Endangered Language Documentation: Bootstrapping a Chatino Speech Corpus, Forced Aligner, ASR

Endangered Language Documentation: Bootstrapping a Chatino Speech Corpus, Forced Aligner, ASR Małgorzata E. ´Cavar, Damir Cavar, Hilaria Cruz Indiana University, University of Kentucky Bloomington, IN, Lexington, KY {mcavar,dcavar}@indiana.edu, hcr224@g.uky.edu Abstract This project approaches the problem of language documentation and revitalization from a rather untraditional angle. To improve and facilitate language documentation of endangered languages, we attempt to use corpus linguistic methods and speech and language technologies to reduce the time needed for transcription and annotation of audio and video language recordings. The paper demonstrates this approach on the example of the endangered and seriously under-resourced variety of Eastern Chatino (CTP). We show how initial speech corpora can be created that can facilitate the development of speech and language technologies for under-resourced languages by utilizing Forced Alignment tools to time align transcriptions. Time-aligned transcriptions can be used to train speech corpora and utilize automatic speech recognition tools for the transcription and annotation of untranscribed data. Speech technologies can be used to reduce the time and effort necessary for transcription and annotation of large collections of audio and video recordings in digital language archives, addressing the transcription bottleneck problem that most language archives and many under-documented languages are confronted with. This approach can increase the availability of language resources from low-resourced and endangered languages to speech and language technology research and development. Keywords: Chatino, Speech and Language Corpus, Speech Technology 1. Introduction Endangered language documentation includes, among oth- ers, the creation of language resources in form of word lists, audio and video recordings, notes, or grammar fragments. Once the data is collected, various institutions offer archiv- ing services for these language resources, however, irre- spective of the availability of archives, another big issue needs to be addressed. Large collections of recordings and data sets face a problem that is commonly referred to as the Transcription Bottleneck. The Transcription Bottleneck is a metaphor that is often used to describe the problem that large amounts of digital audio and video language recordings from language docu- mentation projects are only available in the original audio or video format. In other words, the recordings in digital language archives are often accompanied by some meta- information that includes, for example, information about authors, speakers, origin, or source of the material but, cru- cially, transcriptions of the audio or video material are not available or existent for most of the archived recordings, neither linguistic annotations or even translations. To transcribe the resources collected and recorded in the field or in interactions with native speakers it is necessary but not sufficient to just keep track of the meta informa- tion related to the source, time, location, or purpose of the recording, and the technology that was used. If they are not transcribed, annotated and translated the collected language resources are only accessible to the native speakers of the language or experts. If the recordings can be deciphered only by native speakers or experts, this in itself presents a problem for the low-resourced or endangered languages we work with as …

Content truncated to 500 words