Saturday, May 09, 2009

Research Ideas

(OK, this is probably a post mostly for me as I work things out, but, hey, I'm going to write it anyway. Really, feel free to ignore it because I end up almost talking to myself. Still, my mom might be interested to see the kind of stuff I get up to.)

As always at the beginning of a semester, I think of accomplishing amazing things, far too much for any human to do. The Spring semester is ending and I am neither teaching nor taking classes for the first time ever. This means it's research time. What to do?

1) The obvious #1 thing is continuing my dissertation experiments. Everything takes a backseat to this. However, I spend most of my time for the dissertation waiting for another person to participate in the experiment. This lets me do other stuff in the meantime.

2) Revise Korean apology paper. We got a revise and resubmit response from the journal we submitted to. So we are now revising and resubmitting. This also ties directly into the conference presentation I am making in Melbourne in July, so it's got to happen.

3) On the same theme, we have a second poster for the Melbourne conference about apologies in English and Korean (different topic in the same area). Got to do this one, too.

4) I'm working on a theoretical / philosophical paper about what can and cannot be learned. I wrote several pages of it, and it's a new version of my old "working paper" which is a doctoral requirement. Be good to finish it, but I'm still working through some of the logic.

5) In my working paper, I also came up with a new version of how pronouns work in language. However, I came up with it not because I believed it was the ideal way to understand pronouns, but to prove a philosophical point that it's always possible to come up with reasonable alternate explanations for things. So I just ignored the idea as soon as the paper was complete. Lately, though, I've been wondering if the ideas were really that throw-away. Perhaps they could be developed further and be worth something. (Note to self: would need to build in scope and quantification ideas; also good to test learnability on it.)

6) There were several class projects this semester that could be worth following. First up, we looked at sentences like, "Mr. Bear found a blue fish and Mr. Tiger did, too." For English-speaking adults and kids, this sentence is not ambiguous at all. It's true if Mr. Tiger also found a blue fish, but false if Mr. Tiger found a red fish. There are grammatically similar but different sentences in Korean and Japanese, and based on previous research Japanese children behave very differently, i.e., they will label the red fish possibility as true. My students designed a version of this experiment and ran them with 3 English-speaking children, 1 Korean-speaking child, and 1 Japanese-speaking child. There are some problems with the stories they used, and the Japanese and Korean infants all displayed the famous "yes bias" of child research, which is when the child just says yes to anything. Story: "Elmo had ice cream at the party." Test sentence: "Elmo had cake at the party." Child answer: "Yes." Whatever you say, they say yes. Anyway, I'd like to fix and try this experiment again, but to do it, I need Japanese and Korean speakers. I had them in my class, but my class is no more.

7) Specific Language Impairment is a clinically diagnosed condition in which, by definition, a person tests out as completely normal on all cognitive, hearing, visual, and other tasks, and yet has problems specifically with language (hence the name). It occurs in various forms, in about 5% of the population, and is more common with boys than girls. The earliest we can currently diagnose SLI is at about 3-4 years of age. Speech pathologists are able to help children with SLI to some degree, but there might be limitations at this point. It might be "too late" to really change things. Another student group worked on designing a test for SLI that we could run at 1 year. This project has huge upside, as it could one day help real people and not just be a bit of scientific knowledge, like most of what linguistics does. However, it's also a huge project, probably needing 200 children and about 3.5 years of testing and retesting. This means grants, hiring someone, and more, and would be most appropriate if I'm settled somewhere. There are some preliminary tests however I could start now about how well young children hear verb inflection (kick versus kicked versus kicks). SLI children seem to have particular trouble with verbs. It would be good to start these warm-ups now, which would make serious NSF funding more likely in the future. Funding makes me hirable.

8) Another class project has to do with how children break actions apart into their components. This was started, needs to be fixed, and could be run. The nice thing about this experiment is that it's really quite easy. One could get it going in a few weeks. It's not as important as some of the others, but still. THE MAIN barrier between me and a job is publications. I needs me some.

9) There's a prof in computer science, whom I will start calling Smackdown, that I took a class from who once offered to be my dissertation chair. That really meant a lot to me at the time for reasons I won't put on this public blog. Her being my chair wasn't possible for various reasons, but I really don't want to abandon this connection. I would like to apply some of her machine learning models to language data before it's too late. But I have to find an idea and get her time.

10) I designed an experiment two semesters ago looking at the influence of rhythm on how infants find words in speech. The pros for this experiment are that it's right in my area of specialization where I'd like to get a job, and ties immediately into my dissertation. However, there are some technically challenging components of the experiment, like getting various items perfectly synchronized, that make it a bit of stretch. Moreover, recruiting young infants, like 9 months old, can be difficult and requires installing myself in one of the labs.

To throw one more component into all of this, I've got 3 students who might be interested in helping with research. That's awesome. I don't really think all 3 will stay involved, but maybe 1 or 2 might. For them to help, I need to either pay them as research assistants or find projects that really are likely to end in a publication, which is good for their grad school records. (They are mostly graduating and moving into masters programs in speech pathology.) Since I have no money, the latter option seems best.

OK, now to review all of this, I come to a few conclusions:

a) It's stupid to run any experiments with English-speaking adults, because that's who I need in my dissertation research.

b) Theoretical things make sense as those can neatly fill the time when sitting around hoping to get a research participant.

c) I need publications and that's the way I can "pay" my possible research compatriots, i.e., former students. So choose options that are likely to get a publication in a year, not in 3 years after five failed attempts.

d) Doing experiments in other languages or with children is logical since that doesn't pull from my dissertation pool.

So after all that, what do I think? Well, 1-3 come first. Easy one there. After that? Hmmm.... Lots of variables too long to consider here. Must stew.


McKoala said...

11 Wangle side trip to Sydney to see McKoala for inaugural Minions Downunder event.

fairyhedgehog said...

You have so many interesting avenues to explore, no wonder you're having trouble narrowing it down.

I was particularly intrigued by number four (what can and can't be learned) although they all sound fascinating.