23.11.2008
ICLT Lecture
on
the Linguistic Utility of Tagging
Icelandic
The
next talk in the
ICLT lecture series will be given by
Joel Wallenberg from the University of Pennsylvania, Philadelphia. The
title of his talk is The
Linguistic Utility of Tagging Icelandic
and it will be given in English. The lecture will be held at room 201,
Árnargarður, University of Iceland, on Monday,
November 24th, at 12:00 noon.
The methods for part-of-speech tagging Icelandic have gone through
considerable advances in recent years, and have yielded results that
are not only interesting for language technology, but also for
linguistic theory (see for example, Helgadóttir 2004,
Loftsson 2007a,b, Rögnvaldsson &
Helgadóttir 2008). This talk follows up on this
work with the following two goals. First, I will describe
some recent work adapting the current state-of-the-art pos-tagger for
English to Icelandic (a "bidirectional sequence classification" tagger;
cf. Shen, Satta, & Joshi 2007), and some
modifications we made to the tagger to help it address the tagging of
case distinctions in Icelandic (Dredze & Wallenberg 2007a,b).
The second part of the talk focusses on the reason for my own interest
in part-of-speech tagging, which is the construction of parsed corpora
for linguistic research. I will briefly describe one result
from some recent work of mine on "Quantifier Movement" in Middle and
Early Modern English (Light & Wallenberg 2008). This
construction could not have been studied in historical English without
the help of, first, parsed corpora, and second, earlier careful studies
of modern Icelandic (Rögnvaldsson 1987,
Jónsson 1996, Svenonius 2000).
Joel
Wallenberg is a graduate student in the Department
of Linguistics at the University
of Pennsylvania, Philadelphia.
|
