Spoken Language ‘Grammatical Error Correction’

Lu, Yiting; Gales, Mark J. F.; Wang, Yu

doi:10.21437/interspeech.2020-1852

“…These models operate non-incrementally using whole sentences as inputs, often with a view to remove the disfluencies from transcripts all together. This is also the case for research on disfluency detection in learner speech, which has been applied to improve the downstream tasks of grammatical error detection and correction using bi-directional LSTMs (Lu et al, 2019) as well as end-to-end mod-els (Lu et al, 2020). Approached as a sequence labelling task, disfluencies are flattened and models are trained to detect the reparandum phrase.…”

Section: Related Workmentioning

confidence: 99%

“…With the exception of word timings (Hough and Schlangen, 2017;Hough, 2020, 2021) the incremental approaches outlined above have yet to explore the impact of non-lexical fea-tures on disfluency detection, despite having been successfully integrated into non-incremental settings (Zayats et al, 2016;Lu et al, 2020). Considering the fact that incremental detection begins at repair onset, it seems likely that leveraging paralinguistic information associated with the interruption point will be beneficial to detection.…”

Section: Related Workmentioning

confidence: 99%

“…On average, disfluencies occur at a higher rate in learner speech compared to native speech (Hilton, 2008;De Jong et al, 2013). Learner speech disfluency datasets also contain longer reparandum phrases compared to native equivalents (Lu et al, 2020). This is in part thanks to language learners having a lower degree of 'automatisation' in the language they are learning (Temple, 1992) and is cited by Moore et al (2015) as the reason why disfluencies in learner speech are more difficult to detect automatically.…”

Section: Disfluencies In Learner Speechmentioning

confidence: 99%

See 1 more Smart Citation

Incremental Disfluency Detection for Spoken Learner English

Skidmore¹,

Moore²

2022

Proceedings of the 17th Workshop on Innovative Use of NLP for Building Educational Applications (BEA 2022)

View full text Add to dashboard Cite

Incremental disfluency detection provides a framework for computing communicative meaning from hesitations, repetitions and false starts commonly found in speech. One application of this area of research is in dialogue-based computer-assisted language learning (CALL), where detecting learners' production issues word-by-word can facilitate timely and pedagogically driven responses from an automated system. Existing research on disfluency detection in learner speech focuses on disfluency removal for subsequent downstream tasks, processing whole utterances non-incrementally. This paper instead explores the application of laughter as a feature for incremental disfluency detection and shows that when combined with silence, these features reduce the impact of learner errors on model precision as well as lead to an overall improvement of model performance. This work adds to the growing body of research incorporating laughter as a feature for dialogue processing tasks and provides further support for the application of multimodality in dialogue-based CALL systems.

show abstract

“…The data used in this paper is the same as that described in [24]. This data has been manually annotated with grammatical errors and corrections.…”

Section: Data and Experimental Setupmentioning

confidence: 99%

Ensemble Distillation Approaches for Grammatical Error Correction

Fathullah

¹

,

Gales

²

,

Malinin

³

2021

ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

View full text Add to dashboard Cite

Ensemble approaches are commonly used techniques to improving a system by combining multiple model predictions. Additionally these schemes allow the uncertainty, as well as the source of the uncertainty, to be derived for the prediction. Unfortunately these benefits come at a computational and memory cost. To address this problem ensemble distillation (EnD) and more recently ensemble distribution distillation (EnDD) have been proposed that compress the ensemble into a single model, representing either the ensemble average prediction or prediction distribution respectively. This paper examines the application of both these distillation approaches to a sequence prediction task, grammatical error correction (GEC). This is an important application area for language learning tasks as it can yield highly useful feedback to the learner. It is, however, more challenging than the standard tasks investigated for distillation as the prediction of any grammatical correction to a word will be highly dependent on both the input sequence and the generated output history for the word. The performance of both EnD and EnDD are evaluated on both publicly available GEC tasks as well as a spoken language task.

show abstract

“…The data used in this paper is the same as those described in [24]. This data has been manually annotated with grammatical errors and corrections.…”

Section: Data and Experimental Setupmentioning

confidence: 99%

Ensemble Distillation Approaches for Grammatical Error Correction

Fathullah¹,

Gales²,

Malinin³

2020

Preprint

0

View full text Add to dashboard Cite

Ensemble approaches are commonly used techniques to improving a system by combining multiple model predictions. Additionally these schemes allow the uncertainty, as well as the source of the uncertainty, to be derived for the prediction. Unfortunately these benefits come at a computational and memory cost. To address this problem ensemble distillation (EnD) and more recently ensemble distribution distillation (EnDD) have been proposed that compress the ensemble into a single model, representing either the ensemble average prediction or prediction distribution respectively. This paper examines the application of both these distillation approaches to a sequence prediction task, grammatical error correction (GEC). This is an important application area for language learning tasks as it can yield highly useful feedback to the learner. It is, however, more challenging than the standard tasks investigated for distillation as the prediction of any grammatical correction to a word will be highly dependent on both the input sequence and the generated output history for the word. The performance of both EnD and EnDD are evaluated on both publicly available GEC tasks as well as a spoken language task.

show abstract

Spoken Language ‘Grammatical Error Correction’

Cited by 6 publications

References 20 publications

Incremental Disfluency Detection for Spoken Learner English

Incremental Disfluency Detection for Spoken Learner English

Ensemble Distillation Approaches for Grammatical Error Correction

Ensemble Distillation Approaches for Grammatical Error Correction

Contact Info

Product

Resources

About