Modeling Linguistic Landscapes

Abstract Much work in Linguistic Landscapes approaches multilingual landscapes in qualitative terms; inferential statistical approaches are still underrepresented. The present paper adds to filling this methodological gap by investigating the Linguistic Landscapes of St Martin, a highly multilingual, eastern Caribbean island, divided into a formerly Dutch-colonized and French-colonized part. It does so by employing an inferential statistical method to study the nature of linguistic diversity on signs. Based on 372 and 373 signs respectively for each of the commercial districts of Philipsburg and Marigot, we quantitatively analyze the occurrences of languages on the different signs and statistically model the data by means of Multi-Label Classification (MLC; e.g. Rivolli & de Carvalho, 2018 ). The results show that both shopping districts are characterized by multilingual patterns of Linguistic Landscapes but with differences in their exact linguistic setups, most importantly the interplay and relationships between different languages, motivated by differences in their sociolinguistic setup.