More on the Hausarbeit
The Hausarbeit for this module can take one of two forms.
- An executive summary of a research paper
- A project whereby you analyze some phenomenon
I think that whichever form your HA takes, it’s length should be between 5-10 pages.
Executive Summary
The goal of an executive summary is to convince me that you have engaged with the paper. Ideally you will have understood it in detail. I would like to know
- what the main idea and goal of the paper is
- how they achieve this goal
As I mainly want to know about your understanding of the paper, it would be useful for me for you to explain difficult things in detail. If the paper is very long, you may make the executive decision to focus on a particular part of it. If you choose to do this, please give a short (~2 sentnece) justification for this decision.
Some possible papers to executively summarize
Syntax
- Goodman 99 Semiring Parsing
- Gerdemann 94 Parsing as Tree Traversal
- Resnik 92 LC Parsing and Psychological Plausibility
- Shieber Johnson 93 Variations on Incremental Interpretation
- Johnson Kay 94 Parsing and Empty Nodes
- Roark Johnson 99 Efficient Probabilistic Top-Down and Left-Corner Parsing
- Roark 01 Probabilistic top-down parsing and language modeling
- Hale 01 A probabilistic Earley parser as a processing model
- Hale 09 What a rational parser would do
- Hale 03 The Information conveyed by words in a sentence
- Hale 06 Uncertainty about the rest of the sentence
Phonologie
- Chandlee Heinz 18 Strict locality and phonological maps
- Heinz 09 On the role of locality in learning stress patterns
- Heinz 10 Learning long distance phonotactics
- Kaplan Kay 94 Regular Models of phonological rule systems
- Ellison 94 Phonological Derivation in OT
- Bird Ellison 94 One level phonology
- Karttunen 91 Finite state constraints
Morphologie
- Chandlee 17 Computational locality in morphological maps
- Karttunen 03 Computing with realizational morphology
- Goldsmith 01 Unsupervised learning of the morphology of a natural language
- Koskenniemi 84 A general computational model for word-form recognition and production
A project
Finite state transducers can implement virtually all phonological and morphological processes. CFGs can implement a great deal of syntax. Find some natural language phenomenon and analyse it with the tools we have learned about in this course.
Some examples of this sort of project (here and here) can be found on the web page for last year’s version of this course. These are a little shorter than what I would like for this year.