Prerequisites: To succeed in this class you should have seen at least a bit of (constituency-based) syntax before, such as context-free grammars. Even just a unit in an introductory course is probably enough. If you're not sure, email me.
Time and Place: Tuesdays 12:00-13:30 in C7.2 seminar room U15, April 25-July 25 2017 (14 weeks)
Evaluation (this may change): 7-credit version: 50% paper or project; the rest divided up as follows: 70% presentation, 20% exercises, 5% general participation
Papers due Friday Sept 1
In this seminar we will explore minimalist syntax, from Chomsky's original formulation (The Minimalist Program) through Stabler's formalisation (Minimalist Grammars), and on to Minimalist approaches to linguistic phenomena.
This course will be a combination of lecture-style and seminar-style. Each unit will begin with a class or sometimes two of lecture and in-class excersies to get a handle on the formal material. After that students will present papers from the literature.
- What is minimalism? What are the Chomskians up to these days?
- Original formulation: a chapter of The Minimalist Program (Chomsky 1995)
- A textbook: a chapter of Understanding Minimalism (Hornstein, Nunes, and Grohmann 2005)
- just about anything from "mainstream" linguistics journals such as Linguistic Inquiry
- May 30: Minimalist Grammars: a formalisation of minimalism MGs are a formalisation of minimalist ideas developed by Ed Stabler and others
- Parsing of MGs The first parsers for MGs come from Harkema's dissertation (2001). New ones continue to be developed. Also, John Torr is developing a tree bank based on the Penn Tree Bank.
- Original parsers: a chapter from Harkema (2001) (Anastasija, July 11)
- Original CKY-like parser Stabler 2000
- Top-down beam parser: Stabler Formal paper (2011) or Linguists' paper (2012)
- Mainguy 2010
- Efficient transition-based beam parsing: Stanojevic 2016 (Joan)
- Toward an MG tree bank: John Torr, EACL 2017
- Formal language theory and minimalism Minimalism grew in part out of the proof that context-free grammars with transformations are Turing-complete. MGs are mildly context sensitive, their derivation tree language is regular, and merge-only MGs have subregular derivation tree languages.
Please read through page 9 and send me 2 questions by email by Monday night. Stabler and Keenan 2003