Clearing the Transcription Hurdle in Dialect Corpus Building: The Corpus of Southern Dutch Dialects as Case Study

Abstract

This paper discusses how the transcription hurdle in dialect corpus building can be cleared. While corpus analysis has strongly gained in popularity in linguistic research, dialect corpora are still relatively scarce. This scarcity can be attributed to several factors, one of which is the challenging nature of transcribing dialects, given a lack of both orthographic norms for many dialects and speech technological tools trained on dialect data. This paper addresses the questions (i) how dialects can be transcribed efficiently and (ii) whether speech technological tools can lighten the transcription work. These questions are tackled using the Southern Dutch dialects (SDDs) as case study, for which the usefulness of automatic speech recognition (ASR), respeaking, and forced alignment is considered. Tests with these tools indicate that dialects still constitute a major speech technological challenge. In the case of the SDDs, the decision was made to use speech technology only for the word-level segmentation of the audio files, as the transcription itself could not be sped up by ASR tools. The discussion does however indicate that the usefulness of ASR and other related tools for a dialect corpus project is strongly determined by the sound quality of the dialect recordings, the availability of statistical dialect-specific models, the degree of linguistic differentiation between the dialects and the standard language, and the goals the transcripts have to serve.

Computational sociophonetics using automatic speech recognition

“…Tatman (2017) documents lower ASR performance for women, as well as speakers of non-American dialects like New Zealand English. Ghyselen et al (2020) found that commercial ASR had a 7% word error rate for Standard Dutch, versus 66% for the southern West Flemish dialect. Koenecke et al ( 2020) compared industrial ASR systems and found a 35% word error rate for African American English compared to 19% for white speakers of American English.…”

“…Ghyselen et al. (2020) found that commercial ASR had a 7% word error rate for Standard Dutch, versus 66% for the southern West Flemish dialect. Koenecke et al.…”

Digital Worlds and Our Folding Realities: Implications for Qualitative Inquiry

“…Due to ethical concerns, it is not always feasible for researchers to use the latter option. Even though some transcription applications achieve the desired level of accuracy in transcribing standard major languages such as English, their capacity in transcribing dialects of different languages varies significantly (e.g., Ghyselen et al, 2020;MacKenzie & Turton, 2020). Indeed, it is not hard to recognize the uneven progress that has been made toward embracing linguistic diversity.…”

A Computational Look at Oral History Archives

“…Usually, these interviews are accompanied by field notes, interview questions, questionnaires, reports, and other materials. It is desirable to maintain them in a format where in a given point of an audio or video file the correlated notes, materials, and metadata are made available to the researcher [39]. The sheer size of the archive makes modern computer-based search and retrieval applications and interactive interfaces designed for multimedia retrieval very relevant.…”