This is the story of a master's thesis that could have been. Back in 2007, I was working on my qualifying paper for the linguistics Ph.D. program I was enrolled in at the University of Texas. I was learning a lot of things at that time that had me really challenging some of the core principles of nativism. I was reading empirical-minded linguists like Boersma and Sampson while helping to organize a conference focused on using empiricism in linguistics.

I was evaluating approaches like Optimality Theory for phonology and Minimalism, LFG, and a bevy more for syntax. I could really only get behind the operating concept of Combinatory Categorial Grammar. It performed exactly at mildly context-sensitive without any spurious hacks by doing away with "innate" categories, and focusing on observed interactions. I was also getting more and more into the computational side of linguistics, and applying the various data types set forth by theoreticians was problematic at best. Corpus linguistics, the best evidence we have for how speakers actually perform, seemed to really be at odds with the strict predictions of most frameworks. I encountered a lot of hegemony in what you'd consider fundamental theoretical linguistics, and that was one of the reasons I'm in software now.

During this time, I thought it would be worth writing a master's report simply to pick up an M.A. during the course of my Ph.D. work. This thesis combined several of my conference papers and observations from other empirical linguists into a report on the state of empirical linguistics, and the possibility that frequency played a role in historical change in a variety of facets. I could find examples where frequency promoted changes in some rules and irregularity in others in both phonology and syntax.

Due to some administrative screw-ups with a relatively reluctant graduate advisor, I was unable to get the report signed by the right people at the right deadlines. A year later, I was working on some interesting machine learning stuff that I thought would be more applicable in my future career. At that point, I had determined that academia wasn't the place for me. I cut my losses short, instead of buying into any sort of fallacy of diminishing returns, and published an M.A. that I'm proud of.

At that time, I also had signers for the other thesis -- two old-school professors who had dealt with the current state of linguistics since Chomsky's generative revolution, and were really interested in bucking the system by getting behind my research. One of these professors had actually performed manual quantitative phonology back in the '50s -- before you couldn't publish anything that didn't analyze language from the armchair rather than the field.

Anyhow, I did some "archeology" on my Mac from that era, and found the thesis. I'd like to believe I could improve the prose these days, but I think a lot of the observations are still pretty salient. I don't really follow academic linguistics anymore, but I thought I'd publish it here as a piece of history, during a time in the discipline when we were just starting to consider treating it as a science, rather than a humanity.

Read it here: The Role of Frequency in Historical Change