The LLab at Carnegie Mellon University meets weekly to discuss issues in human language and computation.
Prepare a corpus of East Tusom wordlist and text data for release, with phonetic and phonemic transcriptions and appropriate metadata. Produce a sketch of the phonetics and phonology of East Tusom (Tangkhulic; Tibeto-Burman) based on ibid.
Attempt to leverage neural-based models with dual sources (transcription and translation) to create a hard-to-obtain gloss from an easy-to-obtain parallel corpus.
Contextualize a word translation task; fill the gap between outdated dictionary-based word translation tasks and update contextualized word embedding evaluation.