The world's languages show great diversity, but at the same time, striking commonalities. A central objective of the language sciences is to explain what gives rise to such commonalities. Over the years, a host of cognitive biases have been proposed to impact how languages emerge, evolve, and change over time. In this talk I follow this line of work, focusing on a specific puzzle found in many of the world languages: the prevalence of redundancy in linguistic systems. Linguistic redundancy - using more linguistic material than needed - is attested across languages, even though it seems to add complexity to language systems and to go against principles of efficient communication. I offer a learnability-based perspective on the presence of such redundancy in language. In particular, I make two predictions: (a) redundancy can be beneficial for learning, and (b) speakers increase the use of redundant cues when conversing with learners. I will present a series of studies which aim to test these predictions by looking at both child and adult learners, across different levels of linguistic analyses (morpho-syntax, discourse, full language).
The results support both predictions, providing initial support for the idea that learning constraints help maintain redundancy in languages.
