Associate Professor- Linguistics Program
Taraka defended his PhD thesis in Computational Historical Linguistics from University of Gothenburg under the supervision of Lars Borin and Søren Wichmann. In his thesis, he investigated hypothesis that the phoneme inventory sizes of languages reduce as one moves away from Africa, time-depth of language families, cognate identification, creation of lexical database for more than 300 South Asian languages.
During his PhD, he worked as a research assistant at the Digital Areal Linguistics project headed by Anju Saxena and hosted at University of Uppsala. He was involved in investigating the branching structure of Dravidian languages using Maximum Parsimony.
Taraka was a Post-Doctoral fellow in EVOLAEMP headed by Gerhard Jäger from November 2015 till August 2017 at University of Tübingen. During his stay at Tübingen, he worked in automatic cognate identification (neural networks, Online PMI), dialect classification, discriminating between similar languages, native language identification, and dating of Indo-European language family. He applied Siamese ConvNets for cognate identification using articulatory embeddings.
He served as a reviewer at EACL, ACL, COLING, EMNLP, Information Processing and Management, Language Dynamics and Change, PeerJ Computer Science.
Research Interests: Computational Historical Linguistics, Treebanking for South Asian languages, Medical Informatics, and Bayesian modeling.
List of Publications
Phone: (940) 565-4552
Office: E296C Discovery Park