Mandarin–English code-switching speech corpus in South-East Asia: SEAME

Lyu, Dau-Cheng; Tan, Tien-Ping; Chng, Eng Siong; Li, Haizhou

doi:10.1007/s10579-015-9303-x

Cited by 76 publications

(82 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Of course, typological and comparative works exist (Lim & Gisborne, 2009;Sharma, 2009;Kortmann & Szmrecsanyi, 2011), but they too make extensive use of the concept. However, it is conceivable -and desirableto complement these qualitative approaches with quantitative data: the compilation of non-monolingual corpora (Deuchar et al, forthcoming;Lyu et al, 2010) is a first step in the right direction, and may provide us with data that may have the potential to better inform our understanding of how language variation in multilingual settings is best modelled. Szmrecsanyi, 2011) and measures of phonetic similarity (McMahon et al, 2007), but here too, the starting point is, more often than not, a conceptual linguistic system tied to a particular locale, not least because of the sampling and collection methods employed in the corpora they use.…”

Section: Discussionmentioning

confidence: 99%

The trouble with World Englishes

Leimgruber¹

2013

English Today

View full text Add to dashboard Cite

Ever since the 1980s, when research interest in the field of ‘World Englishes’ began to gather speed, the view of the English language around the world has been largely dominated by the construct of so-called ‘varieties’ of English. These varieties are usually given a geographical label (‘Singapore English’, ‘Welsh English’, ‘South African English’, ‘Fiji English’, etc), and are described in terms of their pronunciation, their grammar, and their vocabulary. The resulting anthologies (see e.g. Wells, 1982; Trudgill & Hannah, 1982; Kortmann et al., 2004) have contributed a lot to our understanding of how English varies globally, as well as to raising the profile of non-inner circle (Kachru, 1985) varieties, which had previously not benefited from as much attention.

show abstract

Section: Discussionmentioning

confidence: 99%

The trouble with World Englishes

Leimgruber¹

2013

English Today

View full text Add to dashboard Cite

show abstract

“…The South East Asia Mandarin-English (SEAME) corpus [17] was used for the following experiments. It can be simply separated into two parts by its literal language.…”

Section: Corpusmentioning

confidence: 99%

Training Code-Switching Language Model with Monolingual Data

Chuang

Sung

Lee

2020

ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

View full text Add to dashboard Cite

A lack of code-switching data complicates the training of code-switching (CS) language models. We propose an approach to train such CS language models on monolingual data only. By constraining and normalizing the output projection matrix in RNN-based language models, we bring embeddings of different languages closer to each other. Numerical and visualization results show that the proposed approaches remarkably improve the performance of CS language models trained on monolingual data. The proposed approaches are comparable or even better than training CS language models with artificially generated CS data. We additionally use unsupervised bilingual word translation to analyze whether semantically equivalent words in different languages are mapped together.

show abstract

“…Even in their data, the percentage of code-switched tweets was barely over a tenth of the total test data. There have been other corpora built, particularly for other language pairs such as Mandarin-English (Li et al, 2012;Lyu et al, 2010), but the amount of data available and the percentage of code-switching data within that data are not up to the standards of other areas of the natural language processing field. With this in mind, we sought to provide corpora for multiple language pairs, each with a better distribution of code-switching phenomena.…”

Section: Related Workmentioning

confidence: 99%

Overview for the First Shared Task on Language Identification in Code-Switched Data

Solorio¹,

Blair²,

Maharjan³

et al. 2014

Proceedings of the First Workshop on Computational Approaches to Code Switching

152

191

View full text Add to dashboard Cite

We present an overview of the first shared task on language identification on codeswitched data.The shared task included code-switched data from four language pairs: Modern Standard ArabicDialectal Arabic (MSA-DA), MandarinEnglish (MAN-EN), Nepali-English (NEP-EN), and Spanish-English (SPA-EN). A total of seven teams participated in the task and submitted 42 system runs. The evaluation showed that language identification at the token level is more difficult when the languages present are closely related, as in the case of MSA-DA, where the prediction performance was the lowest among all language pairs. In contrast, the language pairs with the higest F-measure where SPA-EN and NEP-EN. The task made evident that language identification in code-switched data is still far from solved and warrants further research.

show abstract

Mandarin–English code-switching speech corpus in South-East Asia: SEAME

Cited by 76 publications

References 25 publications

The trouble with World Englishes

The trouble with World Englishes

Training Code-Switching Language Model with Monolingual Data

Overview for the First Shared Task on Language Identification in Code-Switched Data

Contact Info

Product

Resources

About