ThePangloss Collection is adigital library whose objective is to store and facilitate access toaudio recordings inendangered languages of the world. Developed by theLACITO centre ofCNRS inParis, the collection provides freeonline access to documents ofconnected, spontaneous speech, in otherwise little-documented languages of all continents.[1]
For the science oflinguistics, language is first and foremost spoken language. The medium of spoken language is sound. The Pangloss Collection gives access to original recordings simultaneously with transcriptions and translations, as a resource for further research. After being recorded in its cultural context, texts have been transcribed in collaboration withnative speakers.
The archived data is based on robust standards, asopen architecture, in anopen format, and may be downloaded under aCreative Commons license. The software used to prepare and disseminate it isopen-source. The Pangloss Collection is a member of the Digital Endangered Languages and Music Archive Network (DELAMAN). Its hosting platform, theCocoon repository, is one of the archives participating in theOLAC network.
The collection was initially called theLACITO Archive.[2][3] The project originated in 1996 from the collaboration of Boyd Michailovsky, linguist at LACITO, with John B. Lowe, engineer;[4]: 15 they were later joined by Michel Jacobson, engineer, who developed some tools for the project, and brought it online.[1]: 124 [4]
The purpose of the archive was “to conserve, and to make available for research, recorded and transcribed oral traditions and other linguistic materials in (mainly) unwritten languages, giving simultaneous access to sound recordings and text annotation.”[4] The earliest archived corpora in the collection were languages fromNepal, fromNew Caledonia, fromeastern Africa andFrench Guiana.[5]
The archive has grown steadily since the early 2000s,[6] incorporating corpora from various linguists, whether members of LACITO or not. In 2009, the archive had 200 recordings in 45 languages.[7] In 2014, the (newly renamed)Pangloss Collection had 1,400 recordings in 70 languages.[1]: 121
As of April 2021, the Pangloss archive contains5,038 recordings[8] in 196 languages,[9] totalling 780 hours of audio and video recordings.[6]
Languages in the Pangloss Collection includeMwotlap (Austronesian;Vanuatu),[10]Japhug (Sino-Tibetan;Southwest China),[11]Ersu (Sino-Tibetan;Southwest China),[12]Naxi (orYongning Na:Sino-Tibetan;Southwest China),[13] andCèmuhî (Austronesian;New Caledonia).[14]