Dialect Transfer for Swiss German Speech Translation

TL;DR

This paper explores the challenges of developing Swiss German speech translation systems, focusing on dialect diversity and linguistic differences from Standard German, and examines how dialect inclusion affects translation performance.

Contribution

It provides an analysis of how dialect diversity and linguistic differences impact Swiss German speech translation, offering insights into dialect transfer and system performance.

Findings

Dialect diversity significantly affects translation accuracy

Linguistic differences between Swiss German and Standard German influence system performance

Inclusion of multiple dialects improves overall translation robustness

Abstract

This paper investigates the challenges in building Swiss German speech translation systems, specifically focusing on the impact of dialect diversity and differences between Swiss German and Standard German. Swiss German is a spoken language with no formal writing system, it comprises many diverse dialects and is a low-resource language with only around 5 million speakers. The study is guided by two key research questions: how does the inclusion and exclusion of dialects during the training of speech translation models for Swiss German impact the performance on specific dialects, and how do the differences between Swiss German and Standard German impact the performance of the systems? We show that dialect diversity and linguistic differences pose significant challenges to Swiss German speech translation, which is in line with linguistic hypotheses derived from empirical investigations.