Lects in Helsinki Finnish - a probabilistic component modeling approach

Abstract

This article examines Finnish lects spoken in Helsinki from the 1970s to the 2010s with a probabilistic model called Latent Dirichlet Allocation. The model searches for underlying components based on the linguistic features used in the interviews. Several coherent lects were discovered as components in the data, which counters the results of previous studies that report only weak covariation between features that are assumed to be present in the same lect. The speakers, however, are not categorical in their linguistic behavior and tend to use more than one lect in their speech. This implies that the lects should not be considered in parallel with seemingly uniform linguistic systems such as languages, but as partial systems that constitute a network.

Corpus-based dialectometry with topic models

“…The latter would anyway be hard to detect in the examined dataset of US-American tweets. Kuparinen et al (2021) used a Latent Dirichlet Allocation model (see Section 3.3) to discover lects in Helsinki Finnish. Although the aim of the study was similar to the current one, the data was differently pre-processed.…”

“…Secondly, the dialectologically meaningful features are tied to the words, which means we are not actually calculating the frequency of the variants themselves (cf. Kuparinen et al, 2021), but the combinations of words and variants. If we modify the example from before, the occurrences talosa "in a house," koulusa "in a school," and kirkosa "in a church" would all end up as different tokens in the corpus, although they all have the same dialectal variant -sa of the inessive case.…”