Recursive lemons: A review of Recursion across domains

Some introduction

I have always admired those who, given a pile of lemons, know how to make lemonade. This collection of essays is a testament to this admirable skill (one, I should add, that I lack). The lemons in this case consist of the many discussions of recursion by critics of contemporary generative grammar. The lemonade on offer is an interesting program for the crosslinguistic investigation of variation in self-embedding. In what follows, I want to survey the lemons and the lemonade. In addition, I want to briefly consider what theoretical implications the lemonade on offer might have for current syntactic theory.

Before I begin, let me make a confession. I am not really the perfect reviewer for this (kind of) collection. It is chock full of careful empirical crosslinguistic investigations, many of the target languages being understudied. Unfortunately, empirical investigation is not my long suit, as many who know me will be quick to confirm. Facts usually make me itchy, and the book contains subtle descriptions of and diagnostics for many different kinds of self-embedding. It also illustrates how languages can differ in the kinds of self-embedding they allow and avoid, which forms the core of a very large and potentially interesting project. My allergies will lead me to pass lightly over many of the specific empirical findings in what follows. But my distance should not be taken for disdain. The papers are replete with fascinating details that suggest further interesting questions. My concentration on the further questions should not be understood as an invidious evaluation of the underlying careful description.

This review, like Gaul, is divided into three parts. The first reviews the place of recursion within the theory of grammar, in particular Chomsky’s contention that it is the basic property of human language. This claim has generated a very (ahem) lively ‘debate’ between Noam Chomsky and Daniel Everett, the latter denying that recursion is a universal feature of language. I argue that this debate is based on an equivocation and so is largely pointless. In my opinion, the main utility of the ‘debate’ has been to highlight the distinction between two conceptions of universals, Chomsky’s mentalist conception and Joseph Greenberg’s typological one. The second part of §1 highlights the intimate connection between recursion and hierarchy in various generative approaches to structure building. It is universally agreed among generative grammarians that languages are hierarchically organized in the sense of having depth as well as breadth (i.e. constituents contained within constituents within constituents). Given contemporary Merge-based bare phrase structure theories, the fact that even simple clauses are hierarchically organized leads directly to the conclusion that recursion must be a feature of any viable grammar.

Section 2 discusses the project that lies at the heart of the volume. Granted that recursion is part of universal grammar (UG), do languages vary in the kinds that they tolerate? The papers catalogue some of the different kinds of recursion we find within grammars. Some also directly challenge Everett’s claims concerning the lack of recursion in Pirahã.

In §3 I wrap up the discussion and consider some implications of this comparative project. I suggest that the project provides a model of how to fruitfully bridge a divide between two kinds of generative research, the mentalist and the typological.

That’s the outline. Without further ado, let’s begin.

1. The lemons: recursion: what it is and how it is grammatically coded

1.1. The ‘debate’

Though I have not interviewed the editors, I suspect that the main motivation for the volume is the minimalist claim that recursion is a universal (i.e. UG) property of natural language grammars. This claim has been the center of a well-covered ‘debate’ (especially in the highbrow press, e.g. The New Yorker and Harper’s and The Chronicle of Higher Education) involving Everett and Chomsky regarding the latter...