Computational Dialectometry

Instructor: John Nerbonne (course under development)
Course Number:
Seminar room:
Mon. & Wed. 16:15-18, Oct.31 - Nov.9 and Nov.28 - Dec.7

Announcements 2005

For those of you who don't read the New Yorker, I've excerpted a recent issue with a brief story on Bill Labov. Enjoy!
This course overlaps in content a great deal with the course I taught at the 2005 Linguistics Institute (Harvard/MIT). See Aggregate Dialectal Variation. People who took that course will see little new here (but see section on detectic syntactic variation).
On Mon. Nov. 7 the course will not meet. But there will be an SFB-wide colloquium on this topic. Details later.

Description

This course will focus on what Hans Goebl has called the "linguistic management of space," how language variation is structured geographically. We shall begin with basic ideas from categorical data analysis which we'll apply to lexical and syntactic data, then examine a technique from computational linguistics (edit distance or Levenshtein distance) for the analysis of sequences of segments in pronunciation. We shall examine questions of validity and consistency, and we show how to visualize analyses using the L04 package, where the emphasis is on visualization for the purpose of exploration and understanding. Time permitting we shall turn to one or two advanced topics, e.g., explanatory models of the geographic conditioning of language variation, attention to the role of linguistic structure in the geographic distribution, or techniques for analyzing syntactic variation from text corpora.

The course assumes no familiarity with dialectology or computational techniques. Some basic linguistics is helpful, as is an unintimidated attitude toward software. We have three goals:

to show how aggregate analysis works, and to get participants to understand it critically
to compare this to other sorts of analyses

Students are encouraged to try aggregate analysis of language variation on data from American dialects using Peter Kleiweg's L04 package. This presupposes familiarity with UNIX. Note in particular that there is a tutorial on its use.

Schedule

Overhead sheets are available.

Introduction. Concepts & History.
Measuring Lexical Differences
Measuring Pronunciation Differences
Validating and Calibrating.
Location, location, location. Why are there dialect differences?
Linguistic Structure in Dialect Differences
Detecting syntactic variation in corpora (tentative).

Readings.

Introduction. Chambers & Trudgill (1998) Chap. 9; Nerbonne & Kretzschmar (2003).
Measuring Lexical Differences. Nerbonne & Kleiweg (2003).
Measuring Pronunciation Differences. Nerbonne, Heeringa & Kleiweg (1999); Heeringa (2004), Ch.5
Validating and Calibrating. Heeringa, Nerbonne \& Kleiweg (2002); Heeringa (2004) Ch.7
Location, location, location. Why are there dialect differences? Heeringa & Nerbonne (2002); Chambers & Trudgill (1998) Chap. 11; Nerbonne, van Gemert & Heeringa, (submitted).
Linguistic Structure in Dialect Differences. Nerbonne (submitted).

Literature

See literature list.

Some Useful Links

LAMSAS home page
Dialectometric work on LAMSAS at the University of Groningen
Peter Kleiweg's L04, a dialectometric package focusing on the use of Levenshtein distance to assay pronunciation difference. Note especially that there is a tutorial on its use.

An online demo of how Levenshtein distance works.
Hans Goebl's Dialectometry Project.

Exercises

Are available here.

John Nerbonne

Last modified: Wed Nov 16 21:51:28 CET 2005