Projecting Dialect Distances to Geography: Bootstrap Clustering vs. Noisy Clustering

Abstract

Abstract. Dialectometry produces aggregate distance matrices in which a distance is specified for each pair of sites. By projecting groups obtained by clustering onto geography one compares results with traditional dialectology, which produced maps partitioned into implicitly non-overlapping dialect areas. The importance of dialect areas has been challenged by proponents of continua, but they too need to compare their findings to older literature, expressed in terms of areas.Simple clustering is unstable, meaning that small differences in the input matrix can lead to large differences in results (Jain et al. 1999). This is illustrated with a 500-site data set from Bulgaria, where input matrices which correlate very highly (r = 0.97) still yield very different clusterings. Kleiweg et al. (2004) introduce composite clustering, in which random noise is added to matrices during repeated clustering. The resulting borders are then projected onto the map.The present contribution compares Kleiweg et al.'s procedure to resampled bootstrapping, and also shows how the same procedure used to project borders from composite clustering may be used to project borders from bootstrapping.

A Neural Model for User Geolocation and Lexical Dialectology

“…The traditional approach to dialectology is to find the geographical distribution of known lexical alternatives (e.g. you, yall and yinz: (Labov et al, 2005;Nerbonne et al, 2008;Gonçalves and Sánchez, 2014;Doyle, 2014;Huang et al, 2015;Nguyen and Eisenstein, 2016)), the shortcoming of which is that the alternative lexical variables must be known beforehand. There have also been attempts to automatically identify such words from geotagged documents (Eisenstein et al, 2010;Ahmed et al, 2013;Eisenstein, 2015).…”

Holistic corpus-based dialectology

“…By contrast, in linguistics and particularly in corpus linguistics, we find a long and strongly entrenched tradition of looking at individual features in isolation, which is partly a legacy of the discipline's philological origins, and partly a convenience issue. In any event, the one-feature-at-a-time line of analysis -exceptions such as the multidimensional register studies in the spirit of Biber (1988) notwithstanding -has yielded a corpus-based dialectology literature dominated by an abundance of what Nerbonne (2008) has referred to as 'single-feature-based studies'. We will refrain from citing actual studies here (but see the survey in ANDERWALD; SZMRECSANYI, 2009), though fictitious titles such as 'Verbal complementation in West Yorkshire English' or 'The KIT vowel in Appalachian English' are entirely realistic.…”

“…Starting in the 1970s, computationally inclined dialectologists have addressed these worries by developing a methodology known as DIALECTOMETRY. Dialectometry is defined as the branch of geolinguistics concerned with measuring, visualizing, and analyzing aggregate dialect similarities or distances as a function of properties of geographic space (for seminal work, see SÉGUY, 1971;GOEBL, 1982;GOEBL, 1984;KLEIWEG, 1999;NERBONNE, 2005;GOEBL, 2006;KLEIWEG, 2007). Dialectometrical inquiry marshals computational approaches to identify "general, seemingly hidden structures from a larger amount of features" (GOEBL; SCHILTZ, 1997, p. 13) and puts a strong emphasis on quantification, cartographic visualization, and exploratory data analysis to infer patterns from feature aggregates.…”

“…The well-known major intrinsic drawback of the corpus-based method is that it is unable to deal with textually infrequent phenomena (see, e.g., PENKE;ROSENBACH, 2004, p. 489), and data sparsity is a particular concern when the focus is on syntax and lexis; in this case, a questionnaire study may indeed be the more appropriate research design. Nonetheless, one may justifiably wonder if phenomena that are so infrequent that they cannot be described on the basis of a major text corpus should have a place in an aggregate analysis at all.…”

Corpus-based Dialectometry: Aggregate Morphosyntactic Variability in British English Dialects

“…The resulting higher-order structures can be visualised, for example, via socalled composite cluster maps (see Nerbonne et al 2008 for a discussion). These highlight the fuzzy nature of dialect boundaries such that darker borders between localities represent more robust linguistic oppositions (which, thanks to the clustering-with-noise technique utilized, can be considered statistically significant).…”