My research area is natural language processing and the automatic depiction of textual descriptions as 3D scenes. In particular, I'm interested in the ways that textual meaning is resolved by context inferred from the functional and physical properties of the background actions, setting, and constituent objects.

I did my PhD research in the Language and Speech Processing Group in the Department of Computer Science at Columbia University. My thesis advisor was Prof. Julia Hirschberg. My PhD thesis (completed Jan 2017) is "Painting Pictures with Words -- From Theory to System"

I am a co-founder at WordsEye. I have previously worked on computer graphics and user interface design at Symbolics, Nichimen Graphics, and AT&T Research. At AT&T, Richard Sproat and I created the first version of WordsEye, a system for automatically depicting 3D scenes from textual input. We published a paper on WordsEye at Siggraph 2001.
Some of our more recent papers here


  • Wordseye Evaluation Corpus (with Morgan Ulinski)

  • VigNet is a lexical semantic resource developed for and used by WordsEye.
    Request from: coyne at cs dot columbia dot edu.

  • Erdnase-Sanders linguistic correspondences
  • Teaching

  • I have taught CS3102 Dev Tech (Linux) at Columbia in the Spring 2018

  • I have taught CS3101 Programming Language (Lisp) at Columbia in the Spring and Fall 2017
  • Bob Coyne
    Visualizing the
    meaning of language