Douglas Rohde dr@tedlab.mit.edu

"Hi, my name's Madeline, but you can call me Maddie"

I received my PhD in 2002 from the School of Computer Science at Carnegie Mellon University and the Center for the Neural Basis of Cognition. I am now a postdoc in Ted Gibson's lab in the Brain and Cognitive Sciences department at MIT.

My research primarily focuses on computational models of language processing. The main questions I seek to address are how adults comprehend and produce sentences and how children learn these abilities. More specific issues I am concerned with involve the representation of word meanings, how syntax can be represented in a distributed connectionist network, how semantics and syntax interact, and the effects of delayed learning or interference from a primary language in second-language learning.

Most of my work involves developing recurrent connectionist, or neural network, models of language processing. One of the main reasons such models are interesting and relevant to language acquisition is that they are often able to learn complex and abstract behaviors without the need to rely on large amounts of structured, innate knowledge. The most recent effort along these lines, the CSCP model, is described in my PhD thesis.

In addition to the modeling, I am also involved in empirical investigations of sentence processing. One major question I am working on is the extent to which we are sensitive to the actual syntax when reading or hearing a complex sentence versus our expectations of probable messages. Along with Ted Gibson and several others, I am also beginning a study of adolescent sentence comprehension, to better understand the difficulties associated with the transition from spoken to written language use.

My other major academic interest involves computational modeling of a very different sort. I have been developing large-scale models of human mating and migration patterns to investigate human ancestry. The models suggest that the most recent ancestor shared by everyone alive today may have lived just a few thousand years ago.

Projects:

the light, efficient network simulator		a platform for sentence processing experiments
COALS the correlated occurrence analogue to lexical semantics		TGrep2 the next-generation parse tree searcher
My PhD Thesis a real page turner		SLG the simple language generator
the professional MP3 manager		My Photo Album it's searchable...and a cool shade of blue!
Boxcar Blockade I think I can...I think I can...		Death of a Salesman a minute to learn....an afternoon to master

Publications:

• Rohde, D.L.T., Olson, S., and Chang, J.T. (2004). Modelling the recent common ancestry of all living humans. Nature, 431, 562-566.
• Rohde, D.L.T. (submitted). On the common ancestors of all living humans. Submitted to American Journal of Physical Anthropology.
• Rohde, D.L.T. (in preparation). Improving the value of ratings by factoring out the generosity of raters. In preparation.
• Rohde, D.L.T. (2003). Assessing first-pass comprehension of relative clause sentences. Poster presented at the 2003 CUNY Conference on Human Sentence Processing.
• Rohde, D.L.T. (2003). The on-line processing of active and passive structures in English. Poster presented at the 2003 CUNY Conference on Human Sentence Processing.
• Rohde, D.L.T., & Plaut, D.C. (2003). Less is less in language acquisition. In P. Quinlan (Ed.), Connectionist modelling of cognitive development (pp. 189-231). Hove, UK: Psychology Press.
• Rohde, D.L.T, & Plaut, D.C. (2003). Connectionist models of language processing. Cognitive Studies, Japan, 10(1), 10-28. Draft:
• Rohde, D.L.T. (2002). A Connectionist Model of Sentence Comprehension and Production. Unpublished PhD thesis, School of Computer Science, Carnegie Mellon University, Pittsburgh, PA.
• Rohde, D.L.T. (2002). Methods for binary multidimensional scaling. Neural Computation, 14 (5), 1195-1232. Short: Long:
• Rohde, D.L.T. (1999). A Connectionist Model of Sentence Comprehension and Production. Unpublished PhD thesis proposal, School of Computer Science, Carnegie Mellon University, Pittsburgh, PA.
• Rohde, D.L.T. & Plaut, D.C. (1999). Language acquisition in the absence of explicit negative evidence: How important is starting small? Cognition, 72, 67-109.
• Rohde, D.L.T. (1999). LENS: The light, efficient network simulator. Technical Report CMU-CS-99-164, Carnegie Mellon University, Department of Computer Science, Pittsburgh, PA.
• Rohde, D.L.T. (1999). The Simple Language Generator: Encoding complex languages with simple grammars. Technical Report CMU-CS-99-123, Carnegie Mellon University, Department of Computer Science, Pittsburgh, PA.
• Rohde, D.L.T. & Plaut, D.C. (1997). Simple recurrent networks and natural language: How important is starting small? In Proceedings of the 19th Annual Conference of the Cognitive Science Society, 656-661, Hillsdale, NJ. Erlbaum.
• Rohde, D.L.T. (1995). Modeling the dual-pathway system for practice-related verbal associative learning. Unpublished undergraduate thesis, Computer Science Department, Princeton University, Princeton, NJ.

  Curriculum Vitae

Finding Me:

• Email: dr@tedlab.mit.edu
• Phone: (914) 960-0779
• Office: MIT, NE20-437E, 3 Cambridge Center, Cambridge, MA 02139
• Home: 139 6th St., Pelham, NY 10803

Useful Stuff:

• SplitIt: A program to help you quickly split sound files into pieces, merge them, and normalize the volume.
• The Spindle Model: A method of obtaining more accurate reviews by removing reviewer bias.
• DRUtils, my general C library which includes many useful low-level shortcuts.
• SVDLIBC, a C library for computing a matrix singular value decomposition and for converting matrix file formats.
• Pre-compiled UNIX|STAT binaries and modified source code.
• The Temporal Difference Conditioning Simulator
• An Introduction to Using GDB Under Emacs
• Analysis of the National Medical Residency Matching Program

Useless Stuff:

• My Web Calculator
• Recent restorations on Michelangelo's "David" lead to a shocking discovery!!
• The journal from my 1996 cross-country trip with Santosh Vempala
• A view of my parents' house from space
• Drug Lord, a computer game based on Dope Wars.