Using Transparent Machine Learning to study Human Speech   Machine learning, the use of nuanced computer models to analyze and predict data, has a long history in speech recognition and natural language processing, but has largely been limited to more applied, engineering tasks.  This talk will describe two more research-focused applications of transparent machine learning...

In prominent theories pronouns are treated as having (in effect) descriptive content, to explain the possibility of anaphora on indefinites outside their binding domain. We discuss some data that appears to be problematic for these approaches, and consider as an alternative, a view that treats pronouns as simple variables and tracks anaphoric dependencies separately.

Joe Pater (UMass). Faculty host: Claire Moore-CantwellZoom link will be available one week ahead of the talk.

Hannah Sande (Georgetown University). Faculty host: Ethan PooleZoom link will be available one week ahead of the talk.

Phonetic variability and natural class phonotacticsGillian Gallagher, NYUPhonological patterns are stated over classes of sounds, usually defined based on a shared phonetic property. Phonetic sound changes and phonetic variation, however, are both extremely common, and easily result in a language showing a phonological pattern over a class of segments that cannot be easily defined given the...

Jessica Coon (McGill University). Faculty host: Harold Torrence.Zoom link will be available one week before the talk. Email the Colloquium Committee at uclacolloquium@gmail.com to be added to the listserv. For guest outside of UCLA, please submit an RSVP for colloquiums you would like to attend.

Molly Babel (UBC). Faculty host: Claire Moore-Cantwell.Zoom link: https://ucla.zoom.us/j/99599993759?pwd=NkQwQnRwNkpqWGRMTGV1dG1FbUxyUT09  Email the Colloquium Committee at uclacolloquium@gmail.com to be added to the listserv. For guest outside of UCLA, please submit an RSVP for colloquiums you would like to attend.Abstract:When do listeners care about phonetic variation?Spoken language is immenselyvariable. Some of that variation may be associated with social categories andsome...

Prosodic context in speech perceptionWhen listeners process spoken language, they extract information about (1) segmental categories, defining the words intended by a speaker and (2) prosodic features, which convey information about prominence, grouping, and so on (i.e. how the words are said). These two parts of understanding spoken language are often studied as separate, however, their joint...

Zoom link: https://ucla.zoom.us/j/98929850267?pwd=Y3hUdi9RSG5ITG9UT0ZGRUFReU5hdz09Meeting ID: 989 2985 0267 Passcode: colloquium Abstract: Making comparisons and interpreting comparatives in child language Rutgers University – New BrunswickFor many years now, researchers in language and cognitive development have been plagued by children’s divergent non-adult-like productions of comparative constructions, and their apparent inability to properly interpret even the most basic of comparatives (although researchers have also...