Ancestor-to-Creole Transfer is Not a Walk in the Park
Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review
Standard
Ancestor-to-Creole Transfer is Not a Walk in the Park. / Lent, Heather; Bugliarello, Emanuele; Søgaard, Anders.
Proceedings of the Third Workshop on Insights from Negative Results in NLP. Association for Computational Linguistics (ACL), 2022. p. 68-74.Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review
Harvard
APA
Vancouver
Author
Bibtex
}
RIS
TY - GEN
T1 - Ancestor-to-Creole Transfer is Not a Walk in the Park
AU - Lent, Heather
AU - Bugliarello, Emanuele
AU - Søgaard, Anders
PY - 2022
Y1 - 2022
N2 - We aim to learn language models for Creole languages for which large volumes of data are not readily available, and therefore explore the potential transfer from ancestor languages (the ‘Ancestry Transfer Hypothesis’). We find that standard transfer methods do not facilitate ancestry transfer. Surprisingly, different from other non-Creole languages, a very distinct two-phase pattern emerges for Creoles: As our training losses plateau, and language models begin to overfit on their source languages, perplexity on the Creoles drop. We explore if this compression phase can lead to practically useful language models (the ‘Ancestry Bottleneck Hypothesis’), but also falsify this. Moreover, we show that Creoles even exhibit this two-phase pattern even when training on random, unrelated languages. Thus Creoles seem to be typological outliers and we speculate whether there is a link between the two observations.
AB - We aim to learn language models for Creole languages for which large volumes of data are not readily available, and therefore explore the potential transfer from ancestor languages (the ‘Ancestry Transfer Hypothesis’). We find that standard transfer methods do not facilitate ancestry transfer. Surprisingly, different from other non-Creole languages, a very distinct two-phase pattern emerges for Creoles: As our training losses plateau, and language models begin to overfit on their source languages, perplexity on the Creoles drop. We explore if this compression phase can lead to practically useful language models (the ‘Ancestry Bottleneck Hypothesis’), but also falsify this. Moreover, we show that Creoles even exhibit this two-phase pattern even when training on random, unrelated languages. Thus Creoles seem to be typological outliers and we speculate whether there is a link between the two observations.
U2 - 10.18653/v1/2022.insights-1.9
DO - 10.18653/v1/2022.insights-1.9
M3 - Article in proceedings
SP - 68
EP - 74
BT - Proceedings of the Third Workshop on Insights from Negative Results in NLP
PB - Association for Computational Linguistics (ACL)
T2 - Third Workshop on Insights from Negative Results in NLP
Y2 - 1 May 2022 through 1 May 2022
ER -
ID: 340703243