Northwestern University

May
4
Fri 3:00 PM

Dissertation Defense: Jeremy Needle - "Gradient typicality and indexical associations in morphology"

When: Friday, May 4, 2018
3:00 PM - 5:30 PM  

Where: Cresap Laboratory, 101, 2021 Sheridan Road, Evanston, IL 60208 map it

Audience: Faculty/Staff - Student - Public - Post Docs/Docs - Graduate Students

Contact: Talant Abdykairov   847.467.3384

Group: Linguistics Department

Category: Academic

Description:

One of the main ways that a language changes is through the creation of new words, which spread socially from person to person. We use a series of experiments to investigate how people gain new words, and what factors influence that process. Critically, we find that people can break new words down into meaningful parts, and their impressions of those words can be influenced by their gendered experiences with similar known words. At base, people are affected by the many things they already know about words: sounds and spellings, meanings, rarity, context, and more. We show that people are sensitive to similar effects when dealing with new words and old ones: they judge overall similarity, they find meaningful parts, and they are affected by indexical knowledge (i.e., social and contextual experience) about similar words. Two sets of experimental judgments of typicality for real words and novel words are used to replicate gradient typicality patterns for novel words, and extend gradient typicality to real words. In addition to well-known factors of typicality (phonotactic probability, neighborhood density), we show that novel words are judged as better when they contain apparent morphemes. Two additional experiments gather explicit morphological decompositions and gender association judgments for real and novel words. We find that people are highly accurate at decomposition for real and novel words, and argue that they achieve this by spotting known morphemes. Gendered associations for real words coincide with corpus statistics taking into account author gender, and novel words are associated with the gender bias of the real morphemes they contain. Our findings both constrain and augment models of the mental lexicon, the processing of words, the social use of language, and language change.

Add Event to Calendar

Add Event To My Group:

Please sign-in