Quirk Fellow 2021 - Workshops delivered by distinguished Prof. Janet Dean Fodor
When: Wednesday, May 19, 2021, 4:00 PM - 6:00 PM
Where: Online: https://qmul-ac-uk.zoom.us/j/88131111509?pwd=TGR6Y2MybHpPRVNacEw5MnNDRmxpQT09
Speaker: Prof. Janet Dean Fodor
Our department extends a warm welcome to our 2021 Randolph Quirk Fellow, Distinguished Professor Janet Dean Fodor (CUNY Graduate Center). Next to a public lecture, Professor Fodor will deliver a series of workshops.
The workshops will be accessed through this Zoom link.
Monday 17 May, 16:00-18:00 BST
Sentence parsing—universals and differences
A sentence parsing system has two main tasks: (1) combine words into a syntactic tree structure, and (2) deduce the deep structure, which will feed semantic interpretation.
For step 2, Transformational Grammar is an implementation nightmare. Early attempts to ‘de-transform’ a surface string were immensely cumbersome. Transformational rules don’t work well in reverse (surface to deep), especially for a left-to-right on-line parsing system. More practical is an enriched phrase structure grammar, which delivers a single tree structure enriched with notations of ‘fillers’ and ‘gaps’. I propose that this is better linguistics as well as better psycholinguistics. And I add a plea to future researchers: How can linguists help grammars help parsers?
Tuesday 18 May, 16:00-18:00 BST
Prosody is the rhythm and melody of language. It is present in all language use: in speaking, in reading aloud, in silent reading, even in verbal thought. (A mentally rehearsed shopping list has prosody!) At CUNY we have studied the role of prosodic cues in disambiguating the meanings of sentences, spoken or written. We have designed experiments which can reveal effects of silent prosody in reading. We conclude that some ‘syntactic parsing strategies’ previously proposed are really just the prosodic component of the grammar at work.
Wednesday 19 May, 16:00-18:00 BST
Language learnability: How do learners evaluate candidate grammars?
Following Chomsky (1965), models of child language acquisition traditionally rely on an innate evaluation measure (EM). EM guides learners’ selection of a mental grammar which generates the adult language they are exposed to. At CUNY we have proposed that EM is part and parcel of a learner’s growing ability to parse and comprehend the sentences encountered in everyday life. This approach is modelled with units of cross-language variation (parameter values) which are UG-provided tree fragments that a child may incorporate into the innate sentence processing system, when needed to accommodate newly encountered constructions in the ambient adult language.