Phonemic Transcription of Low-Resource Tonal Languages

Phonemic Transcription of Low-Resource Tonal Languages Oliver Adams,♠ Trevor Cohn,♠ Graham Neubig,♥ Alexis Michaud♣ ♠Computing and Information Systems, The University of Melbourne, Australia ♥Language Technologies Institute, Carnegie Mellon University, USA ♣CNRS-LACITO, National Center for Scientific Research, France ������������������������������ ��������������������� ������������������� ���������������������� Abstract Transcription of speech is an important part of language documentation, and yet speech recognition technology has not been widely harnessed to aid linguists. We explore the use of a neural network ar- chitecture with the connectionist tempo- ral classification loss function for phone- mic and tonal transcription in a language documentation setting. In this framework, we explore jointly modelling phonemes and tones versus modelling them sepa- rately, and assess the importance of pitch information versus phonemic context for tonal prediction. Experiments on two tonal languages, Yongning Na and East- ern Chatino, show the changes in recogni- tion performance as training data is scaled from 10 minutes to 150 minutes. We dis- cuss the findings from incorporating this technology into the linguistic workflow for documenting Yongning Na, which show the method’s promise in improving effi- ciency, minimizing typographical errors, and maintaining the transcription’s faith- fulness to the acoustic signal, while high- lighting phonetic and phonemic facts for linguistic consideration. 1 Introduction Language documentation involves eliciting speech from native speakers, and transcription of these rich cultural and linguistic resources is an integral part of the language documentation process. How- ever, transcription is very slow: it often takes a lin- guist between 30 minutes to 2 hours to transcribe and translate 1 minute of speech, depending on the transcriber’s familiarity with the language and the difficulty of the content. This is a bottleneck in the standard documentary linguistics workflow: lin- guists accumulate considerable amounts of speech, but do not transcribe and translate it all, and there is a risk that untranscribed recordings could end up as “data graveyards” (Himmelmann, 2006, 4,12-13). There is clearly a need for “devising better ways for linguists to do their work” (Thieberger, 2016, 92). There has been work on low-resource speech recognition (Besacier et al., 2014), with ap- proaches using cross-lingual information for better acoustic modelling (Burget et al., 2010; Vu et al., 2014; Xu et al., 2016; Müller et al., 2017) and lan- guage modelling (Xu and Fung, 2013). However, speech recognition technology has largely been in- effective for endangered languages since architec- tures based on hidden Markov models (HMMs), which generate orthographic transcriptions, re- quire a large pronunciation lexicon and a language model trained on text. These speech recognition systems are usually trained on a variety of speak- ers and hundreds of hours of data (Hinton et al., 2012, 92), with the goal of generalisation to new speakers. Since large amounts of text are used for language model training, such systems often do not incorporate pitch information for speech recog- nition of tonal languages (Metze et al., 2013), as they can instead rely on contextual information for tonal disambiguation via the language model (Le and Besacier, 2009; Feng et al., 2012). In contrast, language …

Content truncated to 500 words