Learnability in a parametric world
Workshop hosted by Romance Turn IX

With the advent of the Generative Grammar, the study of human language has focused on two central problems. On the one hand, the so-called language acquisition problem: although the linguistic input is limited and fragmented, the development of a child’s language occurs rapidly and systematic. On the other hand, the linguistic variation problem: while it is true that natural languages vary, it is also true that they do not vary in an unlimited way. In this sense, there must be certain constraints on variation (and consequently on change) that are induced by UG?

During Government-Binding times UG is specified for general principles shared by all languages, of which some are open to parametric variation (Chomsky 1981). Grammatical differences between languages are thus the outcome of the interaction of a finite number of ‘parameters’ encoded in UG (Baker 2001). Accordingly, language acquisition is explained by the exposition of a language learner to the linguistic environment, which contains enough data to set the parameters of UG in one way or another.

However, the vast number of parameters that have been attributed to UG during the GB era has increased the structural complexity of UG itself. This overspecification has consequently been subject to severe criticism during the last years (Newmeyer 2004, 2005; Boeckx 2011, 2014). Minimalist constraints and current evolutionary assumptions force us to rethink the concept of parameters and to see the content of UG as a minimized system (Boeckx 2006). Complexity in these terms is transferred to or seen as a consequence of general cognitive skills, the so-called third factor (Chomsky 2005). In this context, and focusing on third factors as an indispensable means of attaining explanatory adequacy, “learnability” seems to play a central role. However, until today, there seems to be no adequate explanation for learnability as a third factor and its effects on language acquisition, language variation and language change as a whole. For instance, it seems that current acquisition approaches that use learnability as a ‘third factor’ continue to use the classical concept of parameters in order to account for their data (Fodor 2017). The problematic tension between parametric theory and minimalist approaches has not yet been solved. In terms of the challenges that emerge from current theoretical proposals, the question remains what the components of learnability are? What role does learnability play in the interaction between acquisition, change and parametric variation?

 

We welcome papers for oral presentation on different aspects of learnability:

  •  What is the state of the art concerning learnability in language acquisition and parametric theory?
  •  What are the essential components of learnability? E.g. algorithmic  mechanisms, parsing, parametric space, etc. How are these components connected to computational efficiency? Does the child “learn” the grammar, as  it learns the lexicon?
  • How does learnability constrain language change and language variation? Are there limits to variation?

  • Are language-specific cognitive mechanisms necessary? What does “third factor” mean for a learnability theory?

 

Invited speaker: Charles Yang (University of Pennsylvania)

 

Organized by Susann Fischer, Mario Navarro and Jorge Vega Vilanova