ABSTRACT
In this chapter, we consider texts mentioning ‘austerity’ from the perspective of systematic skewings of conceptual and semantic associations. On the basis of assumptions made in the theory of systemic-functional linguistics concerning tight relationships between language use and social context, we explore some new techniques for locating potential socio-political differences exhibited in texts, taking the ‘austerity’ corpus as an example. For this, we adopt computational word embedding methods in order to conduct some initial studies of their applicability to the theoretical and descriptive challenges raised. Previous approaches using word embeddings have found suggestive correlations between plausibly selected word groups and external socio-cultural variables. Our main focus here is to find ways in which word embeddings might be used to derive dimensions of socio-political interest without prior assumptions concerning how those dimensions are expressed linguistically.
