Recursion and Ambiguity: a Linguistic and Computational Perspective

In this chapter I will be concerned with what characterizes human language and the parser that computes it in real communicative situations. I will start by discussing and dismissing Hauser, Chomsky and Fitch's(2002) (HC&F) disputed claim that the "only uniquely human component of the faculty of language" be "recursion". I will substantiate my rejection of HC&F's claims, with the fact that recursion only appears in mature and literate language - an opinion also shared by some papers in a book on recursion by Harry van der Hulst (2010). I will then present in detail Chomsky's proposal - now part of the Minimalist Theory (MT) - of the architecture of the human parser as being based on Phases. I will accept this part of the theory and compare it with the computational architecture contained in a system for deep text understanding called Getaruns (Delmonte,2007;2009). I will then argue in favour of what I regard the peculiar component of human language faculty that is "the ability to associate meaning to deficient propositions and generic linguistic expressions, which are highly ambiguous in their structure”. And this is also due to the presence of recursion (but not only). I will then speak in favour of a parser that takes "context" into account and strives for a "complete" syntactic representation. As to the problem of ambiguity, I will introduce the use of a computational device, a lookahead mechanism, which is presented in association with the need to specify UG parameters characterizing a given language. I will discuss the use of psychologically viable Parsing Strategies implemented in the parser to overcome ambiguity and prevent Garden Path, whenever possible. This will be highlighted by reference to peculiar features associated to different languages, Italian and English. Eventually, I will present a theory that encompasses all my previous proposals and is called LSLT