During my postdoc at the University of Wisconsin-Madison, my advisor Dr. Mark
Seidenberg suggest that looking at noun learning was "low-hanging fruit" for a
language modeler. I began looking at learning noun categories, such as the
mass/count distinction (e.g. water versus chair, substance versus object) using
categorizational neural network models. This proved more challenging than
expected, but we did make some interesting findings. However, we had no clear
human data to compare the model results to.
So, a year later while a Visiting Assistant Professor at Albion College in Michigan in 2006, my RA's and I began an experimental research project into the ability of humans to infer the meanings of novel words from the context of that word. This is well known to happen in humans, but details of the process are scarce. We started with the determiners used, plus the plural indicator (thanks to Dr. Maryellen MacDonald for that suggestion). We collected data from 20+ undergraduate subjects, but analysis would have to wait until 2010 while I was an Assistant Professor at Kutztown University. An RA ( Dale Kappus) and I began coding and analysing the data, and making some interesting findings which I am in the process of writing up in 2011
So, a year later while a Visiting Assistant Professor at Albion College in Michigan in 2006, my RA's and I began an experimental research project into the ability of humans to infer the meanings of novel words from the context of that word. This is well known to happen in humans, but details of the process are scarce. We started with the determiners used, plus the plural indicator (thanks to Dr. Maryellen MacDonald for that suggestion). We collected data from 20+ undergraduate subjects, but analysis would have to wait until 2010 while I was an Assistant Professor at Kutztown University. An RA ( Dale Kappus) and I began coding and analysing the data, and making some interesting findings which I am in the process of writing up in 2011