsai/cloth kopaIyai/panther kaxAowI/basket
NNP Proper noun KatO/Kato TixohOI/TixohOI hoagaixoOxai/HoagaixoOxai isaitaOgi/Steve pasabIi/Passar_Bem XaogiosohoagI/Xaogioso
PRP Personal pronoun ti/1 gIxa/2 hi/3 xi/3_feminine
VB Verb igA/speak xai/do a/move ait/sleep Op/give_birth apipaO/dream
JJ Adjective bio/weak AbahIo/useless ogI/big sigi/same itA/hurt bAt/used_up
JJR Comparative adjective pixAagihI/younger
IN Postposition higIo/comitative kA/from xiigihi/near xIosi/inside
RB Adverb aigIa/thus xaI/thus pIai/also gAi/here kaxaO/away IbIg/very
RP Particle go/focus
DT Determiner kapiOxiai/another aihi/that_one aIbA/many oogiO/much oOxiai/other
CD Cardinal numeral hOi/one_or_few
WP Wh-pronoun aOi/who_or_what
UH Interjection xigIa/okay hai/hmm ko/hey a/false_start
FW Foreign word so/only agora/now quase/almost saiu_ai_morar/went_to_live_there

The tagging should be considered descriptive yet tentative. In particular, some of the part-of-speech tags used may not reflect true syntactic categories in Pirahã. For example, several tags used appear with only one lexical item: the only comparative adjective which appears is pixAagihI, translated as "younger", the only cardinal numeral which appears is hOi, meaning "few", and the only wh-prounon is aOi, meaning "who" or "what". It is possible that hOi may be better analyzed as a determiner and aOi as a noun.

Syntactic Tags

We provide a simple shallowing parsing of the corpus in order to characterize the syntactic structure of Pirahã in a broad and theory-neutral way. This parsing was completed based on the judgment of the authors. The formatting used is tgrep2 compatible.

Tag Description
Q Quote
S Sentence
NPsubj Noun phrase, subject
NPobj Noun phrase, object
NPiobj Noun phrase, indirect object
NPloc Noun phrase, locative
NPvoc Noun phrase, vocative
NPtmp Noun phrase, temporal
NPtopic* Noun phrase, topical
JJobj Adjective phrase, complement
V Verb
POS Possessive phrase
PP Postpositional phrase
PAREN Parenthetical
FRAG Fragment
* Omission [see below]
  • NPtopic may include following modifiers to indicate what role the noun appears to play in the context of the sentence: -subj -obj -iobj -loc -tmp

Omissions: An asterisk is used to indicate the omission of something which seems called for by the grammar. That is, it appears in the following cases:

  1. No subject appears before the verb
  2. No verb appears before the verbal affixes, or no verb appears in the sentence
  3. No object appears after a possessor
  4. No object appears before a postposition, such as a comitative marker

ISLRN Number