Transparent and non-transparent languages

Kees Hengeveld, Sterre Leufkens
  • Folia Linguistica, March 2018, De Gruyter
  • DOI: 10.1515/flin-2018-0003

What is it about?

Languages can be thought of as a list of mappings from meanings to forms. Intuitively, it seems to make sense for languages to be fully transparent, that is, to maintain 1:1-relations between meaning and form throughout their grammars. However, all languages violate this principle in some way. This is all the more suprising since there is evidence that non-transparent structures are hard to learn. For example, grammatical gender in Dutch (the fact that soms nouns receive the article 'de' and others 'het' - a formal distinction not grounded in a meaning distinction) is very difficult, for children and adults learning Dutch. For long, linguists have assumed that all languages are equally complex and equally transparent. In this paper, we show that this is not the case. We have measured the degree of transparency of 22 languages from all over the world by counting the amount of non-transparent features in their grammars. It turns out these languages can be ranked from least transparant (French) to most transparent (Sri Lanka Malay). This shows that some languages are in fact harder to learn than others, both for an L2 as an L1 learner. Furthermore, we observe an ordering in non-transparent features: some are only present in the least non-transparent languages, while others are exhibited by all languages.

Why is it important?

In this paper, we establish a relation between the nature of particular features of grammar (non-transparent structures), their learnability (how difficult is it to acquire these features), and their distribution over the languages of the world. This helps us understand why languages develop and maintain certain difficult structures.

The following have contributed to this page: Sterre Leufkens

In partnership with: