Patrick Littell

Dr. Littell currently concentrates on practical technologies for Indigenous languages spoken in Canada -- in particularly, seeking to dramatically lower the data and expertise prerequisites necessary for their development. He has a particular interest in "zero- and few-shot" language technologies, machine learning systems that can process language data without having seen data in that language before.  The ultimate goal is that communities can develop language technologies for themselves, without the need for Big Data-scale resources in their languages, and without necessarily having a rare dual expertise in linguistics and computer science.

Speech-enhanced educational tools

There is growing interest among Indigenous young people and their parents in the availability of Indigenous language education and materials.  However, there are few educational materials available for most languages, and there are even fewer that include spoken recordings, due to the greater expense in making and distributing these.  Dr. Littell is involved in several projects to leverage speech technologies to make their creation easier and faster:

• The ReadAlong Studio project automatically performs time-alignment between audio recordings and text transcriptions, allowing the easy creation of "read-along" and "sing-along" audiobooks for literacy education.  
• The Speech Generation for Indigenous Language Education project seeks to make "text-to-speech" systems for several Indigenous languages spoken in Canada, with the goal of enhancing educational materials for which no audio recordings exist.

Linguistic programming for non-experts

Due to the relative lack of large text corpora for Indigenous languages, most text processing tools for these languages are still rule-based and written by hand.  In practice, however, writing and maintaining these tools requires substantial expertise in both computer science and linguistics.  Dr. Littell's team is developing a new, more intuitive spreadsheet-based programming language and development environment, intended to be more accessible to teachers and subject-matter experts who do not have degrees in computer science, allowing them to create and maintain language tools without necessarily relying on a third-party programmer.