Vol. 27:1 (2022) ► pp.93–125
The Sociolinguistic Speech Corpus of Chilean Spanish (COSCACH)
A socially stratified text, audio and video corpus with multiple speech styles
This paper presents the Sociolinguistic Speech Corpus of Chilean Spanish (COSCACH) v1.0, a 9.3-million-word corpus containing transcribed, lemmatized and morphologically tagged text, audio recordings and videos from 1,237 L1 speakers of Chilean Spanish, as well as a control sample of 21 non-Chilean L1 Spanish speakers. The COSCACH is the first freely available corpus of spoken Chilean Spanish of substantial size, as well as one of the largest speech corpora of any variety of Spanish. Following a review of other Chilean speech corpora, I describe how the COSCACH was constructed, covering corpus design, speaker recruitment and metadata collection, speech elicitation and recording, transcription, lemmatization and morphological tagging, and corpus compilation. I thereby aim to provide a blueprint for creating modern, large-scale speech corpora suitable for phonetic, sociophonetic and sociolinguistic research, in addition to traditional inquiry into semantics, lexis, grammar, pragmatics and discourse.
Article outline
- 1.Introduction
- 2.Other Chilean Spanish speech corpora
- 2.1ESECH and PRESEEA-SA
- 2.2King-ASR-290
- 2.3Additional speech corpora
- 2.4Justification for the COSCACH
- 3.Corpus design and speaker sampling
- 3.1Chilean speaker samples
- 3.1.1Speaker inclusion variables
- A.Locality
- B.Ethnicity
- C.Lingualism
- D.Age/Generation
- E.Sex
- F.Socioeconomic status
- G.Year of recording
- 3.1.2Derived variables
- A.Region
- B.Urbanness
- C.Locality size
- D.Distance and travel time from Santiago
- 3.1.1Speaker inclusion variables
- 3.2Non-Chilean control sample
- 3.1Chilean speaker samples
- 4.Data collection
- 4.1Timeframe
- 4.2Fieldworkers
- 4.3Speaker recruitment
- 4.3.1Recruitment procedures
- 4.3.2Informed consent and institutional review board (IRB) approval
- 4.3.3Further criteria for exclusion of speakers
- 4.4Socio-demographic questionnaire
- 4.5Elicitation tasks
- 4.5.1Sustained pronunciation of isolated vowels
- 4.5.2Reading of minimal pairs or other word lists
- 4.5.3Reading of meaningful sentences
- 4.5.4Reading of meaningful texts
- 4.5.5Conversational interview
- 4.5.6Language attitudes interview
- 4.6Recording
- 4.6.1Audio equipment and configuration
- 4.6.2Audio post-processing
- 4.6.3Video recording equipment and procedures
- 5.Transcription, text processing and corpus compilation
- 5.1Transcription
- 5.2Anonymization and protection of speakers’ privacy
- 5.3Text extraction and annotation
- 5.4Corpus compilation and use
- 6.Availability and access
- 7.Conclusions and future directions
- Acknowledgements
- Notes
-
References
https://doi.org/10.1075/ijcl.19103.sad