STA 290 Seminar: Stefan Gries (UC Santa Barbara)

Joint Statistics – Linguistics Seminar

DAY 1

This seminar was hosted by the Dept of Linguistics. The second seminar was on Thursday April 10 in MSB 1147 - click here for the abstract.

Wednesday, April 9th 2014 at 4:10pm, 53A OLSON HALL

Speaker: Stefan Gries, UC Santa Barbara

Title: "On The Need For (Improving) Quantitative Corpus Methods”

Abstract: Over the last 15 years or so, statistics have become more and more widespread in most areas of linguistics. However, not only is this development not completely uncontroversial - there are still many linguists who are more or less explicitly against the use of statistical tools - but there are also still many ways in which the field has to evolve. In this talk, I will first exemplify the need for more rigorous quantitative analysis by discussing problems of purely qualitative analysis and the results of a rather unfair experiment involving native-speaker linguists. Then, I will discuss several applications of different statistical methods in L1/L2 acquisition/learning contexts.

Specifically, I will discuss:

  • an example involving the detection of temporal stages in Korean L1 acquisition data;
  • an example of exploring non-linear trends in Russian L1 acquisition data;
  • two examples of using mixed-effects models in English L1 acquisition and Japanese L2 learning data;
with an eye to highlight the benefits of more advanced statistical methods in linguistics.