🧪 Test?View on arXiv
CamSpokenNLI: A Pilot Resource for Cross-Variety Transfer Evaluation in Cameroon Pidgin English and Camfranglais
Not specified in the provided content
NLIspeech recognitioncross-variety transferdataset creation
2611.17573
Builder Relevance
19h ago60%
Abstract
The paper introduces CamSpokenNLI, a pilot dataset for spoken Natural Language Inference in Cameroon Pidgin English and Camfranglais.
Reality Card
Core Claim
The dataset provides a foundational resource for evaluating spoken NLI in two underrepresented varieties of English in Cameroon, highlighting the need for disjoint-item evaluation.
Method / Result
XLM-RoBERTa baselines achieved 44.7% accuracy on Pidgin and 46.2% on Camfranglais.
Limitations
The dataset is exploratory and lacks maturity as a benchmark, with significant concerns regarding hypothesis wording affecting accuracy.
Paper to code
Verified implementation resources so builders can test the paper’s claims instead of stopping at the abstract.
No verified implementation link has been attached yet. AIBuzzHub will keep this panel separate from unverified search results.