Looking back on my career in linguistics, I regret spending so much time on theoretical syntax. Not just because Government and Binding was a crock, not just because of the theory itself nor because it was a very elaborate mechanism for handling a very small problem, reflexives, that should not be handled solely in syntax. To put it simply, you don’t need a lot of mechanism or theory if you know that a reflexive has to refer back to a subject, a topic. All of this could easily be resolved in semantics. And I don’t mean resolved in a semantics that has as its output truth values but a semantics that has actually meanings, roles, and knowledge attached to it. Syntax by itself is mostly a crock.
I figured this out some time ago. This lead me to realize that not caring about performance, how the mind actually processes and generates language, is a dead end. All language, all communication, is inherently performance. So, before I even talk about my work on Romance Language clitics or even about my work in phonology, I always point people toward my work on given/new information processing using associative memory and semantic networks. (See my Coling papers.) I regret that I didn’t recast my models, programs and even experimental results into Bernie Baars’s architecture for consciousness and cognition. That would have been interesting. I also regret not casting my research in pre-attentive vision into his architecture.
And before I even talk about my Coling papers, I point people to my work in visualization and automatic code generation. And before I talk about my work in computer science, I always point them to the work I did with Bernie. Our papers on slips of the tongue, criticizing Newell’s Soar, etc. And before I talk about that, I talk about the programs and products I created, designed and shipped.
My consolation is that my experiences have given me more to write about. More fodder for the storyteller inside of me. More color to add to the stories. It’s taken me to where I am today. And that was no small feat.