On April 10, 1969, I submitted my master’s thesis, The Development of Embeddings in the Speech of Young Children. Just now, for a somewhat complicated set of reasons, I have had occasion to re-read it. That was fine. The problem is that I can’t imagine writing it today!
How could this be? I have learned a lot in the intervening 50 years, so what’s the problem? Part of the problem is that I’ve also forgotten a lot over 50 years. In particular, I no longer have scholarly methodology at my fingertips. I no longer have the framework for analyzing linguistic data. For example, on page 30 I was examining three utterances of four-year-old Adam:
(75) Dose are bugs that I throw.
(76) It looks like a popper who pops away from people.
(77) The carrier puzzle that carries people.
What’s up with those? In the thesis I proposed the following as the deep structure of (75), but it’s no longer clear to me how I came up with it:
Or consider, as few pages later, these utterances from four-year-old Sarah:
(101) Know what I’m drawing’? A good thing I didn’t draw before.
(102) I ask her all questions she knows.
These, of course (!), are good examples of the “relative deletion” transformation.
All of these data-specific observations led me to some theoretical considerations, which are way beyond anything I think about these days. And some of my pretentious style that was necessary for thesis-writing is quite foreign to me now. For example:
It is necessary to examine the very foundations of recursion from the points of view of both linguistics and psycholinguistics if one is going to study its development in children’s speech seriously. I therefore have included a discussion of a proposal that I find very attractive: Peter Reich’s (1969) explanation for recursion, which is completely outside the framework of transformational grammar but by so being avoids some of the problems inherent in the transformational approach.
Well, that’s clear, isn’t it?
I won’t bore you with any more of the 78-page material. I just want to conclude with two observations that are not about the thesis itself:
- In order to be able to analyze a corpus of 18,000 utterances, the help of a computer was necessary. Unfortunately we had to write and run the program in Fortran, even though linguistic data clearly called for Lisp or at least Snobol, both of which we used in our courses. But those were both too expensive in terms of computer time (which was charged at the rate of $6.00 per second of CPU time on an ancient SDS-940 computer, if I remember correctly), and Harvard’s ETOIOB accounting system required that the cost came out of my advisor’s grant money. This wasn’t large enough to pay for anything but Fortran, which was fast and therefore cheap. Terribly inappropriate for linguistics, however.
- Not just the data analysis here but the entire effort was all a part of my advisor’s big project. I was lucky to be able to persuade the distinguished Roger Brown to be both my junior tutor and my senior thesis advisor. Read all about him in Wikipedia, including more info about that big project. Be sure to read the personal info near the end. I never knew about his memoir, but I’ve reserved it at the library and will read it as soon as it arrives.
Some of my students may wonder how I persuaded a distinguished professor to do this, when he had never even met me. The answer is that I had found a significant linguistic error in one of his papers, and I politely brought it to his attention in a letter (snail mail, of course). There’s a moral there somewhere.