Professor
Department of Linguistics

Affiliate Professor
Institute for Advanced Computational Science

Stony Brook University
Stony Brook, NY 11794-4376

Email (remove underscores)
mail_@_thomasgraf_._net
Public key for signing and encryption

My research

As a linguist, I study the patterns that occur in languages around the world and seek to characterize what makes them special. For example, the English noun denaturalization is assembled from five components: de + nature + al + ize + ation. There are 120 ways these elements could be arranged, but English allows only the one above. The other 119 patterns would include words like naturdeizalation and deationalnaturize, and all of them are clearly incorrect. Languages differ in what orders they allow, but not everything goes: for instance, no language requires the parts of a word to be ordered by their length. Linguists try to construct a scientific theory of which patterns are possible, why only these patterns occur across languages, and how humans learn to produce the right patterns.

My primary research area is computational linguistics. Computational linguistics means many things to many people. Some consider it computer-assisted linguistics, others a branch of engineering that tries to teach computers to communicate with humans. I have a much more general view: computational linguistics is any kind of linguistics that uses mathematics to study the mental computations that underpin humans’ ability to use language. Mathematics is a wonderful tool to do scientific research, and all the advantages it has for physics, chemistry and biology can also be harnessed for linguistics.

My main goal is to evaluate the complexity of language from a mathematically and computationally informed perspective. What kind of memory architecture and inference rules are required to produce only correct patterns? Can computational limitations of the human brain be invoked to explain why certain patterns never occur in any languages? And can we exploit the limited nature of certain patterns to make them easier to handle for computers? If you’d like to know more, check out the overview of research projects or the list of topics to the left (on a desktop) or the bottom (on a mobile device).