Language Complexity as an Evolving Variable ed. by Geoffrey Sampson, David Gil, and Peter Trudgill (review)

This volume, based on a 2007 workshop in Leipzig, begins with a self-consciously radical manifesto by coeditor Sampson. His "A Linguistic Axiom Challenged" (pp. 1-18) is an attack on models that claim roughly invariant complexity of natural languages, whether formalists or post-Sapirean descriptivists. In the guise of a vengeful angel of global warming, he concludes by dismissing equicomplexity as "a melting iceberg" (p. 18). Most of the following papers also question equicomplexity, though with fewer rhetorical flourishes. They average fifteen pages and in most cases are handbook-like summaries of work published elsewhere at greater length.

Can complexity be meaningfully defined? Should we focus on overt complexity (tabulating the number of units, hierarchical levels, or both in speakers' outputs), or should we focus on listeners' processing difficulty? Can languages (leaving pidgins and creoles aside) differ widely in overall complexity, or is there a trade-off across modules (such as morphology versus syntax) so that all languages end up in a dead heat overall?

Guy Deutscher, "'Overall Complexity': A Wild Goose Chase?" (pp. 243-51), argues that complexity scores cannot meaningfully be calculated for languages, which would make the debate over equicomplexity rather moot. By contrast, Johanna Nichols, "Linguistic Complexity: A Comprehensive Definition and Survey" (pp. 110-25), is optimistic that (overt) complexity can be calculated in phonology, grammar, and lexicon. She includes constraints and dependencies as well as inventory counts. Based on preliminary calculations, she finds no evidence for negative correlations (trade-offs) across domains within a language. Likewise, Östen Dahl, "Testing the Assumption of Complexity Invariance: The Case of Elfdalian and Swedish" (pp. 50-63), denies that the difference in morphological complexity between Swedish and the closely related (and endangered) Elfdalian is compensated for by a countervailing difference in syntactic complexity. David Gil, "How Much Grammar Does It Take to Sail a Boat?" (pp. 19-33), argues that a minimal language (no morphology, no stem-class oppositions, semantic composition by "association") is generally adequate for human communication, and that some languages such as (semidiglossic colloquial) Riau Indonesian attain such minimality in conversational stretches and approximate it elsewhere. Other languages are much more complex (e.g., have cascading syntactic hierarchies), but such complexity is "hugely dysfunctional" insofar as such a language "forces us to say things that we don't want to say" (p. 32). The implication of the papers by Nichols, Dahl, and Gil is that languages can differ greatly in overall overt morphosyntactic complexity.

A different tack is taken by John Hawkins, "An Efficiency Theory of Complexity and Related Phenomena" (pp. 251-68), who injects his trademark crosslinguistic online processing-efficiency calculations into the argument. The basic constituent-order choice of a language (e.g., subject-object-verb) has strong statistical implications for case marking, extraposition, and other morphosyntactic phenomena. Functional trade-offs are quasi-intelligently worked out in linguistic usage by a Darwinian filter. Wide cross-linguistic variation in overall morphosyntactic complexity would make little sense in this model. Matti Miestamo, "Implicational Hierarchies and Grammatical Complexity" (pp. 80-97), takes an intermediate position on trade-offs. His typological survey awards points for the number of input categories (from an implicational hierarchy) that are positive for a grammatical feature in a given language. Data show that agreement and nominal case marking are not inversely related, i.e., one does not compensate for the absence or weakness of the other. However, there is a degree of trade-off between verbalization and use of copula (defined broadly) for various predicate types.

The volume does not focus on emergent pidgin and creole languages. However, John McWhorter, "Oh : A Bewilderingly Multifunctional Saramaccan Word Teaches Us How a Creole Language Develops Complexity" (pp. 141-63), deals with an evolved creole. His general position is that freshly minted creoles have a unique, telltale morphosyntactic signature, and they take many centuries to accumulate the...